(* Content-type: application/vnd.wolfram.mathematica *) (*** Wolfram Notebook File ***) (* http://www.wolfram.com/nb *) (* CreatedBy='Mathematica 12.2' *) (*CacheID: 234*) (* Internal cache information: NotebookFileLineBreakTest NotebookFileLineBreakTest NotebookDataPosition[ 158, 7] NotebookDataLength[ 1336947, 22859] NotebookOptionsPosition[ 1324222, 22650] NotebookOutlinePosition[ 1324719, 22669] CellTagsIndexPosition[ 1324676, 22666] WindowTitle->BERT Trained on BookCorpus and Wikipedia Data-examples WindowFrame->Normal*) (* Beginning of Notebook Content *) Notebook[{ Cell[CellGroupData[{ Cell["BERT Trained on BookCorpus and Wikipedia Data", "ResourceExampleTitle", ShowCellBracket->False, CellMargins->{{27, Inherited}, {10, 30}}, CellGroupingRules->{"TitleGrouping", 0}, PageBreakBelow->False, CellChangeTimes->{ 3.763318266977738*^9, 3.763404216481764*^9, 3.763417091031337*^9, { 3.797848617302459*^9, 3.797848617631496*^9}}, ShowAutoSpellCheck->False, LineSpacing->{1, 4}, MenuSortingValue->1100, MenuCommandKey:>"1", FontFamily->"Source Sans Pro SemiBold", FontSize->32, FontTracking->-0.5, FontColor->RGBColor[{ Rational[31, 85], Rational[146, 255], Rational[11, 85]}]], Cell["Represent text as a sequence of vectors", "ResourceExampleSubtitle", ShowCellBracket->False, CellMargins->{{27, Inherited}, {0, 0}}, CellGroupingRules->{"TitleGrouping", 10}, PageBreakBelow->False, CellChangeTimes->{{3.732976241020545*^9, 3.732976241210205*^9}, { 3.7633182707869587`*^9, 3.763318288497417*^9}, 3.76374049828244*^9, 3.763924034466608*^9}, ShowAutoSpellCheck->False, CounterIncrements->"Subtitle", MenuSortingValue->1550, FontFamily->"Source Sans Pro Light", FontSize->26, FontSlant->Italic, FontColor->RGBColor[{ Rational[1, 3], Rational[1, 3], Rational[1, 3]}]], Cell[CellGroupData[{ Cell["Resource retrieval", "Subsection", CellChangeTimes->{{3.728626116968351*^9, 3.728626120224564*^9}}], Cell["Get the pre-trained net:", "Text", CellChangeTimes->{{3.728401746509698*^9, 3.72840174782841*^9}, { 3.728401830325835*^9, 3.7284018327087107`*^9}, {3.728626101672206*^9, 3.7286261059281063`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetModel", "[", "\"\\"", "]"}]], "Input", CellChangeTimes->{{3.7284011245153217`*^9, 3.728401140115395*^9}, { 3.728401348213561*^9, 3.728401349155575*^9}, {3.728627113743627*^9, 3.728627115125251*^9}, {3.730621122588628*^9, 3.730621128691823*^9}, 3.76331830130238*^9, 3.763334053983881*^9, 3.763334130870682*^9, { 3.763382908067416*^9, 3.763382911546514*^9}, 3.763404216487216*^9, 3.763417093106144*^9, {3.7979224670505247`*^9, 3.797922467165854*^9}}, CellLabel->"In[11]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsvQl8FUW6/h8EZREFFxYVEBdEmXHBhVGZEUdFZRiGGRSvwgwKzOAusjiu wAl72CQGCJiwQwRMwp5gICyC7CREkLAFEEKABMKEy7lz9arzf37n+VD/orpP n07OCUlO3u8n8DldXV1dXf121/t0bbf06N2p12URERH9auC/Tt3f/33fvt0/ fK4uNjq/0+/1V9/5x9/bvdP/H6/+o+/DPaoicAn+dasWEfH/fv9HEARBEARB EARBEARBEARBEARBEIQKxY8//lhYWHjixInjx4/nCoIgCIIgCEKlB47xyZMn 4STDVS5rb124FPzyyy+43Xl5eQUFBf/617/+WxAEQRAEQRAEH3CP4STDVT5z 5gzc5rL23IVSBPc3Pz//9OnTuO/nz5/3CoIgCIIgCIKgAScZrjIcZrjNIo7C GIjfwsJC3OuytjhBEARBEARBKL/AYS70Udb+u1Aq/PjjjwUFBSKLBEEQBEEQ BCEgcJvhPMuYo7Dk7NmzRUVF0olOEARBEARBEAICtxnOM1zosvbihdDD4UVl bWKCIAiCIAiCUDHggKOy9uKF0CMjjARBEARBEATBPRxtVNZevBB6OEd3WduX IAiCIAiCIFQMOI93WXvxQugRZSQIgiAIgiAI7hFlFK6IMhIEQRAEQRAE94gy CldEGQmCIAiCIAiCe0QZhSuijARBEARBEATBPaKMwhVRRoIgCIIgCILgHlFG 4YooI0EQBEEQBEFwjyijcEWUkSAIgiAIgiC4R5RRuCLKSBAEQRAEQRDcI8oo XBFlJAiCIAiCIAjuEWVUoTlz5kyBxs8//6x2iTISBEEQBEEQBPcYygghuqeN zbLw9wVX/PDDD29dzKFDh9TeCqGMdn2/adG2ScX6+/7UPn+pfZt7Mn7Ld8X6 wyH+UjtyJH9hcmax/nbtOlo65SQIgiAIgiCUOoYymj17tu5pz5gxoyxcfsEV //u//4t7lJKScuwCP/30k9pbIZQRlM5rM+4v1t/2g6v8pZb07cHH4pYX6w+H +Ett27ZDvf4xr1h/EEelU06CIAgh4OhR+XojCILghKGMioqKlJs9fPjw6dOn l4HHL7iDymjbtm22eyuQMhqwpOPo9B4B//6Z+LQbZdTui/XPLckK+IdobpRR 37eWDXl/Q8C/D979SpSRIAjlmTfffDMiImLAgAFlnRFBEITyi8M4o/Hjx4sy Ks+EjTKC6pm5c1DAP8+yTm6U0Z+Td3RfeyzgH6K5UUYf9Vk5cfCegH8QR6KM BEG4xGRmZg4cOHDx4sVuIrdt2xbK6MUXXyztXAmCIFRcRBlVXEQZGYgyEgSh UjFnzhyInV69ermJnJOTM2XKFOlQJwiC4IAoo4qLKCMDUUaCIFQqiqWMBEEQ hICIMqq4iDIyEGUkCEKYsXfv3ldeeeXuu+++8cYbu3TpMm3atNOnTyMcb/jW rVs3a9YMyqhBgwaPPvpou3bteEhMTAw209PTBw0a1KpVqzp16mzYsAHhn3/+ OcLnz5+vR1uzZk10dPQTTzxx/fXX33fffQMHDkTKegYKCwvHjRuHxOvVq9ey ZcsxY8bk5OTgwHfffffSloQgCMKlQJRRxUWUkYEoI0EQwoktW7Zce+210D5N mjSBOKpSpQp+P/nkk9h19uxZBF5zzTUIqVWrVuPGje+9914e1bdvXwRCxUT4 qFGjxvr16xH+8ccfY/PTTz/VoyE1/F+7dm1EY/wePXroeYAcY3jdunUvv/xy /Pj973+P/5966qlLWxiCIAiXAlFGFRdRRgaijARBCCdee+01aJA+ffqcP38e m/v3769fvz5C1q1bxwi2vekoecCgQYOOHTumwm2VEZgwYcK5c+dOnjw5YsQI bEJ/HTlyhHFiYmIQcuWVV6akpKBCyc/PHzx4MI8SZSQIQlgiyqjiIsrIQJSR IAjhRIcOHaBBJk2apEKOHz8OfVRYWMhNB2XUuXNnIzVbZdSpUycVAVXG9ddf j8CVK1cypF27dkYGAFIWZSQIQrhSOZVRbGxs//79J0yYcAnO9fPPP0+bNg2n mzx5MrRMCFMWZWQgykgQhHBi1qxZ0CBVq1b985//PGPGjIMHzfeVgzKKjo42 Itsqo6ioKD0OhBICv/zyS242bNgQm8Z5UaOJMhIEIVwpsTL66aefVq1alZqa un37duverKws7MrMzCyRyx8UBQUFmzZtWrBgAd7/ECNpaWk5OTlGHNQF0BTj xo27BPnZvXv3WxdAxkKYcmkoo6UhwuXpRBkJgiD4A6/xyMjIunXrRlygXbt2 nE6BOCgjVH9GarbKyIj28ssvK2V05swZ/L788suNdFavXi3KSBCEcKXEyuh/ /ud/6O337t0b709j77Bhw7Br2rRpJfX6/3927tyZkJCQkZHhJvLatWvffffd tyzMmjXrhx9+UNEupTI6ePCgyoY/FVMyRBkZiDISBCH8OHnyJKTKO++8wxac mjVr5uXlcVepKiPQqFEjbO7bt0+P8/nnn4syEgQhXAleGYG5c+cae0OojObP n4+kYmNjnaP9/PPPU6ZMUVkaNGjQ1KlTY2Ji+vfvzxDUICrypVRGYNWqVcjb 8uXLf/rppxAmK8rIQJSRIAhhzNmzZx9//HGoEtTODKEy6t69ux4thMqoffv2 2BwzZoweh4GijARBCEtCoozeeeed/Px8fW8IlREEjhtltGHDBmbmvffe2717 t57PuLg41AiFhYUq8BIro1JClJGBKCNBEMIGvMBvu+22+vXrb9q0iSHnz5+n KomPj2dIYmIiNu+66y79wBAqIw4puuKKKyDBTp8+nZOT06dPH5mbThCEMCYk ygjMnDlT3+usjH766adjx45t3bp1z549/s7+H18z0A8//IDXOJKaMGGCNfMq 5r///e8PP/wQ0d59990TJ05YkzKmPjCUEfJz4MABSIzvv/9e73Rnm3PEyczM ROYPHTqk50HHX2n/8ssv1sg//vgjduH/4mam9JTRMg1nBbTMgigjQRCEkPD2 229zuaLu3btDyLRq1Qqb9erVQzXHCKhDucgRNFTPnj0ZGEJlBN544w1KIZ4I /OlPfxJlJAhCuBK8MoL8wf94geuSxJ8ygv+/ePFiYyjQ8OHDjx49qkfbsGGD 7XAhnffff1/FX716NQMXLFjgICUUVEYxMTG4/KlTp6KCUMlCYaGusR6Sl5c3 d+7c9957T88DrhqXo0QNgVTxl2evnZIaM2YMdq1ater8+fMuM0NEGRmIMhIE IZw4d+7coEGDOJM2JdIf/vAHY6Y4VGQ333wz9l577bUM6devHzanTJlipPbJ J5/oPfFso0GCITAxMVHPQ0JCAsJbtmzZrl07/EYFLcpIEIRwJXhlBM8cr278 gFev9toqo6KiosjISB7Vu3fvIUOGfPTRR9x855139C5wa9eudZZF4IMPPlDx 58yZw0C9y5wDVEYDBgxgzpkflXKfPn2QVT3++vXrIYJUBFQoun4xhllt2rSp WMqImYmNjXWZGUVpKyM3veaW+LAe5fJ0JVZGbLAzKLEysk1NlJEgCOWEw4cP 79q1y98r/fz583l5eadPnw75eZFmbm6uWj6JcLHXLl26hPx0giAIZU7wyghC YOPGjfx97Ngx7rVVRpMnT2a0efPmeS9ohH379n3yySdv+eZMUJ3H4POfvcDA gQOx97PPPjt7MbpeGDdu3FsXtyI5QzFC4uLiTp06hUC8/HEWBi5fvlyPf+TI EQR+/PHHkGyI/IuP7OxsKjuIJtRKKjKuwshqUlJSQGXkPjOKyqyMYJk5OTlG aiVWRl988cV3331npCbKSBCESs6IESMggu65556CggKG7Nix48Ybb0Qg6vey zZsgCEJpELwyWrdu3U8//cTGIGgf7rUqo8zMTNsWFgCJwV22SyN5PJ63As3A AE2EOKNGjXKIo6PEyIoVK/Rw6JF33nkH4fHx8cYh+/fvN3rNgYyMDKbz9ddf O5wO0saNMnKfGVKZZ2BAmUyYMEF1ticlVkYJCQmwduhfPTVRRoIgVHLy8vJu uOEG6KAaNWq0atXqlltuYb++xx577NSpU2WdO0EQhNATvDJau3YtNrdu3cpN uJf/sVNGCxYsYPPK+fPnramx8WXZsmXWXW6UEYf/IMMOcXQoRkaPHm3dRYln u0vxyy+/nDlzZs+ePWvWrOFVJyUlOcR3o4xKkJlSVUa4FytXrlzlGkTWm5lc ni4YZTRmzBj8r3fzCEYZITXYmF7XizISBEE4fvz43//+9zvvvPOKK66oXbv2 I488gvq6qKiorPMlCIJQKoRKGf38889Dhw59yzeJnPL2dWU0duxYxh9nB3fZ zmXnRhkNGTKE/fEc4ug4zNrNqfBGjBhh3QW3GRoHF9KnT5+3LsbaEKbjRhkV NzP/KWVllJaWdvLkybOuQWQccomVEfjiiy/OnTvH1IJURgAGj3JjaqKMBEEQ FOd9lHUuBEEQSpdQKaP/aF3LDh48SJdeVzoUOM7oczgYBzorI67x+s4777hc StVBjOCqrWLkl19+WbZsmcrnu+++O3r0aFzd4sWLS1UZ2WZGUarKaNWqVdA7 /+saRMYhl14ZAXW64JURWLBgActNlJEgCIIgCEKlIoTKCPIBPjybjSZNmmTb ZtS/f//v/XPmzBnriaiMkKC/nIBFixYxM7YjlawUV4wkJycz/ZEjR+7YsePf //632sX2I1FGl0wZzVsxef369bCxMRqQMwj8dGl6cZURDsGBUNZ6aribCJz3 xUpRRoIgCIIgCJWHECojsGvXLgZyAmpdGc2fP5+7OPeae6i2nGdX2L9/PxMf OnSo7WqqBsUVIx9//DEC4TNbJ2EIV2VUbnvTrdg8b9++fbGxsUrIfPrpp1u2 bEFg/NptxVVGOAQHws5VapDwkEUITE3dKspIEARBEASh8hBaZQRGjx6tep3Z zk0H59NlnzfCub779u3rtZMVRjSKFCP9o0ePQjHp6yUVS4zgSrmYUXJyshF5 69at3BV+yqhCzMBAcGeZWkh606E8mZr0phMEQRAEQahUhFwZqSm437LMqBAX F6f6pO3du/f//u//6N5nZWXBI8Ve2xOlpaXxKGgfldWCgoJvvvnm559/VtFO nTrFOa6pMrZs2XLixImcnJyUlBQ26+B/7wVhUiwx8ssvv3Duuw8++ODgwYMI +fHHH5FybGysutLwU0ZB4vJ0wSujjRs3qtSCV0aQeCo1UUaCIAiCIAiVipAr o/9cmFHNqozOnz8PTaTUBIQM1yEib7/9tr54q36if/7zn29pqKnhDh06pMfc t28fZ/+20rdv3x07dqiYJR5nZIDTffjhh+GqjNyv91pWyig1NVVPLUhllJSU pM+8JMpIEARBEAShUlFiZQS3nB3JrMrowIED/lpSfv7553Xr1nHYji5bZs6c eeLECX/nys/Ph+NqqBLohf379xsxcTkzZsz44IMPVLR+/frFxMQcP35cj8bU OMG49aqtYgTZXrRokT5Z9yeffAJHGqpt9uzZ2ExMTPSX+f8EUkbFzYyikiuj +fPnGxcYjDKCBRordIgyEgRBEARBqFSUWBkFyfnz56FrsrOzz5w542bOBADH FfGPHDly9OhR2+VidQoLC3fv3p2bm6v3uAuSf//734cPH0aySDxUaQZD2Sqj JT7KShktXLjw7NmzRmolVkbIcEFBgZGaKCNBEASAOjdUSaFWcr9K7IkTJ/TV t8MelIxani/kWGtMgxDeZR3UrXDzSiPlklEsCxQqFqF6gspKGQnBU5mVkVqP VafEysg2NVFGlYGcnJyEhISvv/66rDMiCOWUN998MyIiYsCAAcEnBaUTFxe3 du1aN5EPHTrk8XgGDx6cl5cX/KnLD8ePH/9/q0LMmzd79uz09PTDhw+rXZs3 b46Njc3NzQ3h6SCIhg0bdv/99+Mmvvrqq/6iheoub9269eqrr/7zn//Mzb17 91511VXXXXcdrjrIlENCsSywgnLgwAEYmJpLqricOnVq48aNycnJ8P/hAWZk ZNh6krDbxYsXz5gxY+7cuatXr3bQ3d999916O5BPPRrOkpWVlZKSgvMuWrQI +bcK2IBxQvUEiTKquJSeMlpmwZ8CKitlZEuJlZEtoowqA1988QX8AVWPX2Iy MzMHDhyI+qVMzi5UWopleG3btsUz8uKLLwZ5UgicqKgoPHEuK6Y9e/Z4fEAi BXnq8kN2djZ0ikcD0m/79u3ce/78+aSkpOHDh4ew+QYSDLevWrVqTz311Oef f+4vWqjuMpxqpINzcRMaJMLHt99+G2TKwVNcC6xw4LrS0tIiIyNhV59++mkJ UsCzNnr0aM/FzJkzxxAgO3fuNOJAL/gTR7NmzfLYAcdSxUH6UFhGhPj4eL21 0U2cUD1BoowqLqKMDEQZCcWlbJURahycvVevXmVydqHSUizDy8nJmTJlSpCe BuqjST7c10pwcrKysr777rtgzluugGc+ZMgQuHMJCQm4rt27dyunUX3lxlXD 2YuOji4sLAzJSZ9//nnca/hyztFCcpe9FmUEkpOTU1JSgkw2eEpggRULWFdc XJySDCVQRpAeI0aMwLFTp07dtWvXgQMHVq1aRZ21YsUKFa2goGDUqFEInDFj RkZGBsTv2LFjsblo0SLbZFHm2Juammos9bJ3714VZ+HChYiDs8N+cN4NGzbw A8KCBQuKFccboidIlFHFpTLP2m2LKCOhuIgyEiohl97w4L3Ah8nOzvYXIeTj a0KYoL+kdu7cuXjxYn1GU52VK1d+8803egg8Orqd6hCkHBMTg0C9Qy9ECkJm z57tL+Vicdddd+FeHzzot6YLLVZlVE4IaIFeu371uAVu/KiAxubGGoO02Kio KLbdwJZKpozYEgRNoV8yzBiBsbGxKoQlOWHCBBVt//79lGP79u2zJstGKOdi ZBz97mzfvh0huCj1FLiJQ4J/gkQZVVxEGRmIMhKKi6GM8NS0bt26R48eu3bt 6t69++23337TTTe1a9du8+bN6hB4Mo8++uiaNWsWLlz42GOPNWjQ4Le//e2Q IUO+//57FefQoUOI849//EM/17fffotALqzGEzVr1gxnRwoIx1kuyRULFQll bCNGjHj44Yfr16/ftm1bWmNcXNyDDz54/fXXIxx7jQOTkpJef/315s2bN2nS pEOHDnPnzmW4g+HxXOnp6YMGDWrVqlWdOnU2bNiA8M8//xzh8+fPx2+EPPHE E88++6w+bGTy5MkI7N+/v7+r4IihSZMmWXfBY58xY8bIkSMHDx4MybB69er8 /HzuKioqghMyceJEbu7Zs2e8H/SPwwcOHMC1jxs3LjIyEjn/6quvDHcXmTly 5IhzsaelpSFZOPkQPvAwkRSuMTPTfL1PmzYN1/Xll19aHbAVK1bQTdUDU1NT x44da8il5ORkxMT/eiCKAoG4ZOd8ApTe3/72t6ZNm7Zo0eK1117TeyjBk8SN q169Ou41TAW/HXrT6XfZqxke/GTcXJjZfffdN3DgQKMwsYkIMKF69eq1bNly zJgx69evN5TR7373u/bt2+tH4bqQFF6esOfHH3/8ww8/PH36tPW6unbt2qhR o/vvv79fv37WUTPIIZ6Fhg0b/uY3v/n444+3bNniUEr+LBBuP+4R7izKH5lH HLZRQqTgNuGpGT58ON7tuPsoCuuBeKWnpKR89tlnOHDUqFEwPGOuCVwXIsCG EQG3A67R4cOHceC8efNUHJwLZg9bwokQB7fA9r4HtFscDs1y9uxZ6pQSKKPd u3cjhxkZGXogCgep4YFSIVQoqCL1aOznhmIx0oQXysJxOG9BQQGeCxSj/hyd OnWKaotzsLiJo+P+CbJFlFHFpTSU0SVGlJFQthjKCPUaNm+++ebbbrsNP+AP sJM8vAvVq6dv374I+ctf/oL/q1SpouLAQVUdrREZIQ899JB+LniVCHz66ae9 vmHRcFmvueYahNSqVatx48b33nvvJbtqoaJAY2vTpg3+p7WAunXr9u7dGz+u vPLKGjVqMBDeuzoKKoOB8DwRh7/p2zgYHs8F/5bxkTK8XITD7cSmcrQ4IOWV V17hJtyw2rVrV6tWTV902wDuFrvTGOHwW9i7DD4Pm074LZofz5FVbCICI+OZ GncxQ4cO5SHKHYXbyQ42KAGoLe6dM2eO8qbgzjHQGP1tMGXKFNUrif2LmKbx SRyKDD4zdiUmJuoOG+QYZdHJkycdzkLi4+MR2ZgTAK4yk3U+9qOPPuLNuvrq q2kJeCMpFYy9uL+XXXYZwqEg8BvOv7+kjLtMY3jyySfxP+6vMrMePXroR2GT 4TCzK664Aj+gpwxlxBep2oQWht7hUddddx1/wOr0L0vMDMU7T42YuivO82IX DmQil19+udXAFP4scNGiRfothrGxixc7brE9QpnZ8uXLjQNhhPgfJgfbYJyZ M2eqOHDA6J/rcdjxTJeoUNbK0vg44H88Vno+3ditqqFKrIyswKpRRSI1JeX4 VAJDI7OhyujY5r0gXqANIW22b9+OR2PTpk1u5lTJzs7GgVCmDu0+DnFcPkH+ EGVUcRFlZCDKSCgutsoI3HfffTk5OV6fnMFvfVon+gzgpZdeYm2+ZcuWm266 CSFcs8zrQhkR6U0nOENjq1q1akJCgtfnCSjl8uGHHxYVFcHx6N+/PzZbtGjB Q3bv3g1nGFo+LS3N6/Nk4CMhQtOmTVWytoanDHvQoEHHjh1T4YbPDM+WwopD D9q3b89DHK6CH9WtI4bodMHJpGOTm5tLHxWX4LUoIwPspZhKT09nCKo8fvaH O0S3DVllguvWrWMcVJf06xwmBIAuo4MKF4jjbiC4mLJ1zgq8MWJjY3VxVCxZ xA/y8L0NXxE5ZyuDw7GzZs2iOkCukGfkBOWMECgUvY2bvel03WGLrTICFKq4 FpQklZdquYB7z69GycnJiANTHDBgAI/yp4xQRHfffTdC/vjHP/IFi5cntDlC 3n//fcaBqfMQNllCfr7xxhsIad68OSPgdsDCEYEtAkgTxxqvVgN/FkiB4/GN o1Erd6hJEWmHeMrWrFlD5aI8cHUgFAEcLRT+qlWrGKKeHTyANGDoGsSBxdI2 dGVEQcHVOb0+G05KSqIi0zvXubFbRfDKCCWQkZGBvLFVdNSoUSdOnNALB9kz DqElT5482Qhnx7aRI0dyaJLCuWcRrp2PFSRqyeK4eYIcEGVUcRFlZCDKSCgu /pTRjh07VBw4RQh5/PHHuUmfAf6G3odn9erVEb7GJm6KMhJCAo2tW7duKiQ+ Ph4hjRo1UiHwVKv4oByAwwDvRe/thpCGDRviKDVBtIMy6ty5s5EHw2f2XvDJ b7/9djhO+NGqVSvnIRJsx7H2BZoxYwbC4S6qEDyAcMI5EZazMqKq0tuD6BDC b9E/IHNoz+zZs1VRQDXAG3f4EE1fzqPNigCWL1/u8Q05t8ZX4gg+bbFkkWrh glNt7IIjyjw4VOKcWiE6OloP7Nmzp9GAGIwy6tSpk4qAnLB9XPWY6tChAzbH jh2rp/OnP/3JQRnBf6b16uWD29GuXTv16YnXhfurIuBm3XHHHQikFMId5MtW TZiGvOHuO3Q282eBFDiwZz0QqeXn5xuzolFiK9XDA/UWIpgWz6LEC+SP5+IR ZACmqCsjPgJ6V0Ccne1K+pSMbuxWEbwy4kgi1ZKllxsUk8euXyKVCBSQEZ6V lcV0IFJg56gEoXwZYjV7BdvRIKas3SxdxnHzBDkgyqjiIsrIQJSRUFxslVGd OnX0Cggqia1I3KTP8PbbbxtJ4SiE89uvKCMhJNDYPvvsMxXCaZCNUWnsnmSM cdi+ffu8efMmT54cFRV14403KsfS66iMDE/ba6eMwAsvvMBvCDVr1oTz43AJ hYWFdFGsYwE2bdrEXRzHAedKf+4clBEKweMbKq5/neDndwSu1uAoHodeZFbw nOIQYzI3tgj465yDkucEXO5lEV4UbIfCK8jq7qL6Zmpq1JWVW2+9FeVvTCkA nxaBHTt2VCHBKCOjaQBCCYHwSLkJbaIbFZkyZYqDMqKu79q1q0NOeF0oxjgN rsfEllPQvHlzbDZp0mTAgAG4y7YrEiocLJACxxhDROA+7d69e/369enp6dDF 7JypNAIPNIbVQCghkEOicE8plPTmV/DNN9/oyojJIn3daKmM8HQ4XJQDwSsj PNHQ+LjRnNUBz6C6UpQJnzLjkL1793ouHo5EUBo4HPF1/cJReIamNvYOHTrU oetgwDhuniAHRBlVXEQZGYgyEoqLrTJq3LixHgcVjVUZxcTEGEmx7uZ4AVFG Qkigsel9VGyVESxWV0axsbHss2TgRhlZ+8PYKiP4e9WqVUN47969nS9B+aVW F6WoqCgtLW348OHqA/XEiROVzvKnjKAFIiMj4XYa3r7q4GSLddVIf/BbtPFB m5/62b/LliVLlvBEtkrHAEUBX9HjW43FNmMI9OfME76pLr/8ciOcn3GaNWum QoJRRoYxvPzyy0oZqQwYF8vWc3/KqF+/fs59L1WrvS0oEEaDvvj973+vwuvX rz9ixAh/7QsOFkibsa70DV1AUWBgKCPjQLZjUhnhXDzEUG0cMURlpOLYUuKp zkM7zmjz5s3MD1ucoegpSYyYW7du9Vi+J/hD3RHrgKN169YhHA+4Q79BN3EC PkHOiDKquIgyMhBlJBSXEiujDz74wEjqhhtuiLgwOy6VUcuWLfUI6enpooyE YlECZbRp06YqVarUrFmzW7duM2fOXLlyJeoIuschVEajR4+mU9qgQYOAi+Bw 7Lm/aHCTdu7cuXTpUjUSgYMabJXRyZMnGU3v70o45hpOy34LzvMtGNiOSeF8 XBwaY4UjSlBEEyZM8PgmmnMQR7heNjBBwPpbckVNuuUg6FDyVsmzbNkyQ5iU kjIC7FxnpEzT8qeMYmJiIizTOBjUq1cPcWAPqy0Yc4/DGx8zZswzzzxDU3zv vff8penPAm0FDsULDA9Ximdn3759EERs4HOvjAA74KkurIRePZWRalfKzMy0 Gq2blkdbQqiMCOfW4xQryDNH4Rkyk42q1iWNzp07hwfZ2tuWjWXGlCaUVwA/ /GXGTRyvuyfIAVFGFRdRRgaijITiUmJl9Nhjj1njoE7nJirQCN9QaP0zJt6o tsqoe/fupXBlQjhQAmUEQRShTQbi9UkMzlBnKCPD8NwrIzhyNWrUgLVzijB9 NIot9Cqta538tw+1CR+Gk8Kx645VGcG/YgTb4du4Oo+vn4+uSvAb6sO5t5WO +pqtd03kmAXkxHY4lZJFcGXVmCN/4gjXyMno4G0avR91IDc8vgnNHLL67LPP ovCNDn6DBg0yNELpKaMnnnhC3yQwPAdl9M0332DzV7/6lX7fjx8/DjWqmhvw hkQcffrxgKSkpFCk+/N5/FmgrcDhBNR6weJOccHTYikjTl+gN/0oA1a96bg2 6/bt2/V0YK6wwxL7byVWRrhwGLkxqzxg7z51XZz5hBO8EFwXTmctEK/v1ngs 3VDZ8AS5qj+YWVlZLGSHboRu4hA3T5ADoowqLlRGmzdv/ukC+l5RRqKMhICU WBkBNXgBzxrnqtW793PM+9SpU7l58ODBW265xVBGqC8ifJM5lN4FChWaEigj HvLcc8+pvX369KHFKmVka3gulRG8oAceeAAhcPwKCgo42EQfim6Fw8yNb7zQ CJwPWXWJUcqIndasyohzKUNZ2FZtuHx+gVezK0ObcCw5ssqQ06dPIwSXrybY N/juu+88lmEUHHWOvFnjK1mkJu/SJ2QwxBGyzV558C3VTGi2cNy67RkVKL0I 32wGSvXAXaxZsyYC9Tn0Sk8Z8ZCbbrpJ6YXU1NSqVas6KCO8KtnS9M9//lNF 4Ji1Ll26cJNzzjdv3lzNgLF3795bb721Vq1anKkAMgqJIL4qXi4vixesv6uz tUCvH4HDQGgWpg+zpOQprjLikCKPbxwTtEBOTo5KRykjjpcZNWqUulhEY/9S XccFtFsdW2VkTcEasn79eh6oj41iUyz0iBLyHB6IZ415Rimlp6d7fBPWWa0a V8Hed2pAHMqTmlGfuhy3mE1RXCnAFjdxFG6eIAcMZfTzzz8rNxvlI8qoPENl pIPHVu0VZSTKSAhIiZURu3zceeedf/jDH7igBjwEveNEr1696I62aNECaqhO nTrs/aIrI3iqVapUQeBtt93Ws2fP0r9coYJRAmUEL4VG1bJly5deegnmh9/X XnutroxsDc+lMuLMzE8++SQ3oVaYvr+eZt4LcxoYjQveC4ucenxzfM2fP5+r w4wcOZIddQxlxLVLKCumXYzy2SCy+El59OjRCOdQEYSo1WGYE+tXegV9PGNZ Fk5Mt2TJEiPyypUrDVlE9Km89XDKKH4tj7oYYw1WjlpSE5LbAomKl0+Eb10q 3I7WrVtz5Jex5G7pKSPcoFatWjEDzz77LH7DqJo2beqgjLw+A6Z8g9jBi5ed kGvXrq3Ww4LnzGYjvFc7deoEU69bt67eNHno0CEucvTQQw8hn6+88gqbRFGq /q7OnwXaChyqYwoW6CNaEfvjFUsZeX0LLnsuxmgzwk1kGyKsFLsmTZpEAzY+ NQS0Wx1bZWRNwRoCXTN+/HiqHmQAxcV+dB7LoCfONUH9yEOAsUQsgW7iok64 LvxAEbFjKtSf6htZWFioBhtGWdi5c6fLODpuniAHDGWUkJCge9qijMozv/zy CyqL7zR++OEHtVeUkSgjISDz5s2L8C3byk28fiN8Ux7pcfACj9AGDakpvHr0 6MEPpKjoUX0bK12ifodfqtbZfPTRR1NTU/HjmWee0aPBzeNXd/iWpXmhQoWE I9b1L59wxhDSvn17PRosFoFq0ArqcTqcEb7B6XC9YJ8RF89jZjU867nIJ598 EuGbK8zrm2kK7jf8Ui71QjjNst5ganDq1Cn4WvCIjDYUuIUrVqxQS7IOHToU HpdSGZxdSvWHgfPj8YM+yHrHjh24Xvqx9DZ1yQbPFqeDK6XPaq6DF4L1o7Q+ 55gOlJ1VFhGKI2OW78WLF/u7BKOV6rPPPoMk9JdJxenTp3v16sXJ3MA999yD m2XEoTQOOBZMv8teP8bQvXt3o/8eyrZz585qDWK8SNl807ZtWxUHcglGqKcD rxXSifqoevXqjzzyiDFqDDcUb1daNYUV9Lg+YATOPI6iEozwzeCN0na4On8W SP/ZOrEG7jUdeI9PquOi6N4rgWl7INS95+IRcLBwbKLEkD1YEbJNga8LYZgK IrDHGk+HmsLotxnQbnU4gbZhUdYUbNNEZubMmaMWrvX4PjJYF3FGebIfIIGt 6nPvG+BaoBDZ3EMmTJigD/2DIvP3XHguCDc3cXRcPkH+MJTRyZMndU8bj3xZ uPxCCKhAyujdLx77Z2LbgH/vzH3UjTJ6Ynra03PWBvxDNDfK6I3XEvu8tTTg 39uvLxRlVEnQv6bCF83MzPQ3jNrrqxQQwfn9jMo6Ly/PYeEGQSgB8Fr37t3r UAuUwPCQmnXYjm2gAXvg+BsdAEcrNzfXzVotLjl79iy0gO2wIGNwUzBwEJO/ vSgQ52We/MFFMznLpUsOHTpkzA59KcFl7tq1q7jTI+MeQe06D5CHjjBmMNBB CaOsrPOb2eJsgbbALOEDl9haUCBQUkaxLF26FNmYO3euNT6qCYeJ1IK3W2sK /tLETcHbY/fu3baqX4FHDHFwj9y8Q2AkSHPPnj2XoKYrwRNk4DDOSKjQVCBl VKy/gMqoWH8BlVGx/kQZVQb8dToSBMGBhISEqKioEs+4VUkoKCgYO3bs9OnT Q6gTBXKJLZCdJ8eMGUPtBmkAHcEuYdZZDoRQEZInSJRRuFIhlNGZf+XnFhwq 1t+/zvn9Rnem6NyR/DPF+sMh/lL717/O5eaeLtbfmTNuZ0ASKi6ijAShBKA+ WrJkCZyWgGNeKi25ubnjx49PTEws2VTDgjOX2ALz8/PVokjQR5wehF3pHKYl FIIhVE+QKKNwpUIoI0GocCxduvT999+3dr0WBCEgW7ZscTOKvHKyY8cOh8Vk hZBwKS2woKAgKSmJY16gjKCJUlNTRfaWHqF6gkQZhSuijARBEARBEMqW8z7K OheCW0QZhSuijARBEARBEATBPaKMwhVRRoIgCIIgCILgHlFG4YooI0EQBEEQ BEFwjyijcEWUkSAIgiAIgiC4R5RRuCLKSBAEQRAEQRDcI8ooXBFlJAiCIAiC IAjuEWUUrogyEgRBEARBEAT3iDIKV0QZVU74RJferUfi586dc46DCIimL99w 4sSJU6dOlVKWhEsPDKzirldotc9gQDkEfCIqOkePHg1VUsWyHHlvBENlsExB KA1EGYUroowqJ+np6R6PZ8WKFaWR+L59+5B4XFycc7Tp06cj2p49e7h56NAh bA4ePDgvL680chUMOTk5CQkJaWlptntTUlKwNz8/P1Sng6PyxRdf9O/fv0uX Lh988MH8+fORgVAl7pImTZrUr18/mBTgr8IG1q5da4QfOHBg/fr127Ztsx5y +PDhxYsXz5gxY+7cuatXrz579mwwGQgSwz6DZPPmzbGxsbm5uSFJrRzy5ptv RkREDBgwIPik/FmOLeX5vVECCgoK1vvnyJEjemS8c/BSwsMyc+bM1NTUXbt2 2aaZnZ2NaLNmzcJbZc2aNd9//72+N+wtUxBKCVFG4Yooo8rJqlWr4E6gMi2N xKmMPv/8c+do06ZNQ7TvvvuOm3BBPT7g6pRGroIBHgW8vqpVq65bt8669957 78XeHTt2uE8wMzNz4MCBUAHWXXBvbr755oiLqVGjxuTJk0t+AcWnTp06l19+ eYkPh5saFRWFctNfL/gNDy0yMhJ3+dNPPzUO2blzp+dixo8fX4biyLDPIDl/ /nxSUtLw4cND2LBS2jhYqZW2bdvCUF988cUgT2prOQ6U5/dGCThw4IDHP/r7 B5Y5bNgwI8LChQv1BiBYXUpKihFnyJAh27dv1+NUOMsUhPKAKKNwRZRR5aQc KiNU0FlZWaFyREMLlRG4/fbbCwoKjL0lUEZz5szBIb169TLCjx07hlNgV9Om Td94440vv/xyypQpnTt35tnfe++9IC/EPcEoI7xVJvnQ3y3weOPi4pR7Zigj lOqoUaMQPmPGjIyMjLVr144dOxabixYtCuoygiC0ysjrs/D4+Pjo6OjCwsJQ pVmq+LNSW3JycmCrQXrXtpbjTHl+b5SA/Pz8dAt4XQ8ePBjWuHnzZkbD0zR0 6FCExMbGfv3113v37sXLnN8cNm7cqFLDo8THDXtxgxANrzKGnDx5UkWrcJYp COUBUUbhiiijCoSb3uDWOKj1rLfYUEYlS9lfBAdl9N8++LtYnqftVfhLOeQo ZQR69uxp7A2VMioqKnr00UcRfvfddxv6C2WF8OrVqxvdaUqPYJTRggULcGez s7P1wKioKDYDwZGzKiMeMmHCBHUT9+/fTxcO5uTyvO5N1B8u7bPEzwu8U6Q5 e/bsUA1fKlWKpYxCgq3l6IR8REwIE/SX1M6dOxcvXuzvjq9cufKbb75xThnS D8UyatQo1NcMSUxMRMi4ceP0RtWvvvoKgdCVKuTLL79EiN7qh2xATCHQOGnF skxBKA+IMgpXRBmVf1Dhrl69Gn7akCFDRo8ePX/+fH3sQ2FhIbzNmTNn7t27 d+rUqYMHDx4+fDjqTYQfOnQIIUN8xMXF6R6mUkbw5+Pj4xFhzJgxSUlJhw8f Ns5+4MABhKMKjoyMhNhB5atqZ3LmzBmkg+qYdffSpUt37dplVUYIhL5A3pDD KVOm4BIMzxO6ABcyceJE47qQAfw/bNgwHIi93377rZFDuA0JCQlIGVeBlJEg 8oBjcWAwxa5DZXTPPffUqFEDP1Ag+l5bZTRjxoyuXbs2atTo/vvv79evnxpW g9Jr3bp1s2bNcEiDBg0ghdq1a8dd69atQ+A111xj640jGvYOGjSImzExMTg2 PT0dIa1atYKQ2bBhA3fhRsNOEL9evXrY9dprr+mDCJiBHj16IPDNN9+84447 mjRp8sorryxZskQ/HZXRqVOnevfujUuoX7/+7373OzcdqzjuQ3fPCO4O/F44 cpQ8hjJChhFoDJSYO3cuAuE6+jsXduFGwyRSUlI+++wzWiDuDmxSjwZLSE5O RkxmDOZx/PhxI6mA9ukN9CSSgwcP4taPHDkS6eDpQ3xjABoCPaEbvkRoDGvW rBkxYsTDDz+Mm9W2bVu2L+DBf/DBB6+//nqEY69xIMrq9ddfb968OWygQ4cO KHCGO1ipP8PD845wlAl+I+SJJ5549tln9XKePHkyAvv37+/vKvxZjtexVI33 Bgp2vB/0BpGArzVkJuBXiLS0NCS7ceNGPBewZySFa8zMzDSi0ZAgUqyiY8WK Ffxc4HAWHMW3qz7wii2qu3fv1mPiEvACP3bsmArBm9Mqgig/UQUYJyoNyxSE MEaUUbgiyqj8w+9+AI4NXDJ2FId7yb0FBQUIGTp0KPaialY9z6dPnw43j0cx BLuUo0JlFB0drXbxBxxLvZcF6lnugkMCt4Rx5syZo6p4/GDlyzjs8sFuUboy gmpjTw9mlZGZMeV5wmfmpenXhbPzvOoqgC554BuolBkHm3SzQ9jBhsqoc+fO cEjoK+qjmK3K6OOPP2YDE2JSTF133XV0+3GZ8EIhfxBYq1atxo0b43AeBecK gX/9619t84ACwYWrYeZ9+/ZF5JYtW6qBSOvXr0c4MnbTTTchpFq1atBl3AvX V3VzgmpACJzeX//61/hRt25dxMSPKlWqwNLU6eDxIuSpp57CLvjVqsksYPc2 6A7bjprqdliVEW89MLxTti7BkfN3LmSGX85pKjQ/AJtUceDiKvNWNoyz64LF jX16Az2JXp9bznDYCRQEI0+YMEFvTYBXj8DExETnYiwWNIY2bdpQWfNO4c5C 1eLHlVdeSSMEuEx1FK6RgVBSiMPfKEyvo5X6MzzavLqtHHYExc1NlFLt2rVh aXpfLwN/luNcqsZ7A/dr3MXwhgKllwO+1vhtx3jVWIGCVi8l9YJCmkYrJxQZ FBNvui6O2MQDWaS/cq1AankubjDiJeNEqLuRQ1jUsmXLMjIyrB19oYkQMz4+ Xh174sQJvh6tw7JKwzIFIYwRZRSuiDIq59A5RLVLBwz3KykpCSFRUVF0DKgg AKrp06dPI3DdunUMQQ1IBx6VIB0A5ZZQGYGpU6eirvT6ZtzlR3U1pxzONWbM GNaVrFgPHjxIB0ANBOYcd6ijUX3DkOADwLFhykoZoR5nTyr4q5xcF/lhygGV EYAvxBzCB6BfNG/ePMaBq8P8zJ07l44ufB76yaWkjODYUCx06NBB7TWUUUJC AtUEv6WjTN544w3KE3WIbT+lv/3tbwhEWbnJEh1UtiLpX4mZSMeOHekHZmdn UwGp20plRId5+fLluGu4Ke+//z5CrrrqKuXUQRkhBC4xrwuFiUtGyG9+8xvn jLHtxqHwrcqIjQXWC6dP6DD1BJURwGOCC8GlKcNmmcBu6QfOmjWLtodwWpEq EJf2GfBJ9PrsxOMbG0UHODc3l/apf9vHQ8Rn07kYiwWNoWrVqrA9r++mK+Xy 4YcfFhUV4br69++PzRYtWvAQZOmyyy6rXr06Z1xEIVCYN23aVCVra6X+DM9Q RrhMCitOgNm+fXu9xdMWf5bjXKrGe8MAe3m78aZiiJvX2rZt23j3rS3UCtx0 6jW8NvnZAS8fpmxtWoVlsg+bEkcuZREiQwN6Lm4w4vMSHR3NPnU6mzZt0g9H jcDBfShbFGBycrIazWc9V2lYpiCEMaKMwhVRRuUcVGGorbZs2aJCcL/YlYIf /ZSCUK0YiEBvEL6cOop1qOoxRQcSckn/zIianUmx7zo9WNTd+ndOaCuPrzs6 N9kBQ3kdXl9VzjwrZbR3715WuHofJzWKJKAyQn2tjqIQg4/BTRzr8X1J1jvb s8N8KSkjr89/uPbaa7GpsmEoo+effx6bOEQvkzvuuAOBqqeKrc959913I3DZ smUqBB5XzMWojnx0UJklHahIGIbe+II7iJh///vfuamUUXx8vH7gk08+iUB1 16iMli5dqiKg8BFy9dVXO5cVv8Y79ESyKiOOE7d2o6KrBiv1lxSVkd5CBH+V GaBPy3PB9vSeVPDndTt3aZ8Bn0QVB06vioM04frqS/PgBjHlEL54aQzdunVT Ibi5CGnUqJEKwR2p4oO2gYJCtvXebghp2LAhjlJdah2UkdXwDGUEoEYjfJOW cJRcq1atnAf1+LMc51J1VkZUVXp7kJvXGvK5efNm3GuHQTfqPaN3VV2+fLk/ 3aHEER5hl7IIbN++ncapP9FKuHl83w3WrFmDVz1fxcDo1rtr1y69wZ0qyXaO 7tKwTEEIY0QZhSuijMo5bOtBhbtag/4YPw+qXmf6UfxOqA9k5hdv1RWKykh3 KQm/efKrOA+Jjo7WT52cnIxARGN81rmGM7N+/XpdGbENSw1hMC7NWRlhU/em 6Cqrbvns/qGakBSl15uOm7gWbNauXZunMJTRrbfeik1kMk7j/vvvRyA/6Xv9 +JytW7dGoH45KMmIi2nTpg130UHF3bHNcH5+PhcGQgQ4zIjZtWtX7lLKiC1x ilGjRiHwrbfe4iaVkd7lDC8KXDICHdxFCBA6Vw4rb1qV0e7du2lpRkxqFvbv soXKyBiIxO6dHNhFG1bFrmDzKNL3urbPgE8iwA/dX4W5WssKr1zGCeECWDQG eLwqZO3atQhRg4MIe1cag7Dge8PkkOGoqKgbb7wxoH73Z3hWZQReeOEFGlvN mjWzsrIcLsHBcpxL1UEZoRBoV7oudvNac8OGDRs8vk7LeiDfq/46pKHkOWLI pSzCE8d2NGOxAKg2JoIHXIluRKYM1BeSo+JjPtPS0lJSUiZOnMjisi6RVhqW KQhhjCijcEWUUXkGNZTHP6jmvBcUhPFdncqIjh+xVUbGoHuv71Oz58KHcdVV yRbUyCp7xmr1xgwMTIe51WEvfWdlZFzX0aNHdWXElOGsGimzV3/pKSMAoYGQ Rx55BI+ProyU7rAFueLhtj4nVAkC4WGqkGPHjk25wOuvv469zzzzDHfRQbX2 NIPu6NKlixo2ojCUUcOGDY0Dk5KSEN62bVtu2s5N16BBA5fKyMG5siqjvLw8 j2+AjxFz69atVudThwagt416LzQTUBn5sz02Q3Bkuhv7dPMken2zAcD/5Pg+ AkfUUASIw10O4rG4WI3BVhk1btxYV0axsbFspjRwo4yshmerjGDAHMXWu3dv 50twsBznUvWnjLKzsznu0ljYNOBrzTmfCg46M+YxmD17NgLVXChW8MrliWCl AWeBo/0bDUa8NCZiTOLHXnYoCl4FipQj73Rhde7cufnz53vsmmhLwzIFIYwR ZRSuiDIqz6DqZA+TzMzM/Rb4yTEYZWT9Tg7HRlW4bJGBX2o9NQcmI3sc3awP N/Be+FCplJHLlp0SKCNelPUqrGPng8SqjE6cONGkSZMI35B2o82oXr167Ie2 2oLqGWjrc7LTUfv27W3z8NFHH0X4ho1w05+D2rFjR4Tfd999yBjcMDhp0dHR VmVUtWpVY+2SCRMmILx79+7cLJky8vrG9aPwHda1sSojJMghG4ZXTCt1mPMh oDJas2YNfnPCNB2OPaGFuLFPN0+iAgW7c+dOGADHdAC9eQ5uZ3Gd8ICUQBlt 2rSpSpUqNWvW7Nat28yZM1euXIkSu+uuu0KrjFCGVFuwnIBLHTlbjr9StVVG uCOMZp1LP+BrzSW2o6JoM9bmGAJ9R8vn0KHk5GSHRwn1Mls2DfP2XviSAIwp FpW65Dwt7GyMV6hxFr5FPRd6kypKwzIFIYwRZRSuiDIq51Dj6EuWe33dHlAJ 8sYFo4yMr5Gqnzl9Jy4uD79ar1jxG6dWR/G7ujHsF06mrozYVwr+gN4vTo31 CEYZceEkxNFdU679UdrKyOvzc+BbQj6w45nywZ5++umIi4cLWaHPqWQIgeNH N1KfI46gcCB2sEs1kNk6qLjjCESW9O/kKAqrMgL6wA3w8ssvR/g6AXKzxMqI HTIdFiGynbWbUoWzARBYC3W61TNUBFRG/Lpu2B70F53w06dPe13bZ8An0WtZ VAseJh8Qvb8fbo3H0v01SEqgjNjHUnWe9PpsjE2NhjIyrNS9MoKKrFGjRr16 9Xr06IFdnTp1cr4Kf5bjXKpWZYT7yAj6KDmFm9daQJQG0bsm8v1pdABWKFmE l5Uac+QgjtiHEK9oq05RI0m3bt2qh7OxXk2hQGUESzPyw6alyMhI49tIaVim IIQxoozCFVFG5RwueDFq1Cg1ZjYnJ4cdS+hCBKOM+KlcTX5Ld2LChAmMg+qb 38mVN45KnGthTJs2jSEccYz8qKFGalCAUkbIIZuW4AbQ2JAyv5oGqYxgvXSe 4afB4Tl16hT8YdXrprSVEejXr5/qg6SUESdDbt68ubpluBG33nprrVq11Pj9 xMRExLnrrrtsE7zqqqv0qahQFJwDuX79+nTmvX4cVDhd7LykHHh4p5zE26qM brnlFpVD3BoOz+ciON4glBE7qhk+m46tMqLZwN6YJZyCs21ERUVZ5yJWBFRG 8P1owwsXLqTtwWbYZVTZuUv7DPgkwoNlOmpCM+XD692rqNwR7lCGxaUEyoiH PPfcc2pvnz59jN50tlbqUhnhrfLAAw8gBO8KlPDNN9+M39aBjTq2lhOwVK3K CPfa45uq2rZqc/Naw1OGEFy+0aqioOgwBltxIhHbO6tkkWo91CdksD5QKD3O senvswDf53g61DeQvLw8jnpTLaRqNhIoRHUKPBHTp0/3+GYlNdIsDcsUhDBG lFG4IsqonIPajY5cZGQk6qxJkyZx4RXlYwTZZkR3FH4OuzPhtz6dArwRng4x 4TZwcmOEqDVcYD9qTDGcSX71pbuor2ekJlPCLpyLX+wZLRhl5PV959QHINB3 pVy6BMoIl8+udLoygufGZqPrrruuU6dO8E7r1q1rfDOH8wkZgsDbbrutZ8+e KhwODJxVJtioUaMuXbq0adMGkgqbN9xwg76IpD8HlZOK44wdO3bEqWvXrs3V iAxlVLNmzSt9IKvwYJkZz4VhUN4glBFHplubvRS2ysh7YeYEj28IOfsRAXib DucKqIy8viYhZW+wVdo57ErvAurGPgM+iV6fwGQ6CISDSucW59J7CXKkiT6d Y/CUQBkhA7zpLVu2fOmll1q0aIHfnHRRKSNbK3WpjAYMGIDNJ598kptQK0zf X08zr3/LcS5V472hxuDExMRMuxjVvhPwtcaceCxNhArKdmOlLX4msg7eXLly pSGLiD6Vt21R4HXqr2MbwtVcCjBLyBzqR5xFLXmm8unxzUc3b948PBp85yOy saqyt3QsUxDCGFFG4Yooo/IPKlBUnfweSJcgNTVVdZA4ffq0x7IWDF04vV8K q1pVa7PGhEuJWpt1JReXNxZVB3D4kRp9RfqEhm+DihgVrpoYdvbs2XRO9CmS vL5p1pSGgnjZvHkzO90pN4wzI6m+HLbXlZuby1peD4RcwuXAw4EWW7x48cmT J3n5IVRGyCr8uhdeeMG6C+VTvXp17NV9+FOnTvXo0YMDkSJ8axvBUTScHHhu /JAOd1EPLywsfPXVV6mGyDXXXAMX1/Bk2Lpk/cAL76tDhw4QNTz2j3/8I51S tYAslRGcZDhyt99+O6M1a9bso48+0tOBtsJ1GYlzVmdnZYRrx02EUfmLxgkG rZOb4UAqegKXz+jvZ4W+nDHgnQPM9QEmUJQwD2oi6B2YqD4VPAlon95AT6LX p55WrFihFg+F/wln3vCHYb3IiTFCJEisxoBHO8IyZo0GqfpQJSQkQG7TAOrX r4+nBmaG3/q4fquV+jO8Tz75RPXGhNKsVq1ajRo19PcJZ7OHYPd3Ff4sx7lU jffGzp07PX7QJxZwfq0dOXIEp8PLx99tom1wiVuFPimiDpSdVRYRiiPrLN/8 tmOkbwBJiGyoFYr5AreeBVeqFucikFTWZV69pWOZghDGiDIKV0QZVSBQZ5XS rEFQE0afcwPUwhAgDsuRwIqQPX+dTxTQO6i7A07K5BKk870PPcH8/Hz6wA5L 6lwy4GSq1WGsINvQlaqDnA5uNPwi+OFwbEpQXLgRWVlZttMCK2XETTiEetti SGDXOGP0mUtgZnCnUW62xRIMUKYobefCdGmfAZ9E7IWKt6bDtWv1rpJlDg3A oRZwsFJ/cNFnY9iObaCBs+X4K9US4/BaMwY3BQMHMfnbiwJxXubJGVzCvn37 9u/f71yweCvimUJMf6/ocmiZglDOEWUUrogyEiouqOXZkT4tLY1mDBeOa9rK Su7+MJRRKZGQkBAVFRVwxZZKRUFBwdixY6dPnx5C3z78EMu59IhlCkIJEGUU rogyEio0HNrMzlccgMCOJWomAcHg0igjvFWWLFkCd8tYTabSkpubO378eMh2 mRLZGbGcS4xYpiCUDFFG4YooI6Gik5GRERcXFxUVFRkZ+emnn86dO9dhymjh 3Llz77///ogRIy7BubZs2eJvDHtlY8eOHQ4LgAoGYjmXDLFMQSgZoozCFVFG QtggliwIgiAIwiVAlFG4IspIEARBEARBENwjyihcEWUkCIIgCIIgCO4RZRSu iDISBEEQBEEQBPeIMgpXRBkJgiAIgiAIgntEGYUroowEQRAEQRAEwT2ijMIV UUaCIAiCIAiC4B5RRuGKKCNBEARBEARBcI8oo3BFlJEQPLCic+fOlXUuhHIH 3i1FRUVlnYsSApOGYZ8/fz4kqaEcwv4ZOXr0aKiSKpblnDhx4tSpU6E6dWWj MlimIJQGoozCFVFGQpDs27fP4/HExcWVdUZKkZycnISEhLS0NNu9KSkp2Juf nx+q08FR+eKLL/r379+lS5cPPvhg/vz5yECoEndJkyZN6tevH0wK8FdhFWvX rjXCDxw4sH79+m3bthnheBFlZWWhMKdPn75o0SJEKFtVhWzAsPfs2ROS1DZv 3hwbG5ubmxuS1Mohb775ZkRExIABA4JPyp/l2HLo0CHcpsGDB+fl5QV/6jKn oKBgvX+OHDmiR8Y7By+luXPnzpw5MzU1ddeuXbZpZmdnI9qsWbPwVlmzZs33 33+v7w17yxSEUkKUUbgiykgIEiqjzz//vKwzUorAo4DXV7Vq1XXr1ln33nvv vdi7Y8cO9wlmZmYOHDhw8eLF1l1wb26++eaIi6lRo8bkyZNLfgHFp06dOpdf fnmJD4ebGhUVhXLTXy/4DQ8tMjISBvPpp5/q8SGC4OB5LiY+Pv7MmTMlv4bg mDZtGvLw3XffhSS18+fPJyUlDR8+PIQNK6WNg5Vaadu2LQz1xRdfDPKktpbj AKQrrQUSKchTlwcOHDjg8Y/+/oFlDhs2zIiwcOFCvQEIVpeSkmLEGTJkyPbt 2/U4Fc4yBaE8IMooXBFlJARJ5VFG4Pbbby8oKDD2lkAZzZkzB4f06tXLCD92 7BhOgV1NmzZ94403vvzyyylTpnTu3Jlnf++994K8EPcEo4zwVpnkQ3+3wOON i4tT7pmhjODRIXDEiBEbN26Ec7hhwwZ6fQsWLAjqMoIgtMrI63NBofWio6ML CwtDlWap4s9KbcnJyYGtBuld21qOMyjVrKysEN6msiU/Pz/dwqpVqwYPHgxr 3Lx5M6PhaRo6dChCYmNjv/76671796ampvKbA54glVpGRgYfN+zFDUI0vMoY cvLkSRWtwlmmIJQHRBmFK6KMKhBueoOjjgt4QwOm4+ZEKo6zMnKTFIwwYJyy RSkj0LNnT2NvqJRRUVHRo48+ivC7777b0F/w0hFevXp1oztN6RGMMoKcgUlk Z2frgVFRUQgcP348HDmrMho9erRxyPbt2xGCo9yP9AnesP/bB387KKNiPSA6 8E6R5uzZs0M1fKlUKZYyCgm2lqMT8hExIUzQX1I7d+5cvHixvzu+cuXKb775 xjllSD8Uy6hRo9SrMjExESHjxo07e/asivbVV18hELpShXz55ZcI0Vv9kA2I KQQaJ61YlikI5QFRRuGKKKPyDyrc1atXw08bMmQIHMj58+frYx8KCwvhbc6c OfPAgQP4f9iwYYMHD544ceK3335rpHPw4MGFCxciMmrA6Oho1K3G0BikkJyc zAioXpcuXXr8+HEjkTNnzqSmpmIva2rE2bVrl1UZIamkpCRU3JGRkdiFKluX P3AGcJbMzEycbsyYMaH9Ml8aUBndc889NWrUwA9cmr7XVhnNmDGja9eujRo1 uv/++/v166eG1aAcWrdu3axZMxzSoEEDSKF27dpx17p16xB4zTXX2JYGomHv oEGDuBkTE4Nj09PTEdKqVSsImQ0bNnAXtCrsBPHr1auHXa+99po+iIAZ6NGj BwLffPPNO+64o0mTJq+88sqSJUv001EZnTp1qnfv3riE+vXr/+53v3PTsYrj PnT3jMB64ffCkdu/f7+hjCADx44d+9lnn+leGU7Nj9sOg+tpSDD1lJQUHE6b xN0x+uC5MWyYMe7y8OHD8fhMmTIFj5hVGTk/iQRPGW79yJEjkc7UqVMR33jK EBjC4UuExrBmzZoRI0Y8/PDDuFlt27Zl+0JcXNyDDz54/fXXIxx7jQNRVq+/ /nrz5s1hAx06dJg7dy7DHazUn+HhMUc4ygS/EfLEE088++yzejlPnjwZgf37 9/d3Ff4sx+tYqkVFRbizeONxEwU73g96g4jzC4qZCfgVIi0tDclu3LgRzwXs GUnhGvFaM6LRkCBSrKJjxYoV/FzgcBYcxfetPvAKzwtCdu/ercfEJRw+fPjY sWMqBDWCVQRRfq5atco4UWlYpiCEMaKMwhVRRuUffvdjXyO4ZOwoDveSe+FV IgSCCG4D46gOS6j9VSKoMeH1qXT4A9Wr8pnhCahe60yK7qvu16GOZlXr8Q15 ZgcP+KKGMsK5mBQiqKTmzJmjHINFixbp2cDl7N27t9TLMQiojDp37owSo6+o j2K2KqOPP/6YDUyISTF13XXXcXw0pAG8UMgfBNaqVatx48Y4nEehtBH417/+ 1TYPuNG4oWqYed++fRG5ZcuWaiDS+vXrEY6M3XTTTQipVq0adBn3wvVV3Zyg GhACp/fXv/41ftStWxcx8aNKlSqwNHU6eLwIeeqpp7ALfrVqMsO9cy4r6A72 3jHClcSwKiNbsrOzEQ1G6/ARm4YE/5aPAA0SwEpVHDeGDfNjTyTATkpIivap KyPnJ9Hrc8sZDjuBgmDkCRMm6K0J8OoRmJiY6Hz5xYLG0KZNGypr3incWaha /LjyyitphACXqY7CNTIQSgpx+BuF6XW0Un+GR5tXt5XDjqC4uYlSql27NixN 7+tl4M9ynEsVWeWNYGTcr3EXwxsKlF4O+ILi1x7jFWoFClq9bNXbDGnu27dP jwZFBsXEm64bM5t4IIv0jm1WILU8FzcY8ZJxItTdyCEsatmyZRkZGdaOvtBE Ht94PXXsiRMn2D5rHZZVGpYpCGGMKKNwRZRROYddj1Dt0gHD/UpKSvL4ehnR MaAyAvAZUOt5fZ9D6T/MmzePieAo1oYzZsxgLYxqkd/Yk5OT9QizZs3iJ/pj x44xEX3SufT0dNbIqKxhNqjx4cbw7EoZISk2A6GGZXV88OBBug1q+DAdWrBi xQprbV4OUcoIjg3FQocOHdReQxklJCRQTfBbOkrpjTfeoDxRh9j2U/rb3/6G QNxZN1mig8pWJP0rMRPp2LEj/UDoCyogdR+pjOgwL1++HPcRd/z9999HyFVX XaWcOigjhMAl5nXB4cQlI+Q3v/mNc8ZoVw6NgG6UEWybfX4WLlzoEE0ZEh4T XAgubdWqVQxhmbgxbPiZ7OkHPcU4eDpow/qFBHwSvT47QQhyRQc4NzeXlq9/ 28fjgBDkyrkYiwWNoWrVqrA9r++mK+Xy4YcfFhUV4br69++PzRYtWvAQZOmy yy6rXr06Z1xEIVCYN23aVCVra6X+DM9QRrhMCis849hs37693uJpiz/LcS5V QxkZYC9vN95dDHHzgtq2bRvvvrXlXYGbTr0GacPPDhBcTNnatArLpD0rceRS FiEyNKDn4gYjNq6x3d9zMZs2bdIPP336NAf3oWxRgHjb80MWKgLruUrDMgUh jBFlFK6IMirnoApDbbVlyxYVgvvFrhT86KeUEeo1FYcSBnUxN+mLok5ERani IP7s2bPZMYwRUCfqHU7g9jBl1ZWd3S2Uj+H1VdzMoVJGTAo1vv51dOPGjR5f J3Zu0qGFsxqaMip9lDLy+srt2muvxaYqXkMZPf/889jEIepwFMUdd9yBQNVT xdbnvPvuuxG4bNkyFQKPK+ZiVEc+OqjMkg7UMQxD7xqEe4GYf//737mplFF8 fLx+4JNPPolAdR+pjJYuXaoi4L4j5Oqrr3YuK36Nd+iJ5EYZsXXGsFgrNCS9 hQj+KjNAn9aNYe/du5dx9D54PFB31AM+iSoOnF4VB2nC9dWnH8cNYsohfPHS GLp166ZCcHMR0qhRIxWCO1LFB20DBYVs673dENKwYUMcBQ+fIQ7KyGp4hjIC eMAjfJOWcJRcq1atnAf1+LMc51J1VkZUVXp7kJsXFPK5efNm3GuH9koOzAF6 V9Xly5f70x1KHOERdimLvBdG28E49SdaCTcwefLkNWvWQLbz5QyMbr27du3S OxJQJdnO0V0alikIYYwoo3BFlFE5h509UOGu1qA/xs+DVEZwDHSvg1//VPd1 fu5WgwisMAI/OOtwaIbq6sYa1nBd1q9frysjJhUdHa1nODk5GYFjxoxhHDq0 qNCDKppLiK6MAEoSm7Vr16bbbCijW2+9FZsoujiN+++/H4GqhG19ztatWyNQ tfR5fWUbcTFt2rThLjqoKGfbDOfn5y9evBgeGiLAYUbMrl27cpdSRmxhVECG IPCtt97iJpWR3uUMLwpcMgId3EUIEDpXDoODAiojDr4YOnSoc18m7wVDWrly pR7IDp8c2OXGsNetW2f7dPDRU8oo4JMI8EP3V/EYWssKr1zGCeECWDQGeLwq ZO3atQhRg4MIe1cag7Dge8PkkOGoqKgbb7wxoH73Z3hWZQReeOEFGlvNmjWz srIcLsHBcpxL1UEZoRD4LtJ1sZsXlBs2bNiAQ6ZPn64HssnSX4c0lDxHDLmU RXji2I5mLBYA1cZE8IAr0Y3IlIF6Kz8VH/OZlpaWkpIyceJEFpd1ibTSsExB CGNEGYUroozKM6ihPP5BNee9oIzgtukHHj16VFdGdCDh1Pk7ESMwQR1+reUA XpUZY/1NYwYG1cHJFh7LOHBRgiyfS4ahjACEBkIeeeQRPD66MlK6wxZcNQ+3 9TmhShAID1OFHDt2bMoFXn/9dex95plnuIsOqnWRI+iOLl26qGEjCkMZNWzY 0DgwKSkJ4W3btuWm7dx0DRo0cKmMHJwrZ2VEnRIZGenQkUlha0j0D6mM3Bi2 vzgcRUJl5OZJ9PpmA4D/qQb0ATiihiJAHO5yEI/FxWoMtsqocePGujKKjY1l M6WBG2VkNTxbZQQD5ii23r17O1+Cg+U4l6o/ZZSdnQ0rGjZsmLGwqZsXlBvY rGnMYzB79mwEqrlQrCxZsoQngpUGnAVu69atHkuDkffCEDyPZRI/9rJDUfAq UKQceacLq3Pnzs2fP99jN9NFaVimIIQxoozCFVFG5RlUnexhkpmZud8CPzm6 UUb8dqr37zJYs2YNInBeKR120adziMxwLLM+uMB74bOkUkYcxjt9+nRrhlUT QBgooxMnTjRp0iTCN6TdaDOqV68e+6GttqB6PNr6nOx01L59e9s8fPTRRxG+ YSPc9OegduzYEeH33XcfMgY3DE5adHS0VRlVrVrVWLtkwoQJCO/evTs3S6aM vL5x/bi5DuvaOCgjuoIAPxxOoQiojNwYNi1Wb6ojHKCkjD/gk6hAwe7cuRMG wDEdQG+eU3PuuXfCA1ICZbRp06YqVarUrFmzW7duM2fOXLlyJUrsrrvuCq0y QhlSbcFyAi515Gw5/krVVhnhjjCadS59Ny8oN9iOiqLNWJtjCPQdLZ9Dh5KT kx0eJdTLbNm0vifz8vJYAsYUi0pdcp4W5I1Vg3EW1g4erZs0KQ3LFIQwRpRR uCLKqJzD8bP6kuVeX7cHVIK8cW6UEZccGjNmjP7tEbXnV199xYG9/AiJKlvv kpefn09fRY314Fd0Y5AvXEpdGXFJenjjenWM38iwOnsYKCOvz8+Bbwn5wI5n ygd7+umnIy4eLmSFPqeSIQSOH91IfY44Ah8GYge7VMOfrYO6d+9eBCJL+ndy FLVVGQF94AZ4+eWXI3ydALlZYmXEQejG9Fw6/pRRVlYWJ4gzDMyBgMrIjWHv 3r3bGkeNRVKub8An0Xvxckhe30d4PjJ6fz/cGo9vJj2X1+iGEigj9rFUnSe9 PhtjU6OhjAwrda+MoCJr1KhRr169Hj16YFenTp2cr8Kf5TiXqlUZ4T4ygj5K TuHmBRUQpUH0rokcp2N0bFYoWQTVpsYcOYgj9iGE1LLqFJQGJZjxAYHN92oK BSojWJqRHzYt4Vkzvo2UhmUKQhgjyihcEWVUzuGYi1GjRqkxszk5OexYQhfC jTKC88AhQvPnz2ctifseHx/vuTC8AlUkP4kvXLiQ9qAiTJgwQSXL8cU4uxpq pIYAKGWESp9JKR8eVT9X0Jg2bRpDwkMZgX79+qk+SEoZcTLk5s2bq1sGwXLr rbfWqlVLjd9PTExEnLvuuss2wauuukqfigq3mHMg169fX6lUWwcVThc7LykH Ht4pJ/G2KqNbbrlF5RAeGofncxEcbxDKiB3VHBp9bJURiogzfXEKaJcEVEZu DBvFy8ZQFALjoIj4VV9XRgGfRHiwTEf1A1Q+vN69iqt2Itz9ZQakBMqIhzz3 3HNqb58+fYzedLZW6lIZ4T3zwAMPIARPPUr45ptvxm99rgwrtpYTsFStygj3 2uObqtq2anPzgsJThhBcvtGqoqDoMAZbZWRk+LuzShap1kN9QgbrA4XS41z0 /l6SHC0VFRWlvoHk5eVx1JtqIVWzkUAhqlPgiZg+fToCp06daqRZGpYpCGGM KKNwRZRROQe1Gx25yMhI1FmTJk3id3XlY7hRRl7fx3N+J0ddCa+GXihcDrUa y+7duxkBzh7OwghIVu87B2tRI4jhOvIbL51DfT0j+DDM5OjRo+FscEpkhKhz hY0yQoGwK52ujOC5sdnouuuu69SpE7zTunXrGt/M4XxChiDwtttu69mzpwqH AwNnlQk2atSoS5cubdq0gaTC5g033KAvIunPQeWk4jhjx44dceratWtzNSJD GdWsWfNKH8gqPFhmxnNhGJQ3CGXEkenWZi+FVRnBW1OjSKIs7Ny5019SAZWR 151hq8m+EAdFquLryijgk+j1CUymg0A4qHRucS597AxHmugTPAZPCZQRMsCb 3rJly5deeqlFixb4zUkXlTKytVKXymjAgAHYfPLJJ7kJtcL0/fU08/q3HOdS NZSRGoMTExMz7WJU+07AFxRz4rE0ESo4+SfUkx7ID0fGisle33rEhiwi+lTe tkWBF6y/jm0IV3MpwCwhc6gfcRa15JnKp8c3H928efPwaLCxCZG5vJpOaVim IIQxoozCFVFG5R9UoKg6+T2QLkFqaqrqIHH69Gn6k/ohubm5rA31wKysrOnT p6uFLOHdGYv9wfGGn0DXEW7h7Nmz9ZnACapdVK9qGljEoSuiT4gEIBNQX9PD pCepe0Ssgh3GKZc34FTAr3vhhResu3Cl1atXx96MjAwVeOrUqR49enAgUoRv bSM4ioaTA8+NH9LhLurhkAmvvvoq1RC55ppr4OIangxbl6wfeOF9dejQAaKG x/7xj3+kU6oWkKUygpMMR+72229ntGbNmn300Ud6OtBWuC4jcc7q7KyMcO1Q 33DA/EXjxIn693Y187wt/rxTrx9D4gBzfYCJG8Nev369Uv2jRo3avHkzu4kq meAN9CR6feppxYoVavFQPGtw5g1/GE8lcmKMEAkSqzFALUZYxqzRIFUfqoSE BMhtGkD9+vXxwMLM8Fsf12+1Un+G98knn6jemFCa1apVq1Gjhr6QE2ezh2D3 dxX+LMe5VDmjmuoDBh3tz5D0iQWcX1BHjhzB6fBS9XebaBtG+6Y+KaIOlJ1V FhGKI+ss34hvTd8AkhDZUCsU41qgj6xnwZWqxbkIJJV1mVdv6VimIIQxoozC FVFGFQjUWcHPGgT//OjRow4LiyAC5I+z9wubQWb8dTVRIILzucIeOJlqdRgr KGQUte2SPbjR8Ivgh8OxCTiHlRWUPISw7bTAShlxEw6hmpU9VLCPpfvhQpcG N4aNewHfMmCBB3wSsTc3N9eaDjSa5+JVO8scGoBDLeBgpf7gMtDGsB3bQANn y/FXqiXG4QVlDG4KBg5i8rcXBRLMGxKXsG/fvv379zsXbH5+Pt5FiOnvpV0O LVMQyjmijMIVUUaCUKkwlFEpkZCQEBUVFXDFlkpFQUHB2LFjp0+fHkLfPvwQ y7n0iGUKQgkQZRSuiDIShErFpVFGeKssWbIE7paxmkylJTc3d/z48YmJiTIl sjNiOZcYsUxBKBmijMIVUUaCUKk4d+7c+++/P2LEiEtwri1btjiMEqpU7Nix owINrCtzxHIuGWKZglAyRBmFK6KMBEEQBEEQBME9oozCFVFGgiAIgiAIguAe UUbhiigjQRAEQRAEQXCPKKNwRZSRIAiCIAiCILhHlFG4IspIEARBEARBENwj yihcEWUkCIIgCIIgCO4RZRSuiDISBEEQBEEQBPeIMgpXRBkJgiAIgiAIgntE GYUroowEQRCE4Dl69GiokkKtVFRU5DLyiRMnTp06FapTVzZQzufOnSvrXAhC xUOUUbgiykgQ3AMv4osvvnjvvfe6du366quvRkVFHTx4sKwzJQhlz5tvvhkR ETFgwIDgk4LSiYuLW7t2rZvIhw4d8ng8gwcPzsvLC/7UZU5BQcF6/xw5ckSP nJ+fn5aWNnfu3JkzZ6ampu7atcs2zezsbESbNWsW3l1r1qz5/vvv9b2bN2+O jY3Nzc0txasShHBElFG4IspIEFySlZXVuHHjiIupWrXqoEGDSpBaZmbmwIED Fy9eHPJ8kpiYGKR/5syZUkpfCHuKZaJt27bF4/Diiy8GeVIInKioKPjwLium PXv2eHxAIgV56vLAgQMHPP5Zt26divndd98NGzbMiLBw4UK9Aej8+fMpKSlG nCFDhmzfvl2Pk5SUNHz48BA2+QlCZUCUUbgiykgQ3HDkyJFbbrkFvl+TJk1e ffXVOXPmwJd4/PHHL7vsMgTGxsYWN0GkgAN79epVGrkFd911F9I3vg8LgnuK ZaI5OTlTpkwJ0rtGfTTJh/taCY59VlYWZEIw5y0/5Ofnp1tYtWrV4MGDIWo2 b97MaNCPQ4cORQjePF9//fXevXtTU1MjIyMRsnHjRpVaRkYG1RD24gYhGiQn Q06ePKmioQzj4+Ojo6MLCwsv9QULQoVFlFG4IspIEAJy9uzZBx98EF7ir371 K7gu+q7Ro0cjvFq1aikpKcVKU5SRUM4pbRO1smDBAjjt2dnZ/iKEfERMCBP0 l9TOnTsXL14M9WG7d+XKld98841zypB+KJZRo0ahvmZIYmIiQsaNG4dXk4r2 1VdfIRC6UoV8+eWXCNFb/ZANiCkEGieFbkLg7Nmz/eVTEAQDUUbhiigjQQjI mjVr4CJec801tp+mu3Tpgr3PP/88Nw8dOvToo4/+4x//0ON8++23COzTp4/X 92G8devWzZo1w1ENGjRAeLt27RgtJiYGmzjdwoULH3vsMez97W9/O2TIEF3g BEyfiVSvXh3pQ9Dh97x580JZHEI5Q5nNiBEjHn744fr167dt25btC3FxcbCB 66+/HuHYaxyYlJT0+uuvN2/evEmTJh06dJg7dy7DA5poenr6oEGDWrVqVadO nQ0bNiD8888/R/j8+fPxGyFPPPHEs88+e/z4cXWuyZMnI7B///7+roIjhnTH XnHw4MEZM2aMHDly8ODBU6dOXb16tfpAUVRUNH78+IkTJ3Jzz5494/2gN4gc OHAA1w5lERkZiZxDUyjRoTJjDOqxkpaWhmQ3btwI6fHpp58iKVxjZmamEW3a tGm4LogUq+hYsWIFdiERh7PgKJQJoukDr8aOHYuQ3bt36zFxCYcPHz527JgK mTlzplUEUX6uWrXKOBEKFuEoQOerFgSBiDIKV0QZCUJA4EHBRezatavtXviB 2HvbbbdxE+oJmw899JA1ztNPP+31tUDBEYXOQkitWrUaN2587733Mlrfvn0R +Je//AX/V6lSBQ4tRzPBrVUfhwOmP3ToUKTJbn4NGzbEb7jHISwNobxBs2nT pg31O22mbt26vXv3xo8rr7yyRo0aDIT3ro6CymAglBTi8DdM3evCRFu2bMn4 SHn9+vUI//jjj7EJgcBoHHb0yiuvcHP//v21a9euVq2a3tfLYOnSpez3ZYTD Vx8yZAh2QQ5Al7Ez2IQJE9hGg6xy7Awj4+kYdzHsdQbUmDvIB47QQQlAbXHv nDlzlHLZtWsXAyGgHIp9ypQpavAOVCd/IM19+/bp0aDIoJiwKzExURdHbOKB LNI7tlmB1DIajHjJOBHqbuQQOnHZsmUZGRkFBQXGsdBEiBkfH6+OPXHixOjR o22HZSEdZtIhM4IgKEQZhSuijAQhIH/961/h5kVFRdnuhedTtWpVRIDX4XWh XIhtVyW6neCll15iO9GWLVtuuukmhLz11luM4zJ96U1XeaDZwAgTEhK8vrnI lHL58MMPi4qKTp061b9/f2y2aNGCh+zevRvauXr16mlpaV6fsw1RgwhNmzZV yTqb6KBBg/TmCUMZHTx4kMJqxYoV2Gzfvj0PcbiKzz77DJ65tVmWQ2MWLVpE WZGbm0sZwhYTQxkZYC/FVHp6OkNQ5Y0ZM4YSgHoBWWWCan6Dbdu2UeZ8++23 /nILXUa9BmnD0VUQXEzZOmcFRBn7sClx5FIWITI0oNFgxMa16Oho9qnT2bRp k3746dOn4+LiEI6yRQEmJydDYWFzxowZ1nOhHLALuskhP4IgKEQZhSuijAQh IHfffTf8umXLlvmLcOeddyICvB1vKJQRRI3e82f16tUIvPnmm7kpykgwoNl0 69ZNhcTHxyOkUaNGKuTIkSNVfFAOwLGHg633dkNIw4YNcRQ8fIY4mGjnzp2N PBjKCMyaNQsht99++7Rp0/CjVatWzoN62I5j7cMGNx7hfLgIhAYEBRc8clZG VFV6e9D+/fspSfTmm40bN3KUjSqKzZs3b9myxWHQDQfmAH2+6+XLl/vTHUoc JSUluZRFYPv27VQremc/JdzA5MmT16xZ8/XXX7MvHNixY4eewq5du1R7FoFK sp2j+8SJE4wgLoEguEGUUbgiykgQAtK6dWu4dg6jdW688UZE4Afb4JXR22+/ baRfp04dhHPFFlFGggHNBh6vClm7di1C1OAgAqGEQGMid/jeMGw42FFRUTRj NdLEwUSjo6ONPFiVEXjhhRfYwFSzZs2srCyHSygsLKRbbl2zFY+VrgIOHjyo CxYHZYRCYNuK/p0BIoKBqzWSk5MROGbMGIccGuCJwyHTp0/XA1etWuXQIQ0l zxFDLmURqma2o+mTdXt9KxAxEUgwtR4uIlMG6l1nqfiYz7S0tJSUlIkTJ7K4 oOyM08EZYGRjkhlBEGwRZRSuiDIShIBwFcuPPvrIdu+xY8ewt3r16vRSgldG MTExxinuv/9+hLNHjSgjwYBmA+GgQmyVEVfjUsooNjaWjaEGbpSRfi5iq4zw aFSrVg3hvXv3dr4EpYysbjkeK3j1w4cPV60ecO+VzvKnjLKzsyMjI4cNG2Y8 AosWLfL4RwmNgHDaN2Meg9mzZyOQU1LYsmTJEp4IKibgLHBbt261Nhjx0piI MYkfe9mhKHgVKFLO9a0Lq3Pnzs2fP99jN9MFjvInTgVBsCLKKFwRZSQIAWHf JMPPVCxevBh7H3jgAW5SubRs2VKPk56e7l4ZffDBB8YpbrjhBoQfPHjQffqi jCoPJVBGmzZtqlKlSs2aNbt16zZz5syVK1du27aNNhNCZcQ57SN8E9wFXOqI bry/aPDzd+7cuXTpUo6UARzWZ6uMTp48yWhG1zLvhXkGpk+fvt+C83wLBraj oji5gbU5hkDfYS+KiEOHkpOTHcQR6uXx48cj2tdff23sysvLYwnonSG9mrpU jcv4PXLkSOMsKGFG02f8BhBExZWHglCZEWUUrogyEoSAZGRk0MHjCHcduGd3 3HEHdqlptI8cOYLNK6644vTp0yoanBxbZdS9e3c9Nbqdjz32mB4Inw2B9erV K1b69HIppoTwpgTKCIIoQpvWw+uTGJyhzlBGtibqRhllZmbWqFEDdtujRw/s 6tSpk/NVcPoCY2I3r8/90Csp+O2cFA5qzmunjM6dO8cIkFHWs+Dq2JtO1wv4 DVlhNM04oDSI3jWR43SQE9vhVEoWQbWpMUcO4oh9CCG1rDoFpUEJtnXrVj2c U+qpKRSojIYNG2bkh01LkZGRxrqu33//PeO7LARBqOSIMgpXRBkJghvefffd CN8EyPTHyNGjR+F8cnoENW4dcCT71KlTuQl5cssttxjKJTExkZMt6GdRE3+p IQ94Qh999FGEdOzYsVjpc3YyOLchKwKhvFICZcRDnnvuObW3T58+Rm86BxMN qIzgjT/wwAMImTZtWkFBAR4Q/J45c6bDVXCmBcPbhy7gtNtqmjiljNhpzaqM Fi5c6PFNVW1bteHyOdXD8uXLGQJtwiV+kFWGnD59GiG4fKNVRUHRYQy2ysjI QCDyZo2vZBHbubwXT8hgFUcovXHjxtk2GBGOloqKilKNwnl5eVzkiEtKMRFe KRSiOgXU0PTp0xGo3h4Kridrm39BEKyIMgpXRBkJghvgWnCZoQjfIkH/9V// 9dBDD1WpUiXCt3CM0WmnV69ejNmiRQuolTp16jRo0MBQLvA/efhtt93Ws2dP BtLtrFevHv6/8847//CHP1x33XX4fdNNN+nK6/9r735a6dnjOIDfhcTWA/A8 2FpYKCsbW0/BwoZioShK2V87FCsWlKR+2SiibGQlSWHDiRXpfvLtTvOb4xzj Xi53vF4LMebMOXNmHJ/3zPdPme2nmwKtra29vb2vXjynMv5BMtre3k6nXyTo wcHBOJHi+46OjnwyanKKvpmMxsbG4seenp70Y6SVtP1GLc0e/h7TYGVlpbA8 DY8QIlhF2Z8iw9TUVOqRVEhGWR+c+fn5P3+X3d+JkDUxMZFur8TyyBfpHsrp 6Wn+lYT9/f1XX2q8e/HbSE/5hWlgurW1tcLKW1tbhViU5IfyfvWtmJmZadSw LZZnYylEBoyYk/JjPEtqSpd/neMv49EtLy8vLS2lm02x8vHxcWGbqQ9UNrw5 0JxkVFWSEZQUlUxEkmxOzKSrq2tnZ6ewZtQtUW1ma3Z3d29sbMQ3EVLyq0Xx lq6lR8WYlmQDfw0NDaU5ktrb26O+LcyPWWb7Ueb19/enjczOzn7OW8K3MDw8 HEc5f7X/169fsaSvry+/WmdnZyzM2lAtLi6m/mt/vEz2GgV2ugGa79dff4rW P1cyOjoay+fm5h5ehpVuaWlpa2tLUw4lAwMDhVufBVdXV5OTk1G6F+6h1Gq1 zc3NbErWqOojImUpI42olrUBOzo6Gm8gP7DAwcFB7G/q2RSZKHYnH9nOzs7i 6SIxFTryZCJlxAPTFLeZeFWxMPa9sHIku/pYlKRwVD/Kd6xfv/2CiITxMlLE G3+Z+DXyUf2zxJ6mZoqZiFT107w+vPScipzVaJeBAsmoqiQjeJeor3Z3d1dX VyOM1PeJyIuK7vDwsHmlEUXg5eVl1mMof0E+Kth4eKEvwHu3Hw+/uLh4cxQs fqxIBCcnJ03+CxRO0TJia/Xddl5dWJA61xSmK83En97HnswRLs7Pz1/tFlTo 3PRvpE5MjX4bb0jzaZ6ai12IT6HT09Pmb+z19XUE3lizUfvA+CQZ/30+WaA5 yaiqJCP4Pho1VYIfYnFxcXp6+s25fvhANzc3s7OzCwsLLqFAeZJRVUlG8H1I Rvxw8f9obW0tCnUDzv83Li4u5ubmVldXDdYN7yIZVZVkBN/H+vr6yMhIoVcR /DR7e3uNRj/gYx0cHDSZmhZoRDKqKskIAADKk4yqSjICAIDyJKOqkowAAKA8 yaiqJCMAAChPMqoqyQgAAMqTjKpKMgIAgPIko6qSjAAAoDzJqKokIwAAKE8y qirJCAAAypOMqkoyAgCA8iSjqpKMAACgPMmoqiQjAAAoTzKqKskIAADKk4yq qlarSUYAAFBS1M93d3dfXcXz8e7v7+PIxtevPsUAAOC7y4rnr67i+XiPj48a 1AEAQBmpKV2U0F9dxfMpUvIVjgAAoIkomN0wqrbn5+d0lOOrZnUAAFAQRXIq mGu1WhTPX12/84ni+MYRv729TQMyAAAASRpyIUrlyEdi0Q/x9PQU+SgddwAA 4Pbl3kEUyVEqf3W1DgAAAAAAAAAAAAAAAPy//QWJ0F3O "], {{0, 89.}, {559., 0}}, { 0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{559., 89.}, PlotRange->{{0, 559.}, {0, 89.}}]], "Output", CellChangeTimes->{ 3.763743996646669*^9, 3.763744181111668*^9, 3.763744896624071*^9, 3.763809921121181*^9, 3.763815803474024*^9, 3.763914840357451*^9, 3.764005668547656*^9, 3.7640811002179003`*^9, 3.764588560945242*^9, 3.764595064595758*^9, 3.772298332334594*^9, 3.797848653061522*^9, { 3.7979224602396183`*^9, 3.797922468214635*^9}, 3.798178758044959*^9, 3.79818617904491*^9}, CellLabel->"Out[11]="] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell["NetModel parameters", "Subsection", CellChangeTimes->{{3.728644832191423*^9, 3.7286448555810347`*^9}, { 3.7286721153422413`*^9, 3.7286721189663763`*^9}, {3.763323292847169*^9, 3.763323298806449*^9}, {3.7633787684827127`*^9, 3.763378772139513*^9}}], Cell["\<\ This model consists of a family of individual nets, each identified by a \ specific parameter combination. Inspect the available parameters:\ \>", "Text", CellChangeTimes->{{3.725979178627933*^9, 3.7259791789799137`*^9}, { 3.7259792303321257`*^9, 3.725979245908435*^9}, {3.730020487832671*^9, 3.730020494904923*^9}, {3.763325603268447*^9, 3.763325642225912*^9}, { 3.763378776274935*^9, 3.7633787809633427`*^9}, 3.763386725184752*^9, 3.7686596973136578`*^9, 3.773407800080714*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetModel", "[", RowBox[{ "\"\\"", ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{3.76337960821341*^9, 3.763404216494698*^9, 3.763417037611783*^9, 3.797848641991898*^9}, CellLabel->"In[12]:="], Cell[BoxData[ TagBox[ TagBox[ DynamicModuleBox[{TypeSystem`NestedGrid`PackagePrivate`$state$$ = Association[ "InitialData" -> Association[ "Type" -> Association[ "Description" -> "The particular type of net", "Default" -> "BaseUncased", "Allowed values" -> { "BaseCased", "BaseUncased", "LargeCased", "LargeUncased", "MultilingualUncased", "MultilingualCased", "BaseChinese"}], "InputType" -> Association[ "Description" -> "The type of input for the net", "Default" -> "String", "Allowed values" -> {"String", "ListOfStrings"}]], "InitialShape" -> TypeSystem`PackageScope`HeaderShape[ Association[{All, TypeSystem`PackageScope`KeyDummy[0]} -> 1, {All, "Description"} -> 1, {All, "Default"} -> 1, {All, "Allowed values"} -> 1], TypeSystem`PackageScope`Limited[ TypeSystem`PackageScope`ColumnShape[ TypeSystem`PackageScope`KeyValueShape[ TypeSystem`PackageScope`AtomShape[86.4], TypeSystem`PackageScope`Limited[ TypeSystem`PackageScope`RowShape[ Association[ "Description" -> TypeSystem`PackageScope`AtomShape[278.40000000000003`], "Default" -> TypeSystem`PackageScope`AtomShape[105.60000000000002`], "Allowed values" -> TypeSystem`PackageScope`UnknownShape[False]]], DirectedInfinity[1], 3, {All}]]], 2, DirectedInfinity[1], {}]], "InitialType" -> TypeSystem`Assoc[ TypeSystem`Atom[String], TypeSystem`Struct[{"Description", "Default", "Allowed values"}, { TypeSystem`Atom[String], TypeSystem`Atom[String], TypeSystem`Vector[ TypeSystem`Atom[String], TypeSystem`AnyLength]}], 2], "Meta" -> Association["ID" -> 58214187374966], "RowTarget" -> 20, "ColumnTarget" -> 10, "Shape" -> TypeSystem`PackageScope`HeaderShape[ Association[{All, TypeSystem`PackageScope`KeyDummy[0]} -> 1, {All, "Description"} -> 1, {All, "Default"} -> 1, {All, "Allowed values"} -> 1], TypeSystem`PackageScope`Limited[ TypeSystem`PackageScope`ColumnShape[ TypeSystem`PackageScope`KeyValueShape[ TypeSystem`PackageScope`AtomShape[86.4], TypeSystem`PackageScope`Limited[ TypeSystem`PackageScope`RowShape[ Association[ "Description" -> TypeSystem`PackageScope`AtomShape[278.40000000000003`], "Default" -> TypeSystem`PackageScope`AtomShape[105.60000000000002`], "Allowed values" -> TypeSystem`PackageScope`UnknownShape[False]]], DirectedInfinity[1], 3, {All}]]], 2, DirectedInfinity[1], {}]], "Type" -> TypeSystem`Assoc[ TypeSystem`Atom[String], TypeSystem`Struct[{"Description", "Default", "Allowed values"}, { TypeSystem`Atom[String], TypeSystem`Atom[String], TypeSystem`Vector[ TypeSystem`Atom[String], TypeSystem`AnyLength]}], 2], "Path" -> {}, "BaseIndices" -> {}, "DisplayedRowCount" -> 2, "DisplayedColumnCount" -> 3, "DataRowCount" -> 2, "DataColumnCount" -> 3, "SortPaths" -> {}, "SortDirections" -> {}, "HiddenItemsMap" -> Null, "UpdateType" -> 1], TypeSystem`NestedGrid`PackagePrivate`$outputID$$, TypeSystem`NestedGrid`PackagePrivate`$path$$ = {}, TypeSystem`NestedGrid`PackagePrivate`$vPos$$ = 1, TypeSystem`NestedGrid`PackagePrivate`$hPos$$ = 1, TypeSystem`NestedGrid`PackagePrivate`$grid$$ = DynamicModule[{ TypeSystem`NestedGrid`PackagePrivate`renderedGrid = Deploy[ Style[ Grid[{{ Item[ Pane[ Annotation[ Mouseover[ Graphics[{}, ImageSize -> 6, BaselinePosition -> Scaled[-0.15]], Graphics[{ GrayLevel[0.6], Polygon[{{2^Rational[-1, 2], -2^Rational[-1, 2]}, { 2^Rational[-1, 2], 2^ Rational[-1, 2]}, {-2^Rational[-1, 2], 2^ Rational[-1, 2]}, {-2^Rational[-1, 2], -2^ Rational[-1, 2]}}]}, ImageSize -> 6, BaselinePosition -> Scaled[-0.15]]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ All, Keys}, "KeyDummy", True], "Mouse"], ImageSize -> {{1, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Pane[ Row[{ Annotation[ EventHandler[ MouseAppearance[ Mouseover["Description", Style[ "Description", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ All, "Description"}, 1]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ All, "Description"}, "ColumnHeader", True], "Mouse"], " ", ""}], ImageSize -> {{1, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Pane[ Row[{ Annotation[ EventHandler[ MouseAppearance[ Mouseover["Default", Style[ "Default", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ All, "Default"}, 1]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ All, "Default"}, "ColumnHeader", True], "Mouse"], " ", ""}], ImageSize -> {{1, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Pane[ Row[{ Annotation[ EventHandler[ MouseAppearance[ Mouseover["Allowed values", Style[ "Allowed values", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ All, "Allowed values"}, 1]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ All, "Allowed values"}, "ColumnHeader", True], "Mouse"], " ", ""}], ImageSize -> {{1, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}]}, { Item[ Pane[ Annotation[ EventHandler[ MouseAppearance[ Mouseover["Type", Style[ "Type", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ Key["Type"]}, 1]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["Type"]}, "RowHeader", False], "Mouse"], ImageSize -> {{86.4, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Pane[ Annotation["The particular type of net", TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["Type"], Key["Description"]}, "Item", False], "Mouse"], ImageSize -> {{278.40000000000003`, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], ItemSize -> {Full, Automatic}], Item[ Pane[ Annotation["BaseUncased", TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["Type"], Key["Default"]}, "Item", False], "Mouse"], ImageSize -> {{105.60000000000002`, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], ItemSize -> {Full, Automatic}], Item[ Pane[ Annotation[ EventHandler[ MouseAppearance[ Mouseover[ Style[{"BaseCased", "BaseUncased", "LargeCased", "LargeUncased", "MultilingualUncased", "MultilingualCased", "BaseChinese"}, ShowStringCharacters -> False], Style[ Style[{"BaseCased", "BaseUncased", "LargeCased", "LargeUncased", "MultilingualUncased", "MultilingualCased", "BaseChinese"}, ShowStringCharacters -> False], FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ Key["Type"], Key["Allowed values"]}, 4]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["Type"], Key["Allowed values"]}, "Item", False], "Mouse"], ImageMargins -> {{5, 3}, {4, 5}}]]}, { Item[ Pane[ Annotation[ EventHandler[ MouseAppearance[ Mouseover["InputType", Style[ "InputType", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ Key["InputType"]}, 1]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["InputType"]}, "RowHeader", False], "Mouse"], ImageSize -> {{86.4, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Pane[ Annotation["The type of input for the net", TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["InputType"], Key["Description"]}, "Item", False], "Mouse"], ImageSize -> {{278.40000000000003`, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], ItemSize -> {Full, Automatic}], Item[ Pane[ Annotation["String", TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["InputType"], Key["Default"]}, "Item", False], "Mouse"], ImageSize -> {{105.60000000000002`, Full}, Automatic}, ImageMargins -> {{5, 3}, {4, 5}}], ItemSize -> {Full, Automatic}], Item[ Pane[ Annotation[ EventHandler[ MouseAppearance[ Mouseover[ Style[{"String", "ListOfStrings"}, ShowStringCharacters -> False], Style[ Style[{"String", "ListOfStrings"}, ShowStringCharacters -> False], FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][{ Key["InputType"], Key["Allowed values"]}, 4]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][{ Key["InputType"], Key["Allowed values"]}, "Item", False], "Mouse"], ImageMargins -> {{5, 3}, {4, 5}}]]}}, BaseStyle -> { ContextMenu -> Dynamic[TypeSystem`NestedGrid`PackagePrivate`$contextMenuTrigger; Which[TypeSystem`NestedGrid`PackagePrivate`$lastOutputID =!= TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$], {}, TypeSystem`NestedGrid`PackagePrivate`$contextMenuTrigger === TypeSystem`NestedGrid`PackagePrivate`$lastContextMenuTrigger, TypeSystem`NestedGrid`PackagePrivate`$lastContextMenu, True, TypeSystem`NestedGrid`PackagePrivate`$lastContextMenuTrigger = TypeSystem`NestedGrid`PackagePrivate`$contextMenuTrigger; TypeSystem`NestedGrid`PackagePrivate`$lastContextMenu = Block[{TypeSystem`NestedGrid`PackagePrivate`$globalScrollPos = \ {TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$}}, With[{TypeSystem`NestedGrid`PackagePrivate`lastPath$ = TypeSystem`NestedGrid`PackagePrivate`$lastPath, TypeSystem`NestedGrid`PackagePrivate`lastPathType$ = TypeSystem`NestedGrid`PackagePrivate`$lastPathType, TypeSystem`NestedGrid`PackagePrivate`isLeafHeader$ = TypeSystem`NestedGrid`PackagePrivate`$\ lastPathIsLeafHeader, TypeSystem`NestedGrid`PackagePrivate`allHidden$ = TypeSystem`NestedGrid`PackagePrivate`allHiddenQ[ TypeSystem`NestedGrid`PackagePrivate`$lastPath, TypeSystem`NestedGrid`PackagePrivate`$state$$], TypeSystem`NestedGrid`PackagePrivate`anyHidden$ = TypeSystem`NestedGrid`PackagePrivate`anyHiddenQ[ TypeSystem`NestedGrid`PackagePrivate`$lastPath, TypeSystem`NestedGrid`PackagePrivate`$state$$], TypeSystem`NestedGrid`PackagePrivate`sortDirection$ = TypeSystem`NestedGrid`PackagePrivate`columnSortDirection[ TypeSystem`NestedGrid`PackagePrivate`$lastPath, TypeSystem`NestedGrid`PackagePrivate`$state$$[ "SortPaths"], TypeSystem`NestedGrid`PackagePrivate`$state$$[ "SortDirections"]], TypeSystem`NestedGrid`PackagePrivate`haveData$ = Not[ FailureQ[ TypeSystem`NestedGrid`PackagePrivate`datasetInitialData[ TypeSystem`NestedGrid`PackagePrivate`$state$$]]], TypeSystem`NestedGrid`PackagePrivate`isKeyDummy$ = Not[ FreeQ[ TypeSystem`NestedGrid`PackagePrivate`$lastPath, Keys]]}, Join[{ If[ Or[ Not[TypeSystem`NestedGrid`PackagePrivate`haveData$], Not[TypeSystem`NestedGrid`PackagePrivate`anyHidden$], TypeSystem`NestedGrid`PackagePrivate`isKeyDummy$], Nothing, MenuItem[ StringJoin["Show ", Which[ TypeSystem`NestedGrid`PackagePrivate`lastPathType$ == "Item", "", TypeSystem`NestedGrid`PackagePrivate`allHidden$, ToString[ ReplaceAll[ Last[TypeSystem`NestedGrid`PackagePrivate`lastPath$], Key[ Pattern[TypeSystem`NestedGrid`PackagePrivate`x, Blank[]]] :> TypeSystem`NestedGrid`PackagePrivate`x]], TypeSystem`NestedGrid`PackagePrivate`lastPathType$ == "RowHeader", "Row", TypeSystem`NestedGrid`PackagePrivate`lastPathType$ == "ColumnHeader", "Column", True, ""]], KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`updateHiddenItems[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][ TypeSystem`NestedGrid`PackagePrivate`lastPath$, "remove"]], MenuEvaluator -> Automatic]], If[ Or[ Not[TypeSystem`NestedGrid`PackagePrivate`haveData$], TypeSystem`NestedGrid`PackagePrivate`pathEmptyQ[ TypeSystem`NestedGrid`PackagePrivate`$lastPath, TypeSystem`NestedGrid`PackagePrivate`$state$$], TypeSystem`NestedGrid`PackagePrivate`allHidden$, TypeSystem`NestedGrid`PackagePrivate`isKeyDummy$], Nothing, MenuItem[ StringJoin["Hide ", Which[ TypeSystem`NestedGrid`PackagePrivate`lastPathType$ == "Item", "", TypeSystem`NestedGrid`PackagePrivate`allHidden$, ToString[ ReplaceAll[ Last[TypeSystem`NestedGrid`PackagePrivate`lastPath$], Key[ Pattern[TypeSystem`NestedGrid`PackagePrivate`x, Blank[]]] :> TypeSystem`NestedGrid`PackagePrivate`x]], TypeSystem`NestedGrid`PackagePrivate`lastPathType$ == "RowHeader", "Row", TypeSystem`NestedGrid`PackagePrivate`lastPathType$ == "ColumnHeader", "Column", True, ""]], KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`updateHiddenItems[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]][ TypeSystem`NestedGrid`PackagePrivate`lastPath$, "add"]], MenuEvaluator -> Automatic]], Delimiter}, If[ And[TypeSystem`NestedGrid`PackagePrivate`haveData$, MatchQ[TypeSystem`NestedGrid`PackagePrivate`lastPathType$, Alternatives["ColumnHeader", "KeyDummy"]], TypeSystem`NestedGrid`PackagePrivate`isLeafHeader$, Not[TypeSystem`NestedGrid`PackagePrivate`allHidden$]], { If[ TypeSystem`NestedGrid`PackagePrivate`sortDirection$ =!= "Ascending", MenuItem["Sort", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`updateSort[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][ TypeSystem`NestedGrid`PackagePrivate`lastPath$, "Ascending"]], MenuEvaluator -> Automatic], Nothing], If[ TypeSystem`NestedGrid`PackagePrivate`sortDirection$ =!= "Descending", MenuItem["Reverse Sort", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`updateSort[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][ TypeSystem`NestedGrid`PackagePrivate`lastPath$, "Descending"]], MenuEvaluator -> Automatic], Nothing], If[ TypeSystem`NestedGrid`PackagePrivate`sortDirection$ =!= None, MenuItem["Unsort", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`updateSort[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][ TypeSystem`NestedGrid`PackagePrivate`lastPath$, None]], MenuEvaluator -> Automatic], Nothing], Delimiter}, {}], { MenuItem["Copy Position to Clipboard", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentPosition[ TypeSystem`NestedGrid`PackagePrivate`copyClip]], MenuEvaluator -> Automatic], MenuItem["Copy Data to Clipboard", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentData[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`copyClip]], MenuEvaluator -> Automatic], Delimiter, MenuItem["Paste Position in New Cell", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentPosition[ TypeSystem`NestedGrid`PackagePrivate`cellPaste]], MenuEvaluator -> Automatic], MenuItem["Paste Data in New Cell", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentData[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`cellPaste]], MenuEvaluator -> Automatic]}]]]]], FontFamily -> "Verdana", FontSize -> 12}, Spacings -> {0, 0}, Alignment -> Left, Dividers -> All, FrameStyle -> GrayLevel[0.7490196078431373], BaseStyle -> {FontFamily -> "Verdana", FontSize -> 12}], LineBreakWithin -> False, ContextMenu -> {}, NumberMarks -> False, ShowAutoStyles -> False]], TypeSystem`NestedGrid`PackagePrivate`initialQ = True}, Dynamic[ TypeSystem`NestedGrid`PackagePrivate`setupViewPath[ TypeSystem`NestedGrid`PackagePrivate`$path$$, If[ Not[TypeSystem`NestedGrid`PackagePrivate`initialQ], Module[{ TypeSystem`NestedGrid`PackagePrivate`tmpGrid$ = $Failed, TypeSystem`NestedGrid`PackagePrivate`tmpData$ = TypeSystem`NestedGrid`PackagePrivate`datasetData[ TypeSystem`NestedGrid`PackagePrivate`$state$$]}, TypeSystem`NestedGrid`PackagePrivate`tmpGrid$ = If[ FailureQ[TypeSystem`NestedGrid`PackagePrivate`tmpData$], TypeSystem`NestedGrid`PackagePrivate`renderedGrid, TypeSystem`NestedGrid`PackagePrivate`renderGrid[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][ TypeSystem`NestedGrid`PackagePrivate`tmpData$]]; If[ Not[ FailureQ[TypeSystem`NestedGrid`PackagePrivate`tmpGrid$]], TypeSystem`NestedGrid`PackagePrivate`renderedGrid = TypeSystem`NestedGrid`PackagePrivate`tmpGrid$]; Null]]; TypeSystem`NestedGrid`PackagePrivate`initialQ = False; TypeSystem`NestedGrid`PackagePrivate`$vPos$$; TypeSystem`NestedGrid`PackagePrivate`$hPos$$; If[ FailureQ[TypeSystem`NestedGrid`PackagePrivate`renderedGrid], TypeSystem`SparseGrid[ TypeSystem`H["(data no longer present)"]], If[GeneralUtilities`$DebugMode, Row[{TypeSystem`NestedGrid`PackagePrivate`renderedGrid, " ", TypeSystem`NestedGrid`PackagePrivate`formatState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$]}], TypeSystem`NestedGrid`PackagePrivate`renderedGrid]]], TrackedSymbols :> { TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$}], DynamicModuleValues :> {}], TypeSystem`NestedGrid`PackagePrivate`$topBar$$ = Dynamic[ TypeSystem`NestedGrid`PackagePrivate`makeFramedBar[ TypeSystem`PackageScope`SubViewPathbar[ TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$vPos$$, TypeSystem`NestedGrid`PackagePrivate`$hPos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]]]], TrackedSymbols :> {TypeSystem`NestedGrid`PackagePrivate`$path$$}], TypeSystem`NestedGrid`PackagePrivate`$bottomBar$$ = Style[ Framed[ Dynamic[ Replace[ TypeSystem`NestedGrid`PackagePrivate`mouseAnnotation$$, { TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][Null, Blank[]] -> "", TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]][ Pattern[TypeSystem`NestedGrid`PackagePrivate`path$, Blank[]], Pattern[TypeSystem`NestedGrid`PackagePrivate`pathType$, Blank[]], Pattern[TypeSystem`NestedGrid`PackagePrivate`isLeafHeader$, Blank[]]] :> ( Increment[TypeSystem`NestedGrid`PackagePrivate`$contextMenuTrigger]; TypeSystem`NestedGrid`PackagePrivate`$lastPath = TypeSystem`NestedGrid`PackagePrivate`path$; TypeSystem`NestedGrid`PackagePrivate`$lastPathType = TypeSystem`NestedGrid`PackagePrivate`pathType$; TypeSystem`NestedGrid`PackagePrivate`$lastPathIsLeafHeader = TypeSystem`NestedGrid`PackagePrivate`isLeafHeader$; TypeSystem`NestedGrid`PackagePrivate`$lastOutputID = TypeSystem`NestedGrid`PackagePrivate`localHold[ TypeSystem`NestedGrid`PackagePrivate`$outputID$$]; TypeSystem`NestedGrid`PackagePrivate`makePathTrail[ TypeSystem`NestedGrid`PackagePrivate`path$, TypeSystem`NestedGrid`PackagePrivate`makePathElements]), Null :> Spacer[10], Blank[] :> Spacer[10]}], TrackedSymbols :> { TypeSystem`NestedGrid`PackagePrivate`mouseAnnotation$$}], FrameStyle -> None, ImageMargins -> 0, FrameMargins -> 0, Alignment -> Top, ImageSize -> {Automatic, 15}], FontSize -> 1], TypeSystem`NestedGrid`PackagePrivate`mouseAnnotation$$ = Null}, DynamicWrapperBox[ DynamicBox[ToBoxes[ Dataset`DatasetContent[ 1, "Path" -> TypeSystem`NestedGrid`PackagePrivate`$path$$, "Grid" -> TypeSystem`NestedGrid`PackagePrivate`releaseLocalHold[ TypeSystem`NestedGrid`PackagePrivate`$grid$$], "State" -> TypeSystem`NestedGrid`PackagePrivate`$state$$, "VPos" -> Hold[TypeSystem`NestedGrid`PackagePrivate`$vPos$$], "HPos" -> Hold[TypeSystem`NestedGrid`PackagePrivate`$hPos$$], "TopBar" -> TypeSystem`NestedGrid`PackagePrivate`$topBar$$, "BottomBar" -> TypeSystem`NestedGrid`PackagePrivate`releaseLocalHold[ TypeSystem`NestedGrid`PackagePrivate`$bottomBar$$]], StandardForm], ImageSizeCache->{20., {14.634033203125, 20.365966796875}}, TrackedSymbols:>{ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$}], TypeSystem`NestedGrid`PackagePrivate`mouseAnnotation$$ = MouseAnnotation[], ImageSizeCache->{20., {14.634033203125, 20.365966796875}}], BaseStyle->{LineBreakWithin -> False}, DynamicModuleValues:>{}, Initialization:> Block[{$ContextPath = $ContextPath}, Needs["TypeSystem`"]; Needs["Dataset`"]; TypeSystem`NestedGrid`PackagePrivate`$outputID$$ = CreateUUID[]], UnsavedVariables:>{TypeSystem`NestedGrid`PackagePrivate`$outputID$$}], Deploy, DefaultBaseStyle->"Deploy"], Dataset`InterpretDataset[1], Editable->False, SelectWithContents->True, Selectable->False]], "Output", CellChangeTimes->{3.76337884064023*^9, 3.7633796089564543`*^9, 3.763404156781633*^9, 3.763417108845847*^9, 3.763743998299114*^9, 3.763744193893393*^9, 3.7637448979920483`*^9, 3.76380992348105*^9, 3.763815805184484*^9, 3.763914842025527*^9, 3.7640056802398577`*^9, 3.764081101869526*^9, 3.797848657547656*^9, 3.7979224708965*^9, 3.798178760203719*^9, 3.7981861822269497`*^9}, CellLabel->"Out[12]="] }, Open ]], Cell["Pick a non-default net by specifying the parameters:", "Text", CellChangeTimes->{{3.763378864569222*^9, 3.763378874753018*^9}, 3.763386727447794*^9, {3.76458890157792*^9, 3.764588917066121*^9}, 3.768659717518422*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetModel", "[", RowBox[{"{", RowBox[{"\"\\"", ",", RowBox[{"\"\\"", "\[Rule]", " ", "\"\\""}], ",", " ", RowBox[{"\"\\"", "\[Rule]", " ", "\"\\""}]}], "}"}], "]"}]], "Input", CellChangeTimes->{{3.763386371899088*^9, 3.763386375258068*^9}, 3.763404216501815*^9, 3.763417037618573*^9, {3.763417129146069*^9, 3.763417141784073*^9}, {3.763417986203992*^9, 3.763418001642385*^9}, { 3.763744070673006*^9, 3.76374407921362*^9}, {3.763914881125794*^9, 3.763914886338791*^9}, {3.763916144220999*^9, 3.763916145566635*^9}, 3.797848641997761*^9}, CellLabel->"In[13]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsnQl4FVWa/oOgLKKgyKIsIiK0dKOCSqu04mij0jRDNwqjYNOC3e5KC9gg iNywh30PGJZAIKxh3wy7QQlrEgmELYAQCCGQdBju9Oio/X/nvsP5H2q7lXtv ILn5fk/gufXVqVOnTn1V9b11lrqve48Ob90UERHRqwL+69Ctz7/17Nnt05er YqHjR73effujv/6lzUe9//r2X3s+0b0sjKvwr3u5iIj//f0vQRAEQRAEQRAE QRAEQRAEQRAEQRDCgh9++CEvLy87O/vs2bNZgiAIgiAIgiBoIEg+f/48AmaE zTc6cheuBz///DNO97lz53Jzc//xj3/8pyAIgiAIgiAIJhAqI2BG2Hzp0iWE 0Dc6iheKEJzfCxcuXLx4Eef9ypUrXkEQBEEQBEEQbEDAjLAZwTNCaBFKYQyE cF5eHs71jfY4QRAEQRAEQSgZIHjO83GjY3mhSPjhhx9yc3NFIgmCIAiCIAhC oUAIjUBaxiiFJfn5+QUFBdLRThAEQRAEQRAKBUJoBNIIp290RC+EHg5HutEu JgiCIAiCIAglDw5QutERvRB6ZESSIAiCIAiCIAQGRyfd6IheCD2c9/tG+5cg CIIgCIIglDw4N/iNjuiF0CMqSRAEQRAEQRACQ1RSuCIqSRAEQRAEQRACQ1RS uCIqSRAEQRAEQRACQ1RSuCIqSRAEQRAEQRACQ1RSuCIqSRAEQRAEQRACQ1RS uCIqSRAEQRAEQRACQ1RSuCIqSRAEQRAEQRACQ1RSuCIqSRAEQRAEQRACQ1RS uCIqSRAEQRAEQRACQ1RSiebSpUu5Gj/99JNaJSpJEARBEARBEALDoJJg0aNu LN6I2F9wxffff//BtZw4cUKtLREq6cB3O1fsmVqov+9yjtjl9m3W+Rm7Dhbq D5vY5Xbq1IXly1IK9XfgwOmiqSdBEARBEAThumJQSXFxcXrUHRsbeyPCf8EV //3f/41ztG7dujNX+fHHH9XaEqGSoHreiW1eqL+9xzfZ5Zbw7fFnYtYW6g+b 2OW2Z8+Jt/66sFB/EEpFU0+CIAjhT25u7qVLl250KQRBEP4Pg0oqKChQIfew YcNmz559A6J/wR1USXv27LFcW4JU0oBV7Udt7u737+9LX3CjktosSHp5VZrf PyRzo5J6frBmcJ8dfv/6/u1LUUmCIAgBc/jw4dtuu61atWpnz5690WURBEH4 XxzGJY0fP15UUnEmbFQSFNCc1IF+/zxrOrhRSX9Ytq/btjN+/5DMjUrq9/HG KYMO+f2DUBKVJAiCEDDbtm2L8PHtt9/e6LIIgiD8L6KSSi6ikgyIShIEQSi5 LFu2bN26dTe6FIIgCP+HqKSSi6gkA6KSBEEQBEEQhJAgKqnkIirJgKgkQRDC m9jY2C5dutSpU6d58+a9evXC/V+tevfdd5977rmJEycqS05Ozu9+9zsY9+/f j8VJkyY99dRTW7duXb58+TPPPFOzZs3f/OY3gwcP/u6779QmeHC0bNmye/fu SUlJnTp1uueeezp06KDWpqen9+/fH5nUqlUL9smTJxueMocPH37jjTeaNm2K DTt37jxr1qyLFy+6T/D000+3bdtWT489fvbZZyxtmzZthg8ffubMGXNpDxw4 0K1bt4YNG9auXRvJkpOTA65hQRAEhaikkouoJAOikgRBCGOgUDhyB5KhQoUK +FGtWjUIBK7dtGlTmTJlYFfjet5//32k+eMf/8jFnj17chH/I+Vdd93F3J54 4on8/HymuXTpErOtUqUKfpQvXx6ig6uOHTsGaQPjLbfcApXEbfEMUsXbtWvX nXfeCWO9evWgg7AL/H7++efdJ8AiSqUWjx49yj2CGjVq8Mfjjz/Oz5So0t57 7733338/t2UaFPvgwYMhrXtBEEojopJKLqKSDIhKEgQhXImPj6cQ2LFjBxbz 8vLee+89WBo3bqzS9OrVC5ZWrVrhd1JS0k033QQ9dfr0/33HjSoJvPbaa2w/ gmypXbu2LnaoO5gtdqQeIleuXHnooYdgf+uttzhZd3JycvXq1WGZM2cO07zz zjtY/Pjjj5HY69M4lDbbt293mUBXSUjTpEkTWDp06HDy5Emvr12pWbNmsHTs 2NFQ2kceeSQzMxMWlBm/YXn77bdDfw4EQShliEoquYhKMiAqSRCEcOWVV15B 8L9gwQJlgY5o1KgRjIcOHaIlPz+/adOmsEyaNOnhhx/Gj4SEBJWeKunBBx+E wlLGLVu2sDmGi0p3pKam6ntPSUmBEZlT4JDY2FgYu3TpwsV27dphcerUqSrB 2bNnIYXU7vwm0FUS91i/fn3VcgQyMjJuueWWihUrXr58WS/tvn37VJrRo0fD 8uyzz7qsWEEQBDtKp0qKjo7u3bv35MmTr8O+fvrpp1mzZmF306ZNg64JYc6i kgyIShIEIVxp0KABgn88l2M0mjdvDmN8fLxKtnv3bugIaodu3brpOVAlffjh h4ac2bnu3Llz3qu6o0aNGoY0M2fOhP3pp5/W9z548GAYf/nLXzLN3LlzsVi2 bNk//OEPEFDHjxvvrn4T6CqJe3z99dcNaaj+cJiqtCi/rt2gmNi65KdCBUEQ /BGwSvrxxx83bdq0fv36vXv3mtempaVhVUpKSkDhf1Dk5ubu3Llz8eLF48aN gzBJTEzMzMw0pImKioK+GDt27HUoT3p6+gdXQcFCmHNRqKTVIcLl7kQlCYIg +EU1mlji8Xj0xB999BGM5cqVM3yelSpp0qRJhswptbZt26Z2VLduXUMa9uWz 5Oabb2YaPHQiIyOrVq2qVrVp04b9A10m0FUS9zhw4EBDSTp06AA7RJZdaY8e PSoqSRCEkBCwSvqv//ovRv49evTAncqwdujQoVg1a9asQBXA/yc1NTU+Pn7/ /v1uEuM+/7e//e0DE3Pnzv3+++9Vsuupko4fP66KYadoAkNUkgFRSYIghCsc BIS76xYTeqNMdnY2VAM1yOeff67nQJXUt29fQ85333037MzETiUhHoD95Zdf Nu+d8kpx/vz5JUuWQKlxhoeKFSuylcpNAl0lTZgwIcI3DMpQkieffBL2rVu3 2pVWVJIgCKEieJUE5s+fb1gbQpW0aNEiZBUdHe2c7Keffpo+fboq0sCBA2fO nDlp0qTevXvTMm/ePJX4eqoksGnTJpRt7dq1P/74YwizFZVkQFSSIAjhygsv vIDgf82aNc7JunbtimTvvPPO7bffXq5cuZ07d6pVVEnPPPOMnp6aAhKMi3Yq afv27bBDobgvcH5+/rPPPstegi4T6CopKSkJi82bN9c3yc3NxXGVLVuW84eL ShIEoUgJiUr66KOPLly4oK8NoUqC2HGjknbs2MHCfPLJJ+np6Xo5Y2Ji+vfv n5eXp4zXWSUVEaKSDIhKEgQhXBk0aFCEb+q5rKwsWg4fPtygQYNKlSrt2rWL liVLlkT45mfA/Z9tMU2bNlWzfKs57kaPHk0Lkj311FOwtG/fnhY7lQQ7hy8h W2Xs168f9v7KK694fbHE/fffX6NGDaXLrly50rZtW2wyY8YMNwm816okFJt9 8wYPHqz2+B//8R+wtGjRwqG0opIEQQgVIVFJYM6cOfpaZ5X0448/njlzZvfu 3YcOHbLb+798zUPff//9uHHjkNXkyZPNhVcp//nPf3766adI9re//S07O9uc lWHaBINKQnmOHTsGufHdd9/pHfMsS440KSkpKPyJEyf0MujY1fbPP/9sTvzD Dz9gFf4vbGGKTiWt0XBWQ2tMiEoSBEEIOQUFBWxOqlatWocOHdq0aUMRob76 ikcGJ9betOl/b5LQIFATWPz73//OBFRJ7Ln3i1/84ne/+x2ywu/atWtzqm2v vUoC69atu/nmm7H20Ucfff311zmLgt5c9eGHH8IC3dStWzfsi3vH7vBQdpkg 4trvJSUmJpYvXx7GRo0atWvXjt9OqlWrVkZGhkNpRSUJghAqgldJkEL4H3c/ XZ7YqSRogZUrVxqGDg0bNuz06dN6sh07dlgOL9Lp06ePSr9lyxYaFy9e7CAr FFRJkyZNwuHPnDkTt2uVLdQWtJt5k3Pnzs2fP/+TTz7Ry4CjxuEogUMgW+zK 7LVSVaNHj8YqPNfwUHNZGCIqyYCoJEEQwpicnJzu3bvXq1ePTUIQFAMGDIB6 4tpOnTrBiAQq/e7du6Fibrrppm+++cZ7VSVNmDABacqWLcsxQVBbXEvy8vIi fF99tSzAokWLnnnmGc6hh5yffPJJ1YwFLl++PHDgQPVpV6ghCDF9zJTfBGXK lDFMr7ds2bLnnnsO5UT6O+64A1pJ70NoWVpkCGOzZs0KU7WCIAgWBK+SEKXj vocfiPDVWkuVhJt5ZGQkt+rRo8fgwYP79evHxY8++kjvJrdt2zZniQT69u2r 0s+bN49GvVudA1RJeL6w5CyPyvnjjz9GUfX0SUlJEEQqQa9evXQtYxiWhXt4 oVQSCxMdHe2yMIqiVkluetat8mHeyuXuAlZJbMgzELBKssxNVJIgCMWTjIwM 1fqjwD0fT0DDbT8/P18ZqZKmTZvm9emLlJQU/cNJ7rl48WJqaip2Z5cAZTtw 4IDDA8hvAgM4ioMHDxb/b1sIghBmBK+SIAq++eYb/j5z5gzXWqok3JyZbOHC hd6reuHIkSOfffbZB775FlQHM8T/+Vf5/PPPsXbixIn516Jrh7Fjx35wbeuS MxQmJCYmJicnB0Y8L7AXGteuXaunP3XqFIz9+/eHfEPin33gOUWVBwF17tw5 lRhHYShqQkKCX5XkvjCK0qyS4Jn80rpOwCppwYIFeAQbchOVJAhCOKGrJEEQ BMEvwauk7du3//jjj2wkwu2Xa80qKSUlxbLlBUBucJXlp5c8Hs8H/mZvgD5C mpEjRzqk0VHCZMOGDbod2uSjjz6CfcaMGYZNjh49auhZB/bv3898vvrqK4fd Qea4UUnuC0NK8+wNqJPJkyer3uwkYJUUHx8Pb4cW1nMTlSQIQjghKkkQBKFQ BK+Stm3bhsXdu3dzEaHmv6xU0uLFi9nscuXKFXNubJRZs2aNeZUblcThQiiw QxodCpNRo0aZV1HuWa5S/Pzzz5cuXTp06NDWrVt51AkJCQ7p3aikAApTpCoJ 52Ljxo2bXIPEevOTy90Fo5JGjx6N//UeI8GoJOQGH8vJyVG5iUoSBCGcwJ25 T58++igkQRAEwYFQqaSffvppyJAhH/gmo1ORv66SxowZw/RjreAqyznx3Kik wYMHs8+eQxodh5nAOaXe8OHDzasQQkPv4EA+/vjjD67F3ECm40YlFbYw/ypi lZSYmHj+/Pl81yAxNrnOKgksWLDg8uXLzC1IlQTg8KqzvagkQRAEQRCEUkuo VNK/tO5nx48fZ3ivqx6KHWf0+R8MGzqrJH5P9qOPPnL52VYHYYKjNguTn3/+ ec2aNaqcf/vb30aNGoWjW7lyZZGqJMvCKIpUJW3atAna579dg8TY5PqrJKB2 F7xKAosXL2a9iUoSBEEQBEEotYRQJUFKIJ5nc9LUqVMt25J69+79nT2XLl0y 74gqCRnalQSsWLGChbEc2WSmsMJk2bJlzH/EiBH79u375z//qVaxXUlU0nVT SQs3TEtKSoKPjdaAtIFx3OrNhVVJ2AQbQmXrueFswrhwwUZRSYIgCIIgCKWT EKokcODAARo5qbWukhYtWsRVnMPNPVRezjMzHD16lJkPGTLE8sutBgorTPr3 7w8j4mfzBA7hqpKKbY+7DckLjxw5Eh0drUTNuHHjdu3aBeOMbXsKq5KwCTaE n6vcIOchkWBcv363qCRBEARBEITSSWhVEhg1apTqmWY5xx0CUZf94gjnD+/Z s6fXSmIYklGwGPI/ffo01JP+PaZCCRMcKT+WtGzZMkPi3bt3c1X4qaQSMXsD wZllbiHpcYf6ZG7S404QBEEQBKHUEnKVpKb1/sA0G0NMTIzqt3b48OH/+Z// YaiflpaG6BRrLXeUmJjIraCDVFFzc3O//vrrn376SSXLycnhvNlUHLt27crO zs7MzFy3bh2be/C/96pIKZQw+fnnnzmHXt++fY8fPw7LDz/8gJyjo6PVkYaf SgoSl7sLXiXp8zUFr5Ig91RuopIEQRAEQRBKLSFXSf+6OjObWSVduXIF+kgp C4gafueIfPjhh/qHYvUd/f3vf/9AQ00xd+LECT3lkSNHOKO4mZ49e+7bt0+l DHhckgHs7tNPPw1XleT+27I3SiWtX79ezy1IlZSQkAAXVbmJShIEQRAEQSi1 BKySEKKzs5lZJR07dsyuheWnn37avn07h/noEmbOnDnZ2dl2+7pw4QKCWINC gXY4evSoISUOJzY2tm/fvipZr169Jk2adPbsWT0Zc+Ok5eajNgsTFHvFihX6 BOCfffYZgmoouLi4OCwuXbrUrvD/8qeSClsYRSlXSYsWLTIcYDAqCR4Ika7n JipJEARBEASh1BKwSgqSK1euQONkZGRcunTJzXwLAEEs0p86der06dOWn6bV ycvLS09Pz8rK0nvlBck///nPkydPIltkHqo8g+HGqqRVPm6USlq+fHl+fr4h t4BVEgqcm5tryE1UkiAIggE8f0OVFZ5QhndTDmRnZ+tf/Q57UDPqU4Ahx/z0 NBDCs6yD5yxCvqLIOTAK5YFCySJUV9CNUklC8JRmlaS+/aoTsEqyzE1UUmkg MzMzPj7+q6++utEFEYQSwPvvvx8RETFgwIDgs4LqiYmJ2bZtm5vEJ06c8Hg8 gwYNOnfuXPC7Lj6cPXv2f786sXBhXFzc5s2bT548qVYlJydHR0dnZWWFcHcQ R0OHDm3evDlO4ttvv22XLFRneffu3bfffvsf/vAHLh4+fPi2226rVq0ajjrI nENCoTywhHLs2DE4mJqTqrDk5OR88803y5YtgxZANLh//37LqBJ+u3LlytjY 2Pnz52/ZssVBgx88eDDJCpRTT4a9pKWlrVu3DvtdsWIFym8Ws37ThOoKEpVU cik6lbTGhJ0aulEqyZKAVZIlopJKAwsWLEA8oJ7j15mUlJTPP/8cz5cbsndB 8BbSCVu3bo3r5dVXXw1ypxA7UVFRuPpcPqQOHTrk8QG5FOSuiw8ZGRnQLB4N yMC9e/dy7ZUrVxISEoYNGxbCZh3IMZy+cuXK/fa3v/3iiy/skoXqLCPARj7Y FxehRyJ8fPvtt0HmHDyF9cASB44rMTExMjISfjVu3LgAcsC1NmrUKM+1zJs3 zyBGUlNTDWmgHeyE0ty5cz1WIMhUaZA/1JYhwYwZM/RWSDdpQnUFiUoquYhK MiAqSSgsN1Yl4YmDvb/11ls3ZO+C4C2kE2ZmZk6fPj3IqAPPpqk+3D+hEPCk paUdPHgwmP0WKxClDx48GKFdfHw8jis9PV0FkOrtN44agd+ECRPy8vJCstNX XnkF5xpxnXOykJxlr0klgWXLlq1bty7IbIMnAA8sWcC7YmJilHwIQCVBhgwf Phzbzpw588CBA8eOHdu0aRM114YNG1Sy3NzckSNHwhgbG7t//34I4TFjxmBx xYoVltmizrF2/fr1hk/JHD58WKVZvnw50mDv8B/sd8eOHXyZsHjx4kKl8Ybo ChKVVHIpzTOBWyIqSSgsopKEUs71d0JEMohnMjIy7BKEfDxOCDO0yyo1NXXl ypX6LKk6Gzdu/Prrr3ULojuGoGoT5Dxp0iQY9Q7AECywxMXF2eVcKB588EGc 6+PHbZ96ocWskooJfj3Qa9UPH6fATUzl19nceGOQHhsVFcU2HfhSYCqJLUTQ F/ohw41hjI6OVhbW5OTJk1Wyo0ePUpodOXLEnC0bp5yrkWn0s7N3715YcFDq KnCThgR/BYlKKrmISjIgKkkoLAaVhKumZcuW3bt3P3DgQLdu3Ro2bFi7du02 bdokJyerTRDJPPXUU1u3bl2+fPkzzzxTs2bN3/zmN4MHD/7uu+9UmhMnTiDN X//6V31f3377LYz8cBt39MADD2DvyAF27OW6HLFQUlGON3z48CeeeKJGjRqt W7emZ8bExDz22GN33XUX7Fhr2DAhIeHdd99t3LhxvXr12rVrN3/+fNodnJD7 2rx588CBA1u0aFGlSpUdO3bA/sUXX8C+aNEi/Iblueeee+mll/RhJtOmTYOx d+/edkfBEUZTp041r0L0HhsbO2LEiEGDBkE+bNmy5cKFC1xVUFCAgGTKlClc PHTo0Hgb9JfGx44dw7GPHTs2MjISJf/yyy8NoS8Kc+rUKedqT0xMRLYI+CGC EG0iKxxjSorxVj9r1iwc15IlS8zB2IYNGxiy6sb169ePGTPGIJ2WLVuGlPhf N6IqYMQhO5cToPb+9Kc/1a9fv0mTJu+8847eiwlRJU5c+fLlca7hKvjt0ONO P8tezfEQM+Pkws0eeeSRzz//3FCZWEQCuFD16tWbNWs2evTopKQkg0p6+umn 27Ztq2+F40JWuJHCn5999tlPP/304sWL5uPq0qVLnTp1mjdv3qtXL/MoG5QQ 10KtWrV+/etf9+/ff9euXQ61ZOeBkAA4RzizqH8UHmnYdgnBgtOEq2bYsGG4 z+PsoyrMG+L2vm7duokTJ2LDkSNHwvEM81TguJAAPowEOB0Ik06ePIkNFy5c qNJgX3B7+BJ2hDQ4BZbn3a/fYnPol/z8fGqWAFRSeno6Srh//37diMpBbrig lIVqBY9LPRn7wqFaDHkiImXlOOw3NzcX1wWqUb+OcnJyqLw4f4ubNDruryBL RCWVXIpCJV1nRCUJNxaDSsJzDYv33nvv/fffjx+IB9iRHtGF6u3Ts2dPWP74 xz/i/zJlyqg0CFBVZ2wkhuXxxx/X94WoEsYXXnjB6xtGjZD1jjvugKVSpUp1 69Z9+OGHr9tRCyUROl6rVq3wPz0HVK1atUePHvhx6623VqhQgUZE8morKA4a EYUiDX8zznFwQu4LsS7TI2dEvLAjBMWiCro4gOWNN97gIkKyypUrlytXTv/Y twGEXuxyY7AjhmEPNMQ/bFLhO2q+VEdRsYgETIzra+y1DBkyhJuo0BQhKDvh oAagvLh23rx5KrJCaEejYeS4genTp6ueS+yDxDwNr8qhzhA/Y9XSpUv14A3S jBLp/PnzDnshM2bMQGLDfAIIm5mt87b9+vXjybr99tvpCbg7KUWMtTi/N910 E+xQE/gNIWCXleEs0xmef/55/I/zq9yse/fu+lZYpB1udsstt+AHtJVBJfGm qhahi6F9uFW1atX4A16nv3FiYSjkuWuk1MNy7hersCEzufnmm80OprDzwBUr VuinGM7GbmDs3MV2CuVma9euNWwIJ8T/cDn4BtPMmTNHpUEwxlhdT8POabpc hcpWnsbLAf/jstLL6cZv1dMqYJVkBl6NxyVyU7KOVyUw6GU2YBk6v3mvChno RMicvXv34tLYuXOnm/lYMjIysCFUqkN7kEMal1eQHaKSSi6ikgyIShIKi6VK Ao888khmZqbXJ23wW58SijEDeO211/g037VrV+3atWHhN9G8LlQSkR53gnvo eGXLlo2Pj/f6ogKlYj799NOCggIEIb1798ZikyZNuEl6ejoCY2j8xMREry+q QbyEBPXr11fZWjqhcvKBAweeOXNG2Q3xM6JciiwOVWjbti03cTgKvmw3jzBi AIaAk0FOVlYW41UcgtekkgxgLYXV5s2bacHjj80BCI0YwqGozHD79u1Mg0cn YzyHyQSg0RisIhziOB2IL+Zsnu8Cd4/o6GhdKBVKIvFFPeJwQ9yIkrP1wWHb uXPnUimgVCgzSoJ6hgVqRW8HZ487XYNYYqmSAEUrjgU1SRWmWjQQ6vNt0rJl y5AGrjhgwABuZaeSUEVNmzaF5fe//z1vtriRQqfD0qdPH6aBq3MTNmVCir73 3nuwNG7cmAlwOuDhSMCWAuSJbQ23WQN2Hkix4/GNu1FfBlGTK9IPcZVt3bqV KkZF42pDqAMEXaj8TZs20aKuHVyAdGBoHKSBx9I3dJVEccEvgXp9PpyQkEB1 pnfAc+O3iuBVEmpg//79KBtbS0eOHJmdna1XDopn2ISePG3aNIOdnd9GjBjB oUwK595HOHZeVpCrgaVxcwU5ICqp5CIqyYCoJKGw2Kmkffv2qTQIimB59tln uciYAfGG3rdny5YtEb5GKC6KShJCDh2va9euyjJjxgxY6tSpoyyIWsv4oDRA 8IBIRu8RB0utWrWwlZp02kEldezY0VAGQ/zsvRqfN2zYEEEUfrRo0cJ5SAXb d8z9hWJjY2FH6KgsuBgRkHNCLWeVRIWltxMxOEQMo79Y5lCguLg4VRVQEIjM HV5QM67zaDMqgLVr13p8w9XN6ZVQQnxbKImkWr4QYBtWIShlGRwe6JyWYcKE CbrxzTffNDQsBqOSOnTooBKgJGxDV72q2rVrh8UxY8bo+fz7v/+7g0pCLE3v 1esHp6NNmzbqlRSPC+dXJcDJatSoEYyURTiDvPGqiddQNpx9hw5pdh5IsQN/ 1o3I7cKFC4bZ1Si3lQLihnrLEVyLe1FCBlLIc+2IMwBX1FUSLwG9uyD2zvYm fWpHN36rCF4lceSRauHS6w3qyWPVd5GqBGrIYE9LS2M+ECzwczwQoYJpMbu9 gu1rEFbmrpgu07i5ghwQlVRyEZVkQFSSUFgsVVKVKlX0BxAUE1uXuMiY4cMP PzRkha1g53tgUUlCyKHjTZw4UVk4tbJhRBu7MBnGROzdu3fhwoXTpk2Lioq6 5557VJDpdVRJhqjba6WSQKdOnfhuoWLFigiEHA4hLy+P4Yp57MDOnTu5iuM+ EGjp16CDSkIleHzDzPW3FnwtD+MWDY76cehpZgbXLDYxTArHlgK7DjyoeU7k 5V4i4abB9incjsyhLx7lzE2N0jLToEED1L9hOgLEtzC2b99eWYJRSYYmA4gm GBGdchE6RXcqMn36dAeVRI3fpUsXh5LwuFCNMRr83hNbVEHjxo2xWK9evQED BuAsW379UOHggRQ7hjFHBKFUenp6UlLS5s2boZHZgVPpBW5oGIYD0QQjh1Dh nFI06c2y4Ouvv9ZVErNF/rrTUiXh6nA4KAeCV0m4oqH3caI5IwSuQXWkqBNe ZYZNDh8+7Ll2+BJBbWBzpNe1DEftGfS1Ye2QIUMcuhf6TePmCnJAVFLJRVSS AVFJQmGxVEl169bV0+BBY1ZJkyZNMmTFZzfHFIhKEkIOHU/vx2KpkuC9ukqK jo5mvyYDblSSuc+MpUpC7FeuXDnYe/To4XwIKkY1hysFBQWJiYnDhg1TL66n TJmiNJedSoIuiIyMRAhqiPxVJyhLzF+otIPvqA0vutkEwD5glqxatYo7slQ9 BlAViBs9vq+9WBYMRrvAnvCudfPNNxvsfL3zwAMPKEswKsngDH/+85+VSlIF MBwsW9jtVFKvXr2c+2eqln1LUCFMBq3xb//2b8peo0aN4cOH27U7OHggfcb8 hXFoBAoEAwaVZNiQ7ZtUSdgXNzEoOI4wokpSaSwJePr00I5LSk5OZnnYEg11 T3liSLl7926P6d2CHeqMmAcobd++HXZc4A59C92k8XsFOSMqqeQiKsmAqCSh sASskvr27WvI6u677464OssuVVKzZs30BJs3bxaVJARMACpp586dZcqUqVix YteuXefMmbNx40Y8Lxgqh1AljRo1igFqzZo1/X5kh+PW7ZIhZEpNTV29erUa ucBBEJYq6fz580ym948lHK+NAOaoCee5GgxYjmHhvF4cSmOGI1BQRZMnT/b4 JqxzEEo4XjY8QczafdJFTd7lIO5Q82b5s2bNGoNIKSKVBNgBz5AzXctOJU2a NCnCNAWEgerVqyMN/GGLCcN85ojMR48e/eKLL9IVP/nkE7s87TzQUuxQyMDx cKS4do4cOQJxxIY/9yoJsJOe6uZKGOFTJan2ppSUFLPTummRtCSEKolwjj5O z4Iyc9SeQXKysdX8yaTLly/jQjb3yGUjmmE6FEotgB92hXGTxuvuCnJAVFLJ RVSSAVFJQmEJWCU988wz5jR4pnMRD9AI39Bp/ZUm7qiWKqlbt25FcGRCuBGA SoI4itAmFfH65AZnujOoJIMTuldJCOoqVKgAz+dUY/roFUsYYZq/pfKfPtQi 4hlOLsfuPWaVhFiLCSyHfuPoPL6+QLpCwW8oEeceWTrqLbfefZFjHFASy+FX SiIhrFVjlOyEEo6Rk9oh8jT0kNSB9PD4JkZzKOpLL72Eyjd0Ahw4cKBBLxSd Snruuef0RQLHc1BJX3/9NRZ/+ctf6uf97NmzUKaqGQJ3S6TRpzT3y7p16yjY 7eIfOw+0FDuc1FqvWJwpfly1UCqJUx/oTULKgVWPO34Hdu/evXo+cFf4YcCx XMAqCQcOJzfMVA/YA1AdF2dN4eQwBMeF3ZkrxOs7NR5TV1U2SEG66hdmWloa K9mhq6GbNMTNFeSAqKSSC1VScnLyj1fR14pKEpUk+CVglQTUAAdca5zzVh8C wDHyM2fO5OLx48fvu+8+g0rC8yLCNxFE0R2gEDYEoJK4ycsvv6zWfvzxx/Re pZIsndClSkJE9Oijj8KCIDA3N5eDU/Rh7GY4RN3w7hd6gXMsq24zSiWxY5tZ JXF+ZqgMy8ccDp9v5tWMzdApHIeOotJy8eJFWHD4agJ/AwcPHvSYhl1wxDrK Zk6vJJKaBEyfzMEglFBs9txDnKlmVLOEY94t96hA7UX4ZkJQCgihY8WKFWHU 5+IrOpXETWrXrq20w/r168uWLeugknDbZAvU3//+d5WAY9w6d+7MRc5j37hx YzV7xuHDhxs0aFCpUiXOcgBJhUyQXlUvP2WLm63d0Vl6oNdG7NAI/cL84ZaU P4VVSRyC5PGNe4IuyMzMVPkolcTxNSNHjlQHi2Tsg6prOr9+q2Opksw5mC1J SUncUB9LxSZaaBMl6jmcENcay4xa2rx5s8c38Z3Zq3EU7KGnBtChPqkf9enQ cYrZRMWvD1jiJo3CzRXkgEEl/fTTTyrkRv2ISirOUCXp4LJVa0UliUoS/BKw SmJXkF/84he/+93v+JEORAh6h4q33nqL4WiTJk2gjKpUqcJeMbpKQqRapkwZ GO+///4333yz6A9XKMEEoJIQsdDBmjVr9tprr8EV8fvOO+/UVZKlE7pUSZzt +fnnn+cilAvzt+uN5r06H4Kh0cF79YOqHt9cYYsWLeLXZ0aMGMHOPAaVxG+j UGLMuhYVv0Fw8VXzqFGjYOfQEljU12dYEvPbewXjPcNnXzjB3apVqwyJN27c aJBIRJ8eXLdTUvEtetS1GL73ylFOapJzSyBXcSOK8H33CqejZcuWHClm+Lxv 0akknKAWLVqwAC+99BJ+w6nq16/voJK8PgemlIPwwU2YnZYrV66svreFKJrN SbjHdujQAa5etWpVvcnyxIkT/IjS448/jnK+8cYbbCpFrdodnZ0HWoodKmWK F2glehH77BVKJXl9H3f2XIuhLQknkW2L8FKsmjp1Kh3Y8NrBr9/qWKokcw5m CzTO+PHjqYBQAFQX+9p5TIOkOE8FtSQ3AYbP0RJoKH40CseFH6gidl6FElT9 J/Py8tTgxCgTqampLtPouLmCHDCopPj4eD3qFpVUnPn555/xsDio8f3336u1 opJEJQl+WbhwYYTvE7FcxO03wjddkp4GN/AIbZCRmv6re/fufFmKBz0e34Yv aeL5jrhUfcfzqaeeWr9+PX68+OKLejKEeXwDj9iyKA9UKPFwtLv+RhSBGSxt 27bVk8F7YVSDXPBMZ/AZ4RvYjjAMvhpx7XxoZic074t89tlnEb45x7y+GasQ iiNG5adkCKdu1htVDeTk5CDuQnRkaFtBiLhhwwb1+dchQ4Yg+lKKg7NUqT4z CIQ8NugDtPft24fjZUzLyFOXb4hysTuEVfpM6Tq4OZhfVutzl+lA5ZklEqFQ MswcvnLlSrtDMLReTZw4EfLQrpCKixcvvvXWW5wUDjz00EM4WYY0lMl+x47p Z9lr4wzdunUz9PFD3Xbs2FF97xg3VTbrtG7dWqWBdIIT6vkggoWMolYqX778 k08+aRhlhhOKOy29miIL2lwfYILAHltRFUb4ZgVHbTscnZ0HMpY2T8qBc81g 3uOT7TgohvpKbFpuCKXvuXbEHDwci6gxFA9ehGJT7OuiGK6CBOzVxt3hqWHo 2+nXb3U4KbfBo8w5WOaJwsybN099JNfje+Fg/mA06pN9BQl8VZ/P3wCOBWqR zUBk8uTJ+lBBqDO768JzVcS5SaPj8gqyw6CSzp8/r0fduORvRPgvhIASpJL+ tuCZvy9t7ffvo/lPuVFJz81OfGHeNr9/SOZGJb33ztKPP1jt9+/Dd5eLSiol 6G9WEYumpKTYDbv2+h4KSOB8f8bD+ty5cw4fgxCEIEEEe/jwYYcnQgBOiNzM w3wsjQbYS8duNAGCrqysLDffgnFJfn4+dIHlMCLDYKhg4KAnu7WoEOfPSNnB D3Ry5kyXnDhxwjDj9PUEh3ngwIHCTrmMcwTl6zy4HprCMPuBDmoYdWWeJ80S Zw+0BG6JeDhgb0GFQFUZqmX16tUoxvz5883p8chwmJAteL8152CXJ04K7h7p 6emWbwAUuMSQBufIzT0EToI8Dx06dB2eegFcQQYcxiUJJZoSpJIK9edXJRXq z69KKtSfqKTSgF1nJEEQXBIfHx8VFRXwzF2lhNzc3DFjxsyePTuEmlEg19kD 2cFy9OjR1HGQCdAU7DZmniFBCBUhuYJEJYUrJUIlXfrHhazcE4X6+8dl2/d1 lwoun7pwqVB/2MQut3/843JW1sVC/V265Hb2JKHkIipJEIIEz6ZVq1YhgPE7 RqbUkpWVNX78+KVLlwY2fbHgzHX2wAsXLqiPLkErcWoRdrdzmN5QCIZQXUGi ksKVEqGSBKHEsXr16j59+pi7ZwuCUCh27drlZgR66WTfvn0OH64VQsL19MDc 3NyEhASOkYFKgj5av369SOCiI1RXkKikcEVUkiAIgiAIQvHhio8bXQrBLaKS whVRSYIgCIIgCIIQGKKSwhVRSYIgCIIgCIIQGKKSwhVRSYIgCIIgCIIQGKKS whVRSYIgCIIgCIIQGKKSwhVRSYIgCIIgCIIQGKKSwhVRSYIgCIIgCIIQGKKS whVRSYIgCIIgCIIQGKKSwhVRSaUTXtFFd+qR+eXLl53TIAGS6Z+EyM7OzsnJ KaIiCdcfOFjJ/R6i2T+DAfXg94oIJ06fPh2qrArlRXIPKTpKmw8LgntEJYUr opJKJ5s3b/Z4PBs2bCiKzI8cOYLMY2JinJPNnj0byQ4dOsTFEydOYHHQoEHn zp0rilIFQ2ZmZnx8fGJiouXadevWYe2FCxdCtTuEIgsWLOjdu3fnzp379u27 aNEiFCBUmbukXr16NWrUCCYHxKvwgW3bthnsx44dS0pK2rNnj3mTkydPrly5 MjY2dv78+Vu2bMnPzw+mAEFi8M8gSU5Ojo6OzsrKCkluxZz3338/IiJiwIAB wWdl50WWFOd7SJAEedWcPXsWmy9cuDAuLg43f2zivLuDBw8m+Thz5owyliof FoRCISopXBGVVDrZtGkTwon169cXReZUSV988YVzslmzZiEZHsdcRDjq8YFQ pyhKFQzQLIj6ypYtu337dvPahx9+GGv37dvnPsOUlJTPP/8csY151YEDB+69 996Ia6lQocK0adMCP4DCU6VKlZtvvjngzRGmRkVFod702wt+Q2lGRkbiLI8b N86wSWpqqudaxo8ffwOFksE/g+TKlSsJCQnDhg0LYSPL9cTBY820bt0aTvvq q68GuVNLL3KgON9DAib4qyYjI2Po0KF6GgjJvXv32u0ROmjIkCFMuXPnTmUv 6T4sCEWHqKRwRVRS6aQYqiQ8gtPS0kIVlIYWqiTQsGHD3Nxcw9oAVNK8efOw yVtvvWWwnzlzBrvAqvr167/33ntLliyZPn16x44dufdPPvkkyANxTzAqCXeV qT70ewsi3piYGBWnGeI91OrIkSNhj42N3b9//7Zt28aMGYPFFStWBHUYQRBa leT1efiMGTMmTJiQl5cXqjyvG3Yea0lmZib8NshY2tKLnCnO95DACP6qQQ6D Bw+GMT4+HjWTnp4+d+5c5mbZKkQvVXvUVZK3hPuwIBQdopLCFVFJJQg3fcLN afBcM59ig0oKLGe7BA4q6T998HeholDLo7DLOeQolQTefPNNw9pQqaSCgoKn nnoK9qZNmxq0GOoK9vLly586dSrgoygUwaikxYsX48xmZGToxqioKL7o/uqr r8zxHjeZPHmyOolHjx5lnAZ3crlf9y5qh0v/DPh6gXxAnnFxcaEa7nTdKJRK CgmWXqQT8jEyIczQLqvU1NSVK1fanf2NGzd+/fXXuiX4q+abb77B4syZM9VO UbZJkybBiDzNZUABsApqi2LKoJK8JdmHBaHoEJUUrohKKv7gobZlyxbEbIMH Dx41atSiRYv0sRJ5eXl4hs6ZM+fw4cN4FA4aNGjYsGFLly6F/cSJE7AM9hET E6NHm0olIbafMWMGEowePTohIcHcX/3YsWOwjx07NjIyEsLnyy+/hM/oCS5d uoR8pk6digxHjhy5evXqAwcOmFUSjNAaKBtKOH36dByCIQqFRsCBTJkyxXBc KAD+Hzp0KDbE2m+//dZQwrS0tPj4eOSMo0DOyBBlwLbYMJhq16FKeuihhypU qIAfqBB9raVKio2N7dKlS506dZo3b96rVy81oAC117JlywceeACb1KxZE7Ko TZs2XLV9+3YY77jjDsvIHMmwduDAgVxEnINtN2/eDEuLFi0ganbs2MFVONHw E6SvXr06Vr3zzjv6S2MWoHv37jC+//77jRo1qlev3htvvLFq1Sp9d1RJOTk5 PXr0wCHUqFHj6aefdtPhimND4A8GO84Ogrr8/HwGcoZ4DwWGEU6iG+fPnw8j Qke7fWEVTjRcYt26dRMnTqQH4uzAJ/Vk8IRly5YhJQsG9zh79qwhK7/+6fV3 JZLjx4/j1I8YMQL54OpDesOANRhDONzJDB1j69atw4cPf+KJJ3DiWrdunZyc jFW4CTz22GN33XUX7Fhr2BD19u677zZu3Bj+0K5dO1Q+7Q4ea+eEuPZhR/3g NyzPPffcSy+9pNf5tGnTYOzdu7fdUdh5kdexhg33EFTyeBv0phC/tzgUxu/b icTERGQLSYJrBL6NrHCMKSkphmR0qiVLlpglxoYNGyiIdGPwVw1uzpA8BvGF ywFp8L+hDLjecQlgVXp6+sKFCy1VkrfofVgQShyiksIVUUnFHzxS+XoQgQ37 TuB/PDS5Njc3F5YhQ4ZgLR7Nqv/57Nmz+byDnRasUoEKVdKECRPUKv5AkHn+ /Hm1a4gmrkJAgrCEaebNm6ce8fgBCaP6ugNmYlBJUHDsV8+iMjELpqJQRAI8 NP24sHfuVx0F0OUPnuYqZ6bBIoOHEHa8oUrq2LEj4g3Git99951aa1ZJ/fv3 Z8MTUlJYVatWjcEMDhNRKKQQjJUqVapbty4251YIgWB8/fXXLcuACsGBq2Hp PXv2ROJmzZqpgUtJSUmwo2C1a9eGpVy5ctBoXIvQV3V/goKABUHvr371K/yo WrUqUuJHmTJl4Glqd4h4Yfntb3+LVYirVVOa3y5w0CAeq86c6nSY4z2eemCI Tvn+HFGi3b5QGCRAfEtXofsB+KRKgxBXubfyYexdFy9u/NPr70r0+sJy2uEn fF3PV/16ywKiehiXLl3qXI0BQ8do1aoVFTfPGs4y1C5+3HrrrXRIgENWW+F4 aYSqQhr+RsV6HT3Wzgnp/+oUc5gSlDgXUWOVK1eG10FT2B2FnRc517DhHoJz N/Za1HAbpaP93uL4zsdw2zEDZa1uUOpmhTwNLaFQZ1BPdABdKEGaeXwSSb/9 eovmqgHsU2eeE4MKi9s6qKSi9mFBKHGISgpXRCUVc/jIw2OXwRjOV0JCAixR UVEMDKgmAB7TFy9ehHH79u20QCwwmEegyABAhSVUSR5fT4zs7Gyvb+ZevmxX c9NhX6NHj+bTkA/i48ePMwBQkxhwrjwEAykpKXAkxAAIbJizUkl4mrPfCGJX TtKL8jBnvyoJIBZiCRGlMC7C45tpEOqwPHi4M+hFzMOYuYhUEgIbCod27dqp tQaVFB8fT2XB9+qok/fee49SRW1i2X/pT3/6E4yoKzdFYoDK1iV9Hipm0r59 e8aBGRkZVEPqtFIlMWBeu3YtzhpOSp8+fWC57bbbVFAHlQQLQmIeFyoThwzL r3/9a+eCsU3HofLN8R4bDswHDqeC3WHaCqokj6/vEA4Eh6Ycm3UCv6Vknjt3 Ln0PdnqRqhCX/un3SvT6/MTjGxXCADgrK4v+CS2vyoyLiNemczUGDB2jbNmy 8EOvzwGUivn0008LCgpwjL1798ZikyZNuAmKd9NNN5UvX56zOKJCKNjr16+v srX0WDsnNKgkHDJFFifVbNu2rd4qaomdFznXsOEeYgBreepx16LFzS1uz549 9ARzK7YCDkDthlsoX0fgRsSczc2v8NLo6GhdKNlJJJ0QXjVMAMFomAkwLS2N GbK7r4NKKmofFoQSh6ikcEVUUjEnNjYWz6Ndu3YpC84Xh+hyHielJlTrBhIw MtS7neOJDIvqVcVgEtJJH/+CJzuz4hRJfC7j2a2/82Qv97i4OC6y64WKOry+ 1iWWWamkw4cP85Gq94NS/ef9qiQ8kdVWFGWIMbiIbT2+t8r6nE7sNl9EKsnr ixDuvPNOLKpiGFTSK6+8gkVsotdJo0aNYFR9VCxjzqZNm8K4Zs0aZUHENela VGc/Bqgskg4UJRxDf72MM4iUf/nLX7ioVNKMGTP0DZ9//nkY1VmjSlq9erVK gMqH5fbbb3euK76Zd+ihZI739u/f77HqXsVgDF5qlxVVkt5yhHiVBWBMy33B 9/QeVojndT936Z9+r0SVBkGvSoM8Efrqn/vBCWLORXTjpWN07dpVWXCiYalT p46y4OyU8UE/QaXhEPQecbDUqlULW6kuuA4qyeyEBpUEoFIjfJOfcIRdixYt nAcB2XmRcw07qyQqLL2dyM0tDuVMTk7GeXcYhqPuOXrX1rVr13p88yqY0yuh hMvZjUTyhu6qUW1neATodngC3y8pD3dQSUXtw4JQ4hCVFK6ISirmsA0ID9wt GozN+PxSPdP0rTgtkj7wmW/CVXcpqiQ9vCR8/8m35dxkwoQJ+q7ZoR3JmJ7v XQ3BTFJSkq6S2LalhjkYDs1ZJWFRj6YYAKiu++z4oZqWFEXX446LOBYsVq5c mbswqKQGDRpgEYWM0WjevDmMfL3vtYk5W7ZsCaN+OKjJiGtp1aoVVzFAxdmx LPCFCxf4CRUkQMCMlF26dOEqpZLYQqcYOXIkjB988AEXqZL0bmm4UeCQYXQI FyFGGD45fNnTHO+lp6fT0wwpqV/Y78sSqiTDwCV2AeVAMPqwqnYFm02Rv9e1 f/q9EgF+8PCnTZu2detWuKu5rnDLZZoQfmBLh44xceJEZdm2bRssajARYW9M wwCuvXv3wv1Q+KioqHvuucevrrdzQrNKAp06daLjVaxYMS0tzeEQHLzIuYYd VBIqgT6m62U3tzg37Nixw+Pr5KwbeY+165aGmudYTjcSyRuiq+bcuXO8w+OG ZvBM9gGYOXOmsjiopKL2YUEocYhKCldEJRVn8Azy2LNu3TrvVTVheHNIlcQg kFiqJMOAfe/V/up8nai6M1lSUFCgiqe/Kvde7cmvVBLzYWl12JPfWSUZjuv0 6dO6SmLOCFwNObPnf9GpJADRAcuTTz6Jy0dXSUqDWIJScXPLmBMKBUZEmMpy 5syZ6Vd59913sfbFF1/kKgao5n41iKY6d+6shpYoDCqpVq1ahg0TEhJgb926 NRct57irWbOmS5XkED6Z4z0Ebx5fFyBDyt27d5uDTx06gGGqLjYZUCXZ+R6b JDik3Y1/urkSvb7ZAxITEzkekEyZMsWgCJCGqxyEZDCYHcNSJdWtW1dXSdHR 0WzKNOBGJZmd0FIlwZk5Aq5Hjx7Oh+DgRc41bKeSMjIyOGZTH07odXGLcy6n ggPWDK0zcXFxMKo5Vczg9ssdmTWLmeCvGlQmB6LiJm84NNxXOapUrx8HlVTU PiwIJQ5RSeGKqKTiDB6d7B2RkpJy1ARfPwajkszvz/EU9lxthGJLDZ625l1z IDOKx9HQ+pAE79UuK0oluWzxCUAl8aDMR2Eedx8kZpWUnZ1dr169CN8QeENb UvXq1dlXbYsJ1XvQMuZkZ6S2bdtalqFfv34RvqElXLQLUNu3bw/7I488goIh DEOQhtDIrJLKli1r+OLJ5MmTYe/WrRsXA1NJXt88AKh8h2/lmOM9ZMhhHYao mF7qMF+EX5W0detW/OZkazocn0IPceOfbq5EBSo2NTUVDsBpTIDebIfAsrBB eKEIQCUhEi5TpkzFihW7du06Z86cjRs3ovYefPDB0Kok1CeVF7zI76eUnL3I roYtVRLODpOZ5+r3e4tzieUoKvpPZmam5SbQerwKcN15fNPNOV9WQV41qDE2 XUEOmz91xOn1KDkVdHhUHUqI2tbTF7UPC0KJQ1RSuCIqqZhDvWP4TjrOGp50 PHHBqCQ8x/UBLKq3OWMnfsgeMbb++MZv7FptxffthveNfAmpVBJ7huD5rved U2NDglFJ/DAT0uhhKscgF7VK8vriHMSWkBLsnKZisBdeeCHi2uFFZhhzKklC EIowjNTnmiOoHAgfrFINZ5YBKs44jCiS/k4YVWFWSUAf3AH+/Oc/R/g6CnIx YJXELj0OHzmynNOYsoWzBxB4CzW75VddiF+VBL1v9j1ElQzCL1686HXtn36v RK/po12IIXmB6H0CcWo8pi6yISQAlcQ+maqzpdfnb2yONKgkg8e6V0lQlxUq VKhevXr37t2xqkOHDs5HYedFzjVsVkk4p0ygj7BTuLnF+UW1fOndF3kvNXQY ViiJhBuXGqPkLJSCuWpQS+wkAO1j6GNJ1JQ7dhju8EXtw4JQ4hCVFK6ISirm 8C3fyJEj1bjgzMxMdjhhCBGMSvL4XqGrSXQZTkyePJlp8Dzl60QVmeMhzu8Y zpo1ixaOUEZ51NAkNXBAqSSUkE1OCAPobMiZb1CDVEnwXoYEiNMQ8OTk5CA2 Vr1xilolgV69eqm+SUolcVLlxo0bq1OGE9GgQYNKlSqpkdFLly5FmgcffNAy w9tuu02fpBdVwbmUa9SowcDeaxOgIuhipyYVzCM65cTgZpV03333qRLi1HA4 Pz+s4w1CJbEz2+7du+0SWMZ7dBv4G4uEXXCmDjXjliV+VRIiWPrw8uXL6Xvw GUaMys9d+qffKxGxKPNRk6GpGF7vdkUVD7tDHQZDACqJm7z88stq7ccff2zo cWfpsS5VEu4wjz76KCy4b6C27733Xvw2D4rUsfQivzVsVkk47x5fHzPLx5yb WxyuOFhw+PoUMTqcQ8YwPohTK1ieZSWRVAujPpmD3cUV8FWDA2ffP0gqu0sJ 1ZhrglvBefDboPWK2ocFocQhKilcEZVUzMHjiUFdZGQknkpTp07lh11UjBFk WxIfsohz2HkDv/WpGBCNcHdIibCBEybDor4RA/9RY5ARWPINMENH/XtJajZd rMK++CafyYJRSV7fW019kALjWEqn66CScPjsbqerJIQcbE6qVq1ahw4dEJ1W rVrV8P4cwSckCYz333//m2++qeyIcxCsMsM6dep07ty5VatWkFdYvPvuu/WP VNoFqJyoHHts3749dl25cmV+7cigkipWrHirDxQVESwL47k6bMobhEriSHZz c5jCMt7zXp11weObo5uzKwBEmw778quSvL6mIuVv8FX6OfxK7ybqxj/9Xone qx/rpHHRokWcNAz70vtEcTSKPi1kaAlAJaEwdIBmzZq99tprTZo0wW9O5KhU kqXHulRJAwYMwOLzzz/PRSgX5m/XG81r70XONWy4h7AlkQJh1rWoJhW/tziW xGNqRlRQmBi+T8TXR+aBnxs3bjRIJKJPD265l4CvGooyj+97DVHXYvjwtwFe R5bjkorahwWhxCEqKVwRlVT8wQMUj07OpsWQYP369erl3sWLFz2mr2YwnNP7 q/BZr57afLIjvMRTm1qJH7LXP+xCEPwjN8aNjA8Nsc25c+fwPFUfUoyLi2Nw or5HQ5KSkpSegpBJTk5mxzwVhnHeJNWLw/K4srKyYNTn7/L6pBMOBxEOHvor V648f/48Dz+EKglFRVzXqVMn8yrUT/ny5bFWj0xycnK6d+/OgUsRvm8nIVA0 9OFH5MaX6ggXdXteXt7bb79NZUTuuOMOhLj8KK2CrU7m17mIvtq1aweBw21/ //vfMyhVH6ulSkKQjECuYcOGTPbAAw/069dPzwc6C8dlyJyzQzurJBw7TiKc yi4ZJyo0z82FDanuCWJdQ59AM4zWDAPkETx7rh2EAnUJ96A+gvaBi+rTyxO/ /un1dyV6fUpqw4YN6uOkQ4YMQRBriIfhvSiJPu12aDE7Bi7zCNN4NzqnGqIS Hx8PGU5nqFGjBq4guBx+6/Nkmj3Wzgk/++wz1XsTCrRcuXIVKlTQ7y2cLR9C 3u4o7LzIuYYN95DU1FSPDfq0A863uFOnTmF3uBHZnTL6CT+nq9AnWtSByjNL JEKhZDlzuDeIqwa3RLtKsJshk/A6Uo3LOkXtw4JQ4hCVFK6ISipB4KlURHMK QVmYh/Tq5OfnQ4w4fOIEXoTi2XVKUUD7IELwO6GTS5DPdz70DC9cuMB42OGT PdcNBJnqizNmUGxoTNWJTgcnGnEXYnKEcAFUF05EWlqa5fTCSiVxEQGh3uYY EtgRyPIttF/gZginUW+W1RIMUKmobefKdOmffq9ErIWiN+fDD3rq3SmLFXQG hyeCg8fawY9NG4b5WBoNOHuRXQ0HjMMtzjAYKhg46MluLSrE+TNSdhTdVWOm mPuwINwQRCWFK6KShJILAhsOK0hMTKQbI4Tj93Plu/B2GFRSEREfHx8VFeX3 KzClitzc3DFjxsyePTuEsX14I15U3BAfFgRLRCWFK6KShBKN6nU/ePBgDlJg 70HLjiKC93qpJNxVVq1ahYDK8IWaUktWVtb48eMh4WXyZPeIFxUrxIcFwQ5R SeGKqCShpLN///6YmJioqKjIyMhx48bNnz/fYRpq4fLly3369Bk+fPh12Neu XbvsxryXNvbt2+fwgVHBAfGiYoL4sCDYISopXBGVJIQN4smCIAiCIFxnRCWF K6KSBEEQBEEQBCEwRCWFK6KSBEEQBEEQBCEwRCWFK6KSBEEQBEEQBCEwRCWF K6KSBEEQBEEQBCEwRCWFK6KSBEEQBEEQBCEwRCWFK6KSBEEQBEEQBCEwRCWF K6KSBEEQBEEQBCEwRCWFK6KShOCBF12+fPlGl0IoduDeUlBQcKNLEQLg3nDy K1euhCQ31EnYXy+nT58OVVaF8qLs7OycnJxQ7VrQKQ1+KwiBISopXBGVJATJ kSNHPB5PTEzMjS5IEZKZmRkfH5+YmGi5dt26dVh74cKFUO0OociCBQt69+7d uXPnvn37Llq0CAUIVeYuqVevXo0aNYLJAfEqvGLbtm268ezZs0lJSQsXLoyL i9u8efPJkycdcsjLy0vysXv37mBKEjyzZ8+Gkx86dCgkuSUnJ0dHR2dlZYUk t2LI+++/HxERMWDAgOCzsvQiO06cOIHTNGjQoHPnzgW/62LFsWPHcCHs2bPH vAoX0cqVK2NjY+fPn79ly5b8/HxzmkJdd+DgwYO89M6cOaOMYe+3ghAwopLC FVFJQpBQJX3xxRc3uiBFCDQLor6yZctu377dvPbhhx/G2n379rnPMCUl5fPP P0dsY1514MCBe++9N+JaKlSoMG3atMAPoPBUqVLl5ptvDnhzhKlRUVGoN/32 kpGRMXToUI8GAtq9e/faZbJs2TImGz16dMAlCQmzZs1CMRA6hiS3K1euJCQk DBs2LIQNLkWNg8eaad26NZz21VdfDXKnll7kAGQsHQZyKchdFx9w7ImJiZGR kTiucePGGdampqZ6rmX8+PEGoVTY6w46aMiQIUy5c+dOZS+JfisI1wdRSeGK qCQhSEqPSgINGzbMzc01rA1AJc2bNw+bvPXWWwb7mTNnsAusql+//nvvvbdk yZLp06d37NiRe//kk0+CPBD3BKOScFeZ6kO/tyDiHTx4MFwlPj4eciM9PX3u 3LmMxCzfTiONCurCTCV5fQHnjBkzJkyYkJeXF6o8ixQ7j7UkMzMTfhtkLG3p Rc6gVtPS0kJ4mm44uGpiYmLUhWBQSbgXjRw5EvbY2Nj9+/dv27ZtzJgxWFyx YoWeQ6GuO3qm2qOukrwl0G8F4fogKilcEZVUgnDTJxxPMb8n1G8+bnak0jir JDdZwQn9prmxKJUE3nzzTcPaUKmkgoKCp556CvamTZsatBiidNjLly9/6tSp gI+iUASjkhYvXgyXyMjI0I3ffPMNjDNnzlSje+AbkyZNgvGrr74y5ACXQECI VStXrgxMJQXv5P/pg78dVFKhLhYdSAnkGRcXF6rhTkVKoVRSSLD0Ip2Qj5EJ YYZ2WaWmpsKl7c74xo0bv/76a90SFRXF5iFcI2aVxCqaPHmyctSjR49S3eC2 TEuhrjuAAmAV1BbFlEEleUua3wrC9UFUUrgiKqn4g4fali1bEKcNHjx41KhR ixYt0sdH5OXl4Rk6Z86cY8eO4f+hQ4cOGjRoypQp3377rSGf48ePL1++HInx jJswYcLSpUsNQ2mQw7Jly5hg6tSpq1evPnv2rCGTS5curV+/HmuRZuTIkUhz 4MABs0pCVgkJCWPHjo2MjMSqL7/8UpdCCAawl5SUFOwOAXBo39IXBVRJDz30 UIUKFfADh6avtVRJsbGxXbp0qVOnTvPmzXv16qUGFKAeWrZs+cADD2CTmjVr Qha1adOGq7Zv3w7jHXfcYVkbSIa1AwcO5CLiHGy7efNmWFq0aAFRs2PHDq5C gAQ/Qfrq1atj1TvvvKO/NGYBunfvDuP777/fqFGjevXqvfHGG6tWrdJ3R5WU k5PTo0cPHEKNGjWefvppNx2uODYEHmKww20QehmCQPapw/+GxGvWrKFTnT59 2qVKolPB7detWzdx4kT6J84UPFZP5sbJ4dI448OGDcOlNH36dFxuZpXkfFUS XHFwgxEjRiAfhKlIb7jiYAzhcCdCx9i6devw4cOfeOIJnLjWrVsnJydjVUxM zGOPPXbXXXfBjrWGDVFX7777buPGjeEP7dq1mz9/Pu0OHmvnhDhxsKNO8BuW 55577qWXXtLredq0aTD27t3b7ijsvMjrWKsFBQU4s7j7cREVO94GvSnE+WbF wvh9O5GYmIhsIUlwjUDLICscI25xhmR0pCVLlpglxoYNGyiIdCO8C1IoPz+f 8segkuB4MMJddSNOHIy4HLhYqOsO1zvcHqvS09MXLlxoqZK8ReO3glCiEZUU rohKKv7gkcrXgwhs2HcC/+OhybW5ubmwQBwhbGAa1VkCT3+VycmTJ/n409Pg 6aniZ0QCqu86s+JDWY/r8GSHEFM92wFjUYNKwr6YFRKorObNm6cCgxUrVujF wOEcPny4yOsxCKiSOnbsiBpjrPjdd9+ptWaV1L9/fzY8ISWFVbVq1RjMIOBB FAopBGOlSpXq1q2LzbkVahvG119/3bIMONE4oWpYes+ePZG4WbNmauBSUlIS 7ChY7dq1YSlXrhw0Gtci9FXdn6AaYEHQ+6tf/Qo/qlatipT4UaZMGXia2h0i Xlh++9vfYhXiatWUpnfmsQS6A+cUsZmbimXfHsPYfLgigkw4D8p85swZlyqJ ToVYl5cDnRPAY/Wc/To5XJFjQABHZyAr+qqukpyvSq8vRKcdPsNX93ztr7cy IMKHcenSpW7qyiV0jFatWlFx86zhLEPt4sett95KhwQ4TLUVjpFGqCqk4W9U ptfRY+2ckP6vQnoOU4IS5yJqqXLlyvA6aAq7o7DzIudaRVF5IpgY52vstajh Nko7+71Z8S2Q4XZqBmpa3XjVnQ15qjYdAnUG9cSTrgslSDNKpPPnz+vplcuZ VRIPFhg0HVudoK0cSmt53XmvKixu66CSisJvBaFEIyopXBGVVMzhIw+PXQZg OF8JCQmwREVFMTCgSgKIGbKzs72+V6OMH/CYYybYim8dY2Nj+RRGuMj37Xyd qBLMnTuX8+giOmUm+uR1mzdv5qM/JSUFboMnPsIY7l2pJGTF5iE8Q/n4Pn78 OMMGNfUBA1qwYcMG8zCfYohSSQhsKBzatWun1hpUUnx8PJUF36ujlt577z1K FbWJZf+lP/3pTzDizLopEgNUti7p81Axk/bt2zMOzMjIoBpS55EqiQHz2rVr cR5xxvv06QPLbbfdpoI6qCRYEBLzuBCt4ZBh+fWvf+1cMPqVm8ZBeBGViD4j GWqYYSRnFCysSvL4+hHhoHCYmzZtooX148bJEXmyjxO0FdPgSqE/6wfl96r0 +nzG4xshwmA4KyuLV0F6eroqMy4NWFAqv0fnHjpG2bJl4YdenwMoFfPpp58W FBTguHr37o3FJk2acBMU6aabbipfvjzrHJVAwV6/fn2VraXH2jmhQSXhMCmy cL1jsW3btnqrqCV2XuRcqwaVZABrebpxH6PFzc1qz549PPvm1nkFTjq1G2QO X0dAfDFnc/MrPDM6OloXSnYSScesktjcZr5d8LJymOzF8roDaWlpzJD3ZAeV VBR+KwglGlFJ4YqopGIOdA2eR7t27VIWnC8O0eU8Tkol4cml0lDO4FnMRT5h R44cefHiRZUG6ePi4th5jAnw1NM7ojBABWrGJHa0UDGG1xfTsoRKJTErPPH1 N6XsG4/dcZEBLYLV0NRR0aNUktdXb3feeScWVfUaVNIrr7yCRWyiNkdVNGrU CEbVR8Uy5mzatCmMa9asURZEXJOuRXX2Y4DKIulAKcMx9NfLOBdI+Ze//IWL SiXNmDFD3/D555+HUZ1HqqTVq1erBDjvsNx+++3OdcU38357KKl3+NAyuh3S 0uPrEUq5UViVpLccIQfugvGtGyc/fPgw0+j99NRYDxW0+70qVRoEwCoN8kQY rH/6ByeLOYfwJkzH6Nq1q7LgRMNSp04dZcHZKeODfoKKQrH1HnGw1KpVC1up KaMdVJLZCQ0qCeBij/BNfsIRdi1atHAeBGTnRc616qySqLD0diI3NyuUMzk5 GefaYRgOh+p4rp0PYe3atXwxZU6vhBIuZzcSyWulkvbv3++x6pRICTNixAjL fOyuO3gC22GVVzuopKLwW0Eo0YhKCldEJRVz2AkED9wtGozH+PyiSkJgoEcd fFCqLu589a0GGphhAr581uHwDdUdjm9ZDaFLUlKSrpKYFaJcvcDsBq9iXQa0 W7duDapqriO6SvL6+qVgsXLlygybDSqpQYMGWETVxWg0b94cRlXDljFny5Yt YVQtgF5f3UZcS6tWrbiKASrq2bLAFy5c4CdUkAABM1J26dKFq5RKYsujAiIa xg8++ICLVEl6VzTcKHDIMDqEixAgDJ+cv+x57tw5vmlHxeq5YXfsF6rG7BdW JanhGIQdRDkozI2Tb9++3fJK4WWoVJLfqxLgB6ti2rRpcHVckuZ6w+2XaUL4 sS06xsSJE5Vl27ZtsKjBRIS9MQ2Dtvbu3Qv3Q4GjoqLuuecev7rezgnNKgl0 6tSJjlexYsW0tDSHQ3DwIudadVBJqATel3SN7OZm5QZK+9mzZ+tGNmXadUtD zXN0pxuJ5LVSSenp6Sy8ISWVPntLGrC77ry+4Use3yQPyuKgkorCbwWhRCMq KVwRlVScwTPIY8+6deu8V1WS4c0hx7wrlcQAEkGd3Y6YgBnq8M0tR/6qwugv w71X++0rlaQ6PlnCbZnGcoal4olBJQGIDliefPJJXD66SlIaxBIcNTe3jDmh UGBEhKks0AjTr/Luu+9i7YsvvshVDFDN/WoQTXXu3FkNLVEYVFKtWrUMGyYk JMDeunVrLlrOcVezZk2XKskhfMIqhHZIM2PGDIMvcc4ujvpXNVAolWRwKjYf UCW5cXK7NBx1QpXk5qr0+mYSSExMVIMBwZQpUwzqAGm4yllUFgqzY1iqpLp1 6+oqKTo6mk2ZBtyoJLMTWqoknEqOgOvRo4fzITh4kXOt2qkkiO7IyMihQ4fq wwm97m5WbuAgNUPrTFxcHIxqThUzq1at4o7MmsWMWSVB8nh8HecMKXfv3u0x STav43WH5wXHmer146CSisJvBaFEIyopXBGVVJzBo5O9I1JSUo6a4OtHNyqJ 71H1PmAGtm7daohOCbvxMzhEYTj2WR+A4L3aQUWpJA7sxTPaXGA1/DkMVFJ2 dna9evUifEPgDW1J1atXZ1+1LSZUr0jLmJOdkdq2bWtZhn79+kX4hpZw0S5A bd++PeyPPPIICoYwDEEaQiOzSipbtqzhiyeTJ0+GvVu3blwMTCV5ffMA4OTa fSsHO+UrdITlhgJcvnyZoRc00ZSrsKHH4xuhb9l5SeFXJblxcnqv3pxHOKBJ XQh+r0r9eFNTU+EMnOQE6E14CDILG5D7JQCVhEi4TJkyFStW7Nq165w5czZu 3Igae/DBB0OrklCHVF7wIr+fUvLrRZa1aqmScEaYzDxXv5ublRssR1HRZzIz My03gdaj6oFje3zjQ50vK7NKQnoOhjJoSbZhGWZZcbjuvFen16PkVNDJUXUo IWpbT18UfisIJRpRSeGKqKRiDj8paPhOOs4annQ8cW5UEj9phOBTH65y7ty5 L7/8ktMcZWRk8BGsd9vDw5exihrNxDfqhreLfOWoVBLCKo+vH4j+0MdvFFjt PQxUktcX5yC2hJRg5zQVg73wwgsR1w4vMsOYU0kSglCEYaQ+1xxB+Afhg1Wq QdAyQD18+DCMKJL+ThhVbVZJQB/cAf785z9H+DoKcjFglcQuPYapvQhiKk6u hRjM0NfLq6kkO5xblPyqJDdOzl5MhjRq7JIKg/1eld5rP7fEY+flo/cJxGny +GbkcziuwhKASmKfTNXZ0uvzNzZHGlSSwWPdqyQoygoVKlSvXr179+5Y1aFD B+ejsPMi51o1qyScRybQR9gp3Nys/KJavnSX5sgdQ0dohZJIUHBqjJKzULKc CZwCn3NuqOPlh8b0C8H5uvNe7W7ngOGeXxR+KwglGlFJ4YqopGIO3/KNHDlS jQvOzMxkhxOGEG5UEoIHDilatGgRn9o473xucggGHvR8c7h8+XL6g0owefJk lS3HI2PvamiSGiagVBKewsxKxfN49LMn1axZs2gJD5UEevXqpfomKZXESZUb N26sThnES4MGDSpVqqRGRi9duhRpHnzwQcsMb7vtNn2SXpxizqVco0YNpVgt A1QEXezUpAJ4RKecGNysku677z5VQkRoHM7PD+t4g1BJ7MC2e/dugx1+xT5I CO3sJjbE0eVeCwU+/B+/ze/AdfyqJDdOjr2wwRQVwjSoLr7t11WS36sScSnz UROjqXhe74LFWcVgdziuwhKASuImL7/8slr78ccfG3rcWXqsS5WEe86jjz4K C+4AqOF7770Xv/V5NsxYepHfWjWrJJxrj6+PmeVjzs3NCj4JCw5fTWJjAF7h MY0P4tQKlmdWSSTVqqhP5mB3cVmqJN5+cQj0Q2zLeXvUPHVed9cdqjHXBLeC 8+C3QesVhd8KQolGVFK4IiqpmIPHEwO5yMhIPJWmTp3Kj7moGMONSvL6XqTz nTkeqYhq2FUDIYf6wkt6ejoTINjDXpgA2er96+AtasQxQke+72VwqH8vCTEM Czlq1CgEG5xaGRa1r7BRSagQdrfTVRJCDjYnVatWrUOHDohOq1atanh/juAT kgTG+++//80331R2xDkIVplhnTp1Onfu3KpVK8grLN599936RyrtAlROVI49 tm/fHruuXLkyv3ZkUEkVK1a81QeKigiWhfFcHTblDUIlcSS7uTmMwaHHN5N8 1LUYPkms4/6rsn5Vktedk6uZn5EG1avS6yrJ71XpvfrhThoXLVrECcSwL71/ FEem6JNGBk8AKgkFoAM0a9bstddea9KkCX5zIkelkiw91qVKGjBgABaff/55 LkK5MH+73mheey9yrlWDSmLrIQXCrGtRTSp+b1YsicfUdKigMDF8n4gvlAxf avb6vn1skEhEnx7cci+WKsl7dX4Sj282e9U9FRpNJQjsuvNevXYsxyUVhd8K QolGVFK4Iiqp+IMHKB6dnEGLIcH69evVy72LFy96TF/NyMrKglGf58rrewE4 e/Zs9aFMRHdq1mKCIBxxAkNHhIVxcXH67OLk3LlzeHqqzyYiDUMR/bNKAJIB YSQjTEaSekTEh6zDuObixsKFCxHXderUybwKR1q+fHms1SOTnJyc7t27c+BS hO/bSQgUDX34EbnxpTrCRd2el5f39ttvUxmRO+64AyEuP0qrYKuT+XUuoq92 7dpB4HDb3//+9wxK1cdqqZIQJCOQa9iwIZM98MAD/fr10/OBzsJxGTLn7NDO KgnHDiWOmNOQbOXKlR4b7Gbq8171ZDi/wx6JpVMhkPZcOyDFjZMnJSWptwEj R45MTk5mt1IlGbz+rkqvT0lt2LBBfagU1x0CWkNsjCsUJdGn4A4es2NAOUaY xrvROVXzXHx8PGQ4naFGjRq4eOFy+K0mG/RaeaydE3722Weq9yZUZ7ly5SpU qKB/KIqz5UPI2x2FnRc51yrnXlM9wVJTU+1cTp92wPlmderUKewON1i700Tf 4Od0FfrkijpQeWaJRCiU7AbfcdpS85WCA2H/TwJ3MvSkDey68169dlTjsk5R +K0glGhEJYUropJKEHgqBT+nEGL106dPO3ysBAkghZwjYfgMCmPXBUWBBM77 CnsQZKovzphBJaOq9Y9YKXCiEXchDkcI53f+KzOoeYhiy+mFlUriIgJCNdN7 qGBHIMu30MUEN06O84Jo1m/l+70qsRZaz5wPP+6pd6284dAZHJ4IDh5rBz8/ bRjmY2k04OxFdrUaMA43K8NgqGDgoCe7taiQwO6WKDlEKO42hTo1gVEM/VYQ bjiiksIVUUmCUKowqKQiIj4+Pioqyu9XYEotubm5Y8aMmT17dgjj/PBDvKi4 IX4rCJaISgpXRCUJQqni+qgk3FVWrVqFgMrwhRrB6+tGOH78+KVLl8pEys6I FxUrxG8FwQ5RSeGKqCRBKFVcvny5T58+w4cPvw772rVrl92Y99LMvn37StCg vBuOeFExQfxWEOwQlRSuiEoSBEEQBEEQhMAQlRSuiEoSBEEQBEEQhMAQlRSu iEoSBEEQBEEQhMAQlRSuiEoSBEEQBEEQhMAQlRSuiEoSBEEQBEEQhMAQlRSu iEoSBEEQBEEQhMAQlRSuiEoSBEEQBEEQhMAQlRSuiEoSBEEQBEEQhMAQlRSu iEoSBEEQQsvp06dDlRWeUAUFBS4TZ2dn5+TkhGrXgg7OwuXLl290KQShOCIq KVwRlSQI7kGcsGDBgk8++aRLly5vv/12VFTU8ePHb3ShBKF48f7770dERAwY MCD4rKB6YmJitm3b5ibxiRMnPB7PoEGDzp07F/yuixXHjh1LSkras2ePedXJ kydXrlwZGxs7f/78LVu25Ofnm9OcPXsWmy9cuDAuLm7z5s3YxHl3Bw8eTPJx 5swZZUxOTo6Ojs7KygryWAQh/BCVFK6IShIEl6SlpdWtWzfiWsqWLTtw4MAA cktJSfn8888R3oS8nGTSpEnI/9KlS0WUv1CqKJS7tm7dGpfGq6++GuROIXai oqIWLFjg8iF16NAhjw/IpSB3XXzAsScmJkZGRuK4xo0bZ1ibmprquZbx48cb hFJGRsbQoUP1NBCSe/futdsjdNCQIUOYcufOncp+5cqVhISEYcOGhbChUBDC A1FJ4YqoJEFww6lTp+677z7EfvXq1Xv77bfnzZuHaOHZZ5+96aabYIyOji5s hsgBG7711ltFUVrw4IMPIv/vvvuuiPIXShWFctfMzMzp06cHGUvj2TTVh/sn FML4tLS0gwcPBrPfYgV0YkxMjFI3BpWUm5s7cuRI2GNjY/fv379t27YxY8Zg ccWKFXoOgwcPhjE+Ph41k56ePnfuXOZm2SqEOpwxY4bao66S1NoJEybk5eUV 0SELQklEVFK4IipJEPySn5//2GOPIUr85S9/eeHCBX3VqFGjYC9Xrty6desK laeoJKEEUdTuambx4sWI0jMyMuwShHyMTAgztMsqNTV15cqV0BqWazdu3Pj1 11/rlqioKDYPffXVV2aVxCqaPHmyeogfPXqU6ubIkSO0fPPNN1icOXOm2inK NmnSJBiRp7kMKABWQW1RTBlUktcngWGPi4uzOwpBKIWISgpXRCUJgl+2bt2K EPGOO+6wfE3duXNnrH3llVe4eOLEiaeeeuqvf/2rnubbb7+F8eOPP/b6XpK3 bNnygQcewFY1a9aEvU2bNkyG6AWL2N3y5cufeeYZrP3Nb34zePBgXez4zZ+Z lC9fHvlD3OH3woULQ1kdQjFGudDw4cOfeOKJGjVqtG7dOjk5GatiYmLgD3fd dRfsWGvYMCEh4d13323cuHG9evXatWs3f/582v266+bNmwcOHNiiRYsqVars 2LED9i+++AL2RYsW4Tcszz333EsvvXT27Fm1r2nTpsHYu3dvu6PgCKOpU6ea Vx0/fjw2NnbEiBGDBg1C8L9lyxb14qKgoACCYsqUKVw8dOjQeBv0ppBjx47h 2MeOHRsZGYmSf/nllzhkQ2FOnTrlXO2JiYnIFpIEIghaBlnhGFNSUgzJZs2a heNasmSJWWJs2LCBgkg34tqHFMrPz6f8MaikUaNGwXjgwAHdiBMHIwQXF9ev Xw/JYxBfy5YtQxr8byhDTk7OsGHDsCo9PR03DUuVBFDtWIXqda4TQSg9iEoK V0QlCYJfEEEhROzSpYvlWsSBWHv//fdzEUoKi48//rg5zQsvvOD1tUwhEIXm gqVSpUp169Z9+OGHmaxnz54w/vGPf8T/ZcqUQUDL0U8Ia9VYA7/5DxkyBHmy K2CtWrXwG+FxCGtDKM7QhVq1akVdT/+pWrVqjx498OPWW2+tUKECjYjk1VZQ HDRCVSENf8PtvS7ctVmzZkyPnJOSkmDv378/FlVIz2FKb7zxBhcR8FeuXLlc uXLQFHZHsXr1asThiPANdkTm7D+GyJ8NImxMYdsNiopFJGBiXCljr0UNt1Hj 9U6ePMkxO6gBKC+unTdvnlIx0CA0Qkw5VPv06dNVLzUoUDX8R7XpEKgzqCes Wrp0qS6UIM0okc6fP6+nV69lzCqJBwsMmo6tTtBWDqVlnzrznBhUWNzWQSVB mfIQHHYhCKUKUUnhiqgkQfDL66+/jjAvKirKci0in7JlyyJBdna214WKIZZd mBh2gtdee43tR7t27apduzYsH3zwAdO4zF963JVO6EJwyPj4eK9v5L5SMZ9+ +mlBQUFOTk7v3r2x2KRJE26Snp4OTV2+fPnExESvL/xGNI4E9evXV9k6u+vA gQP1ydAMKun48eMUWRs2bMBi27ZtuYnDUUycOBFxuLnpdsGCBRx3Q4mRlZVF SYJD8JpUkgGspbDavHkzLXj8jR49mgE/tQaKygy3b9/ONHv27KEY+fbbb+1K C41G7QaZw9FYEF/M2TzfBQRadHS0LpTsJJKOWSWxuc18U0pJSYEdWswuKyaA YDTMBJiWlsYMc3NzvY4qCbWEVaNGjbLbhSCUNkQlhSuikgTBL02bNkVct2bN GrsEv/jFL5AA0Y43FCoJAkfvEbRlyxYY7733Xi6KShIcoAt17dpVWWbMmAFL nTp1lOXUqVNlfFAaIMhHyK33iIOlVq1a2EpNGe3grh07djSUwaCSwNy5c2Fp 2LDhrFmz8KNFixbOg4DYvmPu5xYbGws7LzQC0QFxwQ8qOaskKiy9nYjSA/JE b9bhQJ64uDhVFcnJybt27XIYhsOhOob5ENauXct5FczplVBKSEhwI5G8Vipp //79lp0SKWFGjBhhmY9qO9u0aZNuhyeMHTsWdhwpLQ4qKTs7m8crwYMgEFFJ 4YqoJEHwS8uWLRHaOYzuueeee5CAEUXwKunDDz805F+lShXY+e5XVJLgAF1o 4sSJyrJt2zZY1GAiAtEEo2Gi+L1798LJp02bFhUVRZdWY08c3HXChAmGMphV EujUqRMbnipWrJiWluZwCHl5eQzCzd+HxSXGVSjk1q1boQh08eKgklAJWGWY nI2d02DcosExO6NHj3YooQFcfdhk9uzZuhEyxKFbGmoeAofH4lciea1UUnp6 OgtvSHn48GHY2VvSAG4gbOGCYDSIvvXr13OSB2VxUEkIG1hyw1Q2glBqEZUU rohKEgS/8CuZ/fr1s1x75swZrC1fvjxfaAevkiZNmmTYRfPmzWHnOAJRSYID dCG9w5WlSuKXv5RKio6OZoOpATcqydy5y1Il4TIpV64c7D169HA+BKWSzEE4 LrHExETOMECmTJmiNJedSsrIyIiMjBw6dKjhclixYoXHHl7ObliyZIm5dSYu Lg5GTmdhyapVq7gjs2YxY1ZJkDzsOGdIuXv3brNkA6hMSCqsmjFjhuHQTp8+ PciHXj8OKgmb28lYQSidiEoKV0QlCYJf2GfJEGcqVq5cibWPPvooF6limjVr pqfZvHmze5XUt29fwy7uvvtu2I8fP+4+f1FJpZMAVBIi4TJlylSsWLFr165z 5szZuHHjnj176D8hVEmcMz/CN1Ge308pIWJHEG6XDDIqNTV19erV/FoQ4JBA S5V0/vx5Jtu3b58hH85CAEFx1ITzXA0GLEdRcQK6zMxMy02g9ah6Jk+e7PFN N+cslMwqCek5GMqgJdmGpX8yyeurMTZdQQ6bP3XE6fUoORXsmIeqQwlR23p6 iKPCCklBCG9EJYUropIEwS/79+9ngMcR8ToIzxo1aoRVamruU6dOYfGWW265 ePGiSvb/2rufkKi+MIzjBiG2ChJpEQTRRtpGC124aREUKGQQLlNwbbRwk2Bg oAthaFmitBHCgUAJhRAhWmgoupFcmoggE6iDsyri93BfOhzPvTOO92dpd76f hYzXO3/uzBXfx3vOe3K5XGJKevLkif9oVna2tbX5G1UgaWNTU9OJHt+qXAtW qB0pUpLCUZ3XHqQUxQ3rdBekpMTTtZqUtLq62tDQoHO4u7tbP3r48GHlo7CB YUGDuFJUivh/sFSlW3M5a3wdT0nFYtF2UKSKP4uOzgat+QlFt5UjgsZxFbgr X/7wRZu5o1eSOP3KRSQlODdHqXJQSuwEbs0orOeGO17tM3h0LSS9S9bUTtkn GGNpbLhdBcEVpW/fvmmjYlQVbw9QE0hJWUVKAqrR19dXFzVSdguRlKKRKio+ rbWCm+cuNvPdDfJXVLlx40aQYvL5vDVq8J/FNQ1z0yL0G9ra2qotHR0dJ3p8 62ym4vbU3gL8C1KkJLtLZ2en++nTp0+DEXcVTtdjU5JK99u3b2vL+Ph4oVDQ L4tuv337tsJRWJeGL1+++BtV7Vsrb9duzqUkG9gWT0nv37+3MWaJf+Z0+HbF 5MOHD7ZFOcWWatVLtS3fv3/XFh2+a8UfWF9fH4zND7LWCnpt8f1dRLLrX6Wj zRzKBaXElGSztHQI1jVC952fnx/0+tSVouLNxv4pUrmNAb2NhRi7l04e3Q6y nnXDSzw6oDaRkrKKlARUQxWILWNUFy1C9Pjx4zt37ly4cKEuWowmGMzT29tr e966dUvJ5fLly1evXg1SjOpPu/vNmzd7enpso5WdTU1N+trc3Hz//v3Gxkbd vnbtmp/Cqnl8u0BQX19/7969xH+kI5NSpCSV1nYqKll3dXXppNLtK1eu+Cmp wul6bEoaGBjQt3fv3rVvlVzs8cuNRiv97ocwNTUVbLfWCqKQ9e7dO2vLNjw8 bKPOgpT09etX21kBYfwod0lFgevFixeDUV9rbVe+0G1tUSrxX4ksLy8nvlQL JsH6RNbgbnp6Otj548ePQUQyfnvwxGdJTEmi98Fe3ps3b3K5nN1WRnM7WCgb jBZvGjnq9evXic9lrCVg4rwkm1HlGqoDICVlFSkJqJIqGcUTt+amaWlpWVhY CPY8ODhQten2bG1tnZ2d1Q0FFn83FW/2f3VVjLbFNQ3r7u62NZguXbqk+jZY f7Oax1dZ1d7ebg8yOjr6Z94SnDvPnj3TJ+7/n//Tp0/a8uDBA3+369eva6Ob ojI5OWlz3+qihWXHxsbsIqmChrtL/HSNP5d5/vy5tqtoL0XrDV28eLGhocGW NDKPHj0KLo8Gdnd3X758qeQSXFspFotzc3Nu+dehoSHFBJc4rPeaGwm2trZW bgiZ33ZgZWVFx2szoZSPdDh+fNvc3NTTKVP4ndJ91uXAltN1LLzo2IOdlfLi EclYUErsHF763d873tFOB6J85I5LCdFvk16KZk2WexPij+ZTCNU+i4uL8R+9 evVKT1TuDQFqECkpq0hJwImoLPn8+XM+n1cwic+b8KmiW11drVxLqAjc2dlx M4z8f86rgtXd41OtT/T4uvv29vaxHbSAUrTuz8bGRoW/CMHpWg09WnyaT+LG gA0nS7yWUYp+DU/3xN7b29va2kqcRhRMhvo/bNJTuZ/qDam8jFQ5euUKoYq0 J/po0rFFaa3fJgBDSsoqUhJwfpQbwgTUoMnJyZGRkWPXEsJfUygURkdHJyYm +McL4CMlZRUpCTg/SEmAo79N09PTKstpaH8ebG9v53K5fD5PA3AgQErKKlIS cH7MzMz09/cHs5CAWra0tFSucwL+ppWVlQqL5AK1jJSUVaQkAAAAIB1SUlaR kgAAAIB0SElZRUoCAAAA0iElZRUpCQAAAEiHlJRVpCQAAAAgHVJSVpGSAAAA gHRISVlFSgIAAADSISVlFSkJAAAASIeUlFWkJAAAACAdUlJWkZIAAACAdEhJ WUVKAgAAANIhJWUVKQkAAABIh5SUVaQkAAAAIB1SUlYVi0VSEgAAAJCCaumD g4Ozruhx+g4PD/XJ6utZn2IAAADAv8QV0mdd0eP0/fjxg0F3AAAAwEnZcDuV 02dd0eOPsBRMUAIAAACqpOKZC0nZ9uvXL/uU9ZWhdwAAAEAFKpiteC4Wiyqk z7qWxx+kz1ef+P7+vjVzAAAAABBn7RpUNisrEZFqxM+fP5WV7HMHAAAAEFBK UsGssvmsK3cAAAAAAAAAAAAAAAAAAAAAAAAAteU/3BZjUg== "], {{0, 89.}, {561., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{561., 89.}, PlotRange->{{0, 561.}, {0, 89.}}]], "Output", CellChangeTimes->{ 3.763378891801656*^9, 3.763404159273611*^9, 3.763417110525358*^9, 3.763417146630848*^9, {3.763417997817902*^9, 3.76341800481327*^9}, 3.7637439990587597`*^9, 3.763744433832979*^9, 3.763744904434998*^9, 3.76381006440456*^9, 3.763815810210495*^9, 3.7639148421017237`*^9, 3.76391613589751*^9, 3.763916334190456*^9, 3.7640056875941963`*^9, 3.764081104920312*^9, 3.764595508407181*^9, 3.7646011107064953`*^9, 3.797848670112473*^9, 3.7979224728880653`*^9, 3.798178762640615*^9, 3.7981861850753527`*^9}, CellLabel->"Out[13]="] }, Open ]], Cell["Pick a non-default uninitialized net:", "Text", CellChangeTimes->{{3.763378864569222*^9, 3.763378874753018*^9}, 3.763386727447794*^9, {3.7638154760185623`*^9, 3.763815481882222*^9}, { 3.7638155196419477`*^9, 3.763815521121449*^9}, 3.764588924998267*^9, { 3.7646144909325933`*^9, 3.7646145005309887`*^9}, {3.768659746344496*^9, 3.768659767912966*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetModel", "[", RowBox[{ RowBox[{"{", RowBox[{"\"\\"", ",", RowBox[{"\"\\"", "\[Rule]", " ", "\"\\""}], ",", " ", RowBox[{"\"\\"", "\[Rule]", " ", "\"\\""}]}], "}"}], ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{{3.763386371899088*^9, 3.763386375258068*^9}, 3.763404216501815*^9, 3.763417037618573*^9, {3.763417129146069*^9, 3.763417141784073*^9}, {3.763417986203992*^9, 3.763418001642385*^9}, { 3.763744070673006*^9, 3.76374407921362*^9}, {3.763812710677999*^9, 3.763812712131291*^9}, {3.763812852566313*^9, 3.763812869030425*^9}, { 3.763815448554677*^9, 3.7638154557443953`*^9}, {3.763914898124238*^9, 3.763914901808754*^9}, {3.763916148808392*^9, 3.763916150184573*^9}, { 3.764081305538369*^9, 3.7640813122614737`*^9}, {3.7686597742412987`*^9, 3.768659782576825*^9}, 3.797848641999659*^9}, CellLabel->"In[14]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsnQl8FEXa/4PAcqkgyKEcshyysi8qqKzCKq4uq8iy7KLwKrqsoOutKKAI iEw4DYcQgwZMgHCFhFwkQAIJOSDhJhcJJJCDI4Eck8OEzLr6V/f/yzybeivd Mz2dzEwgw/P9BD7T1dXV1dVPVz2/rqN/PW36hDducXNzm9kW/02YOvsPM2ZM nfN8J2xM/GDm229+8M/Xx3ww659v/nPGo9NaIjAC/z5o5eZW+/s/DMMwDMMw DMMwDMMwDMMwDMMwDMMwEj/++GNFRUVRUdGVK1cKGYZhGIZhGIZpIHCki4uL 4VTDtb7e3j3TFPzyyy+43VevXjUajd999901hmEYhmEYhmEaBdxpONVwrcvL y+FmX29Pn3EiuL+lpaVlZWW47zU1NSaGYRiGYRiGYewATjVcazjYcLNZTLkw EMsVFRW419fb4hiGYRiGYRjGdYCDXWHmevv7jFP48ccfjUYjyyiGYRiGYRiG cThws+Fs85wpl6SysrKqqooH9TEMwzAMwzCMw4GbDWcbLvf19voZx0PTo663 iTEMwzAMwzCMa0ITpq631884Hp4hxTAMwzAMwzDOg2ZLXW+vn3E8tOb59bYv hmEYhmEYhnFNaF306+31M46HlRTDMAzDMAzDOA9WUq4KKymGYRiGYRiGcR6s pFwVVlIMwzAMwzAM4zxYSbkqrKQYhmEYhmEYxnmwknJVWEkxDMMwDMMwjPNg JeWqsJJiGIZhGIZhGOfBSspVYSXFMAzDMAzDMM6DlZSrwkqKYRiGYRiGYZwH KylXhZUUwzAMwzAMwzgPVlLNmvLycqPEzz//LHaxkmIYhmEYhmEY56FQUgiR PXNsXg99wOjihx9+eK8++fn5Ym/zVVIXLlw41kCKi4utpVZQUNDQ1HCItdRw ooamhstxTjkxDMMwDMMw1xOFktq6davsmfv5+V0PicDo4t///jfuUWRkZEEd P/30k9jbfJUU1Mf6BnL+/HlrqaWnpzc0NRxiLTWcqKGp4XKcU04MwzA3EUaj sby8/HrngmEYph4KJVVVVSXc8qVLl27atOk6KARGH6SkTp48aXFvc1dSUIjH dRASEqJHSUVEROhJDdH0KCmcVE9quARWUgzDMPaTnZ192223denS5cqVK9c7 LwzDMP+HxjypNWvWsJK6kXFtJQUlclUH+/fv16OkDh48qCc1RNOjpHBSPanh ElhJMQzD2E9CQoKbmdOnT1/vvDAMw/wfrKSaL6ykWEkxDMPcJISGhkZGRl7v XDAMw9SDlVTzhZUUKymGYRiGYRjmesFKqvnCSoqVFMMwjMDPz+/ll1/u1avX sGHDZs6cidZB7Hr77befeuqpr776SoSUlJQ899xzCExJScGml5fXiBEj4uPj w8LCnnjiie7du//+979ftGjRpUuXxCFoVkaOHDlt2rTExMRJkybdfffdEyZM EHszMzPnzZuHRHr06IHwtWvXKtqg7OzsV199dciQIThw8uTJGzduLCsr0x/h 8ccfHzt2rBwfZ/zss88ot2PGjFm2bJm88qrIbUZGxtSpUwcMGNCzZ09E4zqZ YRgHwkqq+cJKipUUwzAMARVDM4kgK9q2bYsfXbp0gYigvQcOHGjRogXCxTyj d999F3H+9re/0eaMGTNoE/8j5p133kmpPfroo5WVlRSnvLycku3YsSN+tGnT BsKEduXk5ED+IPBXv/oVlBQdixZKZA/1YefOnRHYp08faCWcAr+ffvpp/RGw iVyJTdTGdEbQrVs3+vHII4/QB1xEbu+5557+/fvTsRQH2T5z5oxDy55hmJsX VlLNF1ZSrKQYhmGAv78/iYWkpCRsVlRUvPPOOwgZNGiQiDNz5kyEjBo1Cr8T ExNvueUWaK7Lly/TXlJS4KWXXqJ+KNRgPXv2lAURaRNKFicSTUxNTc3999+P 8DfeeIMWKke917VrV4Rs3ryZ4rz11lvY/OijjxDZZK5LSf6g1tUZQVZSiDN4 8GCETJgwgb7Zl5mZOXToUIRMnDhRkdsHH3wwLy8PIcgzfiPkzTffdPw9YBjm poSVVPOFlRQrKYZhGPDCCy9AIOzYsUOEQGvce++9CDx79iyFVFZWDhkyBCFe Xl4PPPAAfoSEhIj4pKTuu+8+qDARGBcXR906tCm0SVpamnz21NRUBCJxEkGE n58fAl9++WXaHDduHDa/+eYbEeHKlSuoUcXpbEaQlRSdsW/fvqIHCmRlZf3q V79q165ddXW1nNvk5GQRZ+XKlQh58skndRYswzCMNjenkvL29p41a9batWub 4Fw///zzxo0bcbp169ZB+zgwZVZSrKQYhmFAv379IBDQavtIDBs2DIH+/v4i 2okTJ6A1SF9MnTpVToGU1Pvvv69ImQbyoUIz1WmTbt26KeJs2LAB4Y8//rh8 9kWLFiHwt7/9LcXZsmULNlu2bPnXv/4VIis3N1eRiM0IspKiM77yyiuKOKQQ cZkit8i/rO+gqqiXykaBMgzD6KPRSuqnn346cOBAVFTUqVOn1Hvhi2JXampq oySCXRiNxqNHj+7cuXP16tUQL9HR0Xl5eYo4Hh4e0CBffvllE+QnMzPzvTqQ MQem7AwltdtB2GOTrKQYhmH0IzpfLGIwGOTIH3zwAQJbtWql+MQtKSkvLy9F 4iTHEhISxIl69+6tiEPjBi3SunVrioMmyd3dvVOnTmLXmDFjaCyizgiykqIz LliwQJGTCRMmIBxCzFpuUYezkmIYxoE0Wkn961//InUwffp01FeKvUuWLMGu jRs3NlYl/B9paWn+/v4pKSl6IqO2//DDD99TsWXLlh9++EFEa0ollZubK7Jh TfU0DlZSrKQYhmEATUpC3RunQu7cKSoqgrIgnfL555/LKZCS+vTTTxUp33XX XQinRKwpKXgLCH/++efVZycJJiguLg4KCoKao1Up2rVrd9Xc26UngqykPD09 3czTshQ5eeyxxxAeHx9vLbespBiGcSz2Kymwfft2xV4HKqnAwEAk5e3trR3t 559/htcqsrRgwYINGzZ4eXnNmjWLQrZt2yYiN6WSAgcOHEDe9u7d+9NPPzkw WVZSrKQYhmHAn/70JwiEPXv2aEebMmUKor311lu33357q1atjh49KnaRknri iSfk+KQ7INNo05qSQs2JcKgY/RmurKx88sknaUSizgiykkpMTMTmsGHD5EOM RiOuq2XLlrR2OisphmGaAIcoqQ8++KC0tFTe60AlBUGkR0klJSVRZj7++OPM zEw5nz4+PvPmzauoqBCBTayknAQrKacqqeKUlNLMTEVq9ispY3Z2kXkMP8Mw jKNYuHChm3lJvcLCQgrJzs7u169f+/btURFRSFBQkJt5TQm0DtSnM2TIELHC uVi7b+XKlRSCaCNGjEDI+PHjKcSakkI4TadCsiJw7ty5OPsLL7xgMnsa/fv3 79atm9BuNTU1Y8eOxSG+vr56IpjqKylkm8YBLlq0SJzxf//3fxEyfPhwjdyy kmIYxrE4REmBzZs3y3u1ldRPP/1UUFBw4sSJs2fPWjv7f8zdTD/88MPq1auR 1Nq1a9WZFzG///77OXPmINqHH35YVFSkTkqx1INCSSE/OTk5kCSXLl2SBwFa zDnipKamIvP5+flyHmSslfYvv/yijvzjjz9iF/5vaGacp6T2SGgrpj0qnK2k Mrdti5sw4fL589pKKj80NPHFF6uKi9VKSpGCNSUV3qdP9MiR+FGel3do4sSC AwdMjlBSBydM2NmmjT3lwzAMo6Cqqoq6pbp06TJhwoQxY8aQ0BBfzkWDQouK HzBXZdApUBzY/OSTTygCKSkaJfib3/zmueeeQ1L43bNnT1pm3GRdSYHIyMjW rVtj70MPPfTKK6/Qyg9yt9f777+PEGirqVOn4lx0dpwOTbbOCG71vycVHR3d pk0bBN57773jxo2jb0v16NEjKytLI7espBiGcSz2KynIJfyPOlCWMNaUFPRC eHi4YirT0qVLL1++LEdLSkqyON1JZvbs2SJ+XFwcBe7cuVNDeghISXl5eeHy N2zYgEpbJAtFBn2nPgSO8fbt2z/++GM5D7hqXI4QQQSkjbU8mywpr5UrV2IX Wjc0bTozQ9ycSirprbcCbrklLylJW0mdnD4d0UrSa9nw5pthv//9xbQ0iynY VFKFCQkBbm7pCxeaWEkxDHOjUlJSMm3atD59+lDXEkTH/PnzobBo76RJkxCI CCL+iRMnoHRuueWWI0eOmOqUlKenJ+K0bNmS5ihBkdFeoqKiws385VyLGQgM DHziiSdobUCk/Nhjj4nuMFBdXb1gwQLxeVwoJog1eQ6XzQgtWrRQLBsYGhr6 1FNPIZ+If8cdd0BPyeMVLeYWCSJw6NChDSlahmEYq9ivpODJo/bDD6gAsdei kkKV7u7uTkdNnz590aJFc+fOpc0PPvhAHpKXkJCgLaPAp59+KuJv27aNAuUh fBqQkkIrQzmn/IiUP/roI2RVjp+YmAjRJCLMnDlT1juKaWKoyRukpCgz3t7e OjMjcLaS0jOKL8KM+ig95/p/VjAajTk5OWVlZf+yBA6sKioSmxAsiIwmGAdC iopEIG//VVaGHzCJQx99BClUevq0xRRk4IogNRyCA6GkDjz+OKVmMhrpB06E CDipxcMV4BIQ2Vh3LJH4wgtQUv9NlmEYxtFkZWWJXiQBWgTUbIpGobKyUgSS klq3bp3JrEFSU1PlD0vpB/VeWloaTmctAvKWkZGh0TzZjKAAV3HmzJlm+tUP hmGaO/YrKQiHI0eO0O+CggLaa1FJoYqmaAEBAaY6TXHu3LnPPvvsPfMaEWIw GzRCZR2ff/459n711VeV9ZH1xZdffvle/V4qbUi8ED4+PvCfEYhWA2ehwL17 98rxL168iMB58+ZB4iHyL2bQWpEShMiCay0i4yoUWQ0JCbGppPRnRtDcldQP 339/ePLkHB8foTLS5sxJmzsX0iMtNDRh4sQrCQknp0+PefzxUzNmlKSmkjzJ 2bw56cUXq0tK8Dtz1apDU6emBQUlTp4c99RTx5csqbxyhZLK37YNiUMBHZw0 KXzQICip+OeeO71kiSKFWvWUnHz8nXdwluNvvZUdEnL+3DlSUmG9eu0fMQI/ vsvNRVKF+/bh9+mVK6PHj4+bMAEp0N+xN98UAi1ny5bESZMO/OEPyZ98Um00 CiWFzBwcNy7u6aczFi8+NH48KymGYW40ZCXFMAzD6MR+JXXw4MGffvqJOptQ CdNetZJKTU212IMDIElol8VPUxkMhvdsrTgBDYU4y5cv14gjI8TLvn375HC4 0B988AHCfX19FYecP39eMYoPpKSkUDqHDh3SOB2kkB4lpT8zRHNfceLfNTUQ OMdff10oqcghQ6IeeADS48RXX2EXFEd4794xv/99wC23BHXoUPPdd7C6kx9+ iF0VFy7g96EJE/A7sE2bqIcf3muWS9F/+cv333+PpFJmzsRm9aVL8X/9664B A/A7bvToNINBkUJhXFxQ+/ahXboc/NvfcK7a/Hh6CiW155FHkM+SI0cQnrVm DQKT587d++ij+0aOPDBq1O6BAxG+u39/UlLJs2djM6Jv38gHHsAP/G8093Cl LFmCzfCePROeew4noutiJcUwzA0FKymGYZhGYL+SSkhIwOaJEydo8+LFi/+x pKR27txJ3Tc1NTXq1KhzZ8+ePepdepQUTV9ChjXiyJB4WbFihXoXSUKLuwS/ /PJLeXn52bNn4+Pj6apDQkI04utRUo3IjFOVFO5FTEzMAd0gstyNpedcNpVU /NixppoaWNqZNWuweTE83KKSOrZ8OY3ui3z00aDbb7+Qm/vDDz+QkrpWWKge 3SencOL998N79SrNyMDvmqqqkM6ddw8bJiup/Pz8oqQkoaTE6L6a6uroRx+F xCuIiSFFhjhJkydThlPnz8dmpo9PdlpacMeOOAUNNUR+Ivr0YSXFMMyNBurt 2bNny7OiGIZhGJs4Skn9/PPPixcvfs+8yJ5QB7KSWrVqFcX/0hK0y+Jaf3qU 1KJFi2h8oEYcGY1V0GmpwGXLlql3lZSUQBPhQj766KP36qPuaJPRo6Qampn/ OFlJRUdHFxcXV+oGkXGIY5VUfnAwaZ/StDRspn72mVpJBbZpI+ZJJc2YgV2n o6IKCgqSzb9tKinCeObMpb178wIDI4cN29m2raykkPiZiAi1kjplTjN94UJK Ic3dHZvnfHyKjh3DX35QEDaPvPtumvnH0Vdf5XlSDMMwDMMwLoajlNR/pKFu ubm5JAFkZUSCSBt5zQrFgdpKir7J+8EHH+j89K2GeMFVq8XLL7/8smfPHpHP Dz/8cMWKFbi68PBwpyopi5kROFVJHThwAPro37pBZBziJCVlPHsWmylz52or qSOfflqrpCIjEZL09tt6lFTV1avRI0ZgE3+hXbsGtm6tVlJpISEKJZWxZUvt cMFnnsFVUJrUO6b4S3j+ebqQ897erKQYhmEYhmFcDAcqKcgN+PzULfXNN99Y 7JOaNWvWJeuUl5erT0RKCglaywnYtWsXZcbiTCs1DRUvoaGhlP4XX3yRnJz8 /fffi13UP8VKqtFK6thrr9mvpPLz8y9evJj4ySdCScW//nqthDl5Ers0lNTh V17B7+x1674rLMRm7NixUFKUWmjPnrKSOrJgAQKx68zBg8GdOu3q2fO7ggLR q0Xp5O/cSX1S9FeYnp5s1lzp8+ezkmIYhmEYhnExHKikQEZGBgXSgt6ykgoM DKRdtDadfkidaa8mcf78eUp88eLFFr9+q6Ch4mXevHkIXLlypXrRCVdVUk0w ug9SIrxnz/3Dh5Os+Hd1NTYbp6SKiorKysqSVErqytmzV65cISUlVv+TU4Ag CrvrLgqvqarae//9UFI4BKkplNSpJUsQeOXy5d0PPhjQsmVhXJy84HnGF18g Tpq7uwi5cugQLuRMUlJAixYxI0fSNcJ+8JuVFMMwDMMwjAvgWCUFVqxYIUbB WVy7D3pE5xg8gtZOnzFjhsmSDFFEI1GjSP/y5ctQWPL3qhokXnCl9DGp0NBQ ReQTJ07QLtdTUk2w4gSkxIm33oLQSP3kkxxf3wOjRkGMOGp03/EPPhCj+06u W1fb+fXPfxYcOKBIIe7ZZ2vnMb3++pk1a2oXCTSPyiu5cMHa6L6T5ulXkcOH n/P1FX9VxcXVJSW77r4bEunwlCl5O3Ycx3W5uWWuW1er6cwD/3Cl+du3J02a xGv3MQzDMAzDuAYOV1JiSfP3VCtI+Pj4iDFy2dnZ8CRJDqSnp0NeYa/FE0VH R9NR0Eoiq3C2Dx8+/PPPP4toJSUltGY4qZLjx48XFRXl5eVFRkZStxH+N9UJ mQaJl19++YXWBvz0009zc3MR8uOPPyJlb29vcaWup6TsRM+5YADlGRmHxo8P bNWqdgHzRx/d3a9f1IMPCiV1ITRUVlK04sQps5apvHSpVkk9/3w9JTV3Limp srKylI8/xu+aq1ehpLLT06MeeaT2FI89pkihLCsLiQS1a1cr4oYOjRk9ulbB RUfXKqnevfcMH47Ec/fvrx0B6OmJwJgnnlDPhypNT69dFiM9PXrkSAhDhIR1 757m7k7fkyrKzz84diyF773vvthRo3a2bctKimEYhmEYprnjcCX1n7oV59RK qqamBhpKqA8IH/oOFPH+++/LH9uVT/TJJ5+8JyGWzsvPz5djnjt3jlZTVzNj xozk5GQRs9HzpBTgdHPmzHFVJaX/+7yNVlL/FRQlJbRIOAElBQECGfIvHVy9 elUoqYsXL+J3cXHx/5OAkkIghHZFbm61lTSrjcbKy5fxo8T8BShacaKgoAC/ r9R96pcQa/dZyw+SqsjLo9/iy7w4EBvX6ifFSophGIZhGKZZ02glBTeeBrap lRS8R2s9NT///PPBgwdp2pEsczZv3lxUVGTtXKWlpStXrlSoGOiL8+fPK2Li cvz8/D799FMRbebMmV5eXvCH5WiUGi3Yrr5qtXhBtnft2iUvfv7ZZ5+FhITA W966dSs2g4ODrWX+P7aUVEMzI3AZJaXAHiVVWFj4448/WlRSelJTKKnLly8r UrOppGRkJWWRBj+uDMMwDMMwzA1Do5WUndTU1EAHZWVllZeX61kjAlRVVSE+ vGX4txY/7ysDZzgzMxN+tTwC0E6+//77CxcuIFkk7qg07eH6KqkIM/YoKWsc O3Zs/fr1x48fv6qD/fv3IzJsCQdGRUXhqhWppaenIwL0u57UEA2RcQgOjImJ gRRSpIYTIQJOqic1XAIi43LsKQ2GYZjmAlpnRyWF9guNvs7IRUVFJSUljjr1 jQ9Kprq62kmJV1ZWakdw4F2WMRqNcAidkXLjaJAFMs0RRz1H10tJMfbDSkqh pNQyymSHkrKYGiupm5a8vDx/f/9Dhw5d74wwzA3Ku+++6+bmNn/+fPuTgjLy 8fFJSEjQEzk/P99gMCxcuBC1rv2nvnG4cuVKYmJiQEDA1q1bY2NjL1y4IHah WfH29i4sLHTg6SCglixZMmzYMNzEN99801o0R93lEydO3H777X/9619pMzs7 +7bbbuvSpQuu2s6UHUKDLLCZkpOTAwODD9m4wzXsU1BWVnbkyJGQkBCoCXiG p06d0vZLUeyw7aCgoM2bN+/bt+/cuXPqOEjh7NmzcMP8/PyCg4OTkpKsaf/S 0tLo6Ojt27cjtaioqIyMDEUERz1HrKSaL85TUntUWFNMN5SSskijlZRFWEnd tOzYsQP+g2j3m5jU1NTPP/88PDz8upyduWlpkOGNHj0az8iLL75o50lReXp4 eOCJ09mEwa0ymIGksvPUNw5ZWVnQNQYJSEU4orS3pqYG3unSpUsd2D0Elxi3 r1WrVn/84x+//fZba9EcdZfhYCMdnIs2oVnczJw+fdrOlO2noRbY7MB1QWK4 u7vDrlavXt2IFLTtk7h06RISN9Rn48aN6tE+BPwrRZoA+YS1yznHfVHE+eqr r9QPwpkzZ9SphYWFyZ1QjnqOWEk1X1hJsZJimozrq6S2bduGs7/xxhvX5ezM TUuDDC8vLw81np0+CVqub8zob7/gDqHShuNkz3lvKNB8LFq0CI6fv78/risz M3PLli3kCor357hqX19fT0/PiooKh5z0hRdewL2G16cdzSF32aRSUiA0NDQy MtLOZO2nERbYvIB1+fj4CHHRCCWlxz6Bl5cXyRw8nhcvXoQvtGzZMoRAvKjT RAQSPrAuuLVnz56F50lpwlREtKCgIIRAA8INy87Ohpe1du1ahHz55ZeyREIO Fy9ejHBvb+9Dhw4hZlRUFClHOTWTg54jVlLNl+a+Cro1WEkxNyCspJibkKY3 vJ07d8LbycrKshbB4fODHJigtaTS0tLCw8PlV+syMTExhw8flkPg7KEQNmzY IA5ByuSXygOMIWoQsnXrVmspN4j77rsP9zo3N9f+pPSgVlI3CDYt0GRp8D9u gR6Py6ax6bFGOy3Ww8MDFwgPH7bUOCWlxz6vXLlCkqe0tFQcCOcKgStWrFBb 7L59+7ALAlYeqgeXTFZJKPaFCxciRO78MhqNuBaFRAoODqYD5dTgudEpFKe2 /zliJdV8YSXFSoppMhRKCs/XyJEjp02blpGRMXXq1AEDBvTs2XPMmDHy7UbL MmLEiPj4+LCwsCeeeKJ79+6///3vFy1adOnSJREnPz8fcf75z3/K5zp9+jQC 6RN4dKKBAwfi7EgB4ThLk1wx05wQxrZs2bJHH320W7duo0ePJmv08fF5+OGH 77zzToRjr+LAkJCQt99+e9CgQX369Bk3btz27dspXMPw6FyxsbELFiwYPnx4 x44dk5KSEP7tt98iPDAwEL8R8tRTTz377LPytJd169YhcNasWdaugmY8qV0d AA/fz8/viy++gCsFFy4uLk54aFVVVXBXvv76a9o8e/bsGivIr51zcnJw7fC1 4Owh56jVFe4xMnPx4kXtYo+OjkaycOEglOCRIilcY2pqqiLaxo0bcV1BQUHW HEgkIgdGRUWtWrVKIa9CQ0MRE//LgSgKBOKStfMJUHp///vf+/btO3jw4Lfe emvPnj1iFzxb3Lg2bdrgXsNU8FtjdJ98l02S4Xl6euLmwswefPDBzz//XFGY 2EQEmFDXrl2HDh26cuXKxMREhZJ6/PHHx44dKx+F60JSqDxhz08++eScOXPU A8NwXS+//HKvXr2GDRs2c+ZM9awf5BDPQo8ePX73u9/NmzcP7bJGKVmzQKhd 3CPcWZQ/Mo841AcKDx+3CU/N0qVLUbfj7qMo1AeiSo+MjPzqq69w4PLly2F4 irU1cF2IABsmoQEn6sKFCzgwICBAxMG5YPawJZwIcXALLN53m3aLw6EWITHg zzROSemxTzywdLFyHJJXixcvVotB6iajmkSAR5s6v2iKE43jVaRpMj+GCEfh ixBkDyGZmZlyNBghSrWgoEB9RfqfI4uwkmq+OENJ3QiwkmJuQBRKCu0gNu+5 557+/fvjB/wHGuQPb0SMMpoxYwZC/va3v+H/Fi1aiDhwaMWLMkRGyCOPPCKf C60JAv/0pz+ZzNPA4eLecccdCGnfvn3v3r0feOCBJrtqprlAxjZq1Cj8T9YC OnXqNH36dPzo0KFD27ZtKRDevjgKqoQC4akiDv2GuDBpGh6dC/4wxUfK8IoR DjcVm8Ixowk1r776Km2i8rz11ltbtWqlGF0jAwcS/gz8NEU4PBxyqOAg0atv sHbtWvLHkFVsIgJFxjP1ZX1onA8Q7iscKhpKhBKAOqO927ZtE0oHnhsFQnBp FDtqeDFKikYuUZqKmfJQcHDzsCs4OFgWU/SSHL5WcXGxxlkIX19fRFasgQDX mpLVPnbu3Ll0s26//XayBNRIQjVjL+7vLbfcgnAoDvyGWLCWlOIukzE8/fTT +B/3V5jZtGnT5KOwSeEws1/96lf4Af2lUFJUkYpNuOLQR3RUly5d6AesTn4T RZkhsU+nRkx5YQE6L3bhQEqkdevWagMTWLPAXbt2ybcYxpadnY3wsLAwCvHw 8BBmtnfvXsWBMEL8D5OjLhWwefNmEQeuGnnychzSArKkpYFtlA16HPC/wvnR Y7eihWq0krKIwj5h55B7QgQR9N4AD5r6cCoi9VsISoRqGHhT9OAr4lAfGR5k 2qQKAcWIgkU54BnZs2dPSkqK0Wi0lnmdz5E1WEk1X1hJsZJimgyLSgo8+OCD eXl5JrP8wW952SvyMcBLL71ErT9MomfPngihr8uZdCgpgkf3MdqQsbVs2dLf 399kng8ulM6cOXOqqqpKSkpmzZqFzcGDB9MhmZmZcJ6h/aOjo01m9wM+FSL0 7dtXJGvR8IRhL1iwQH7Bq/Cx4QmTEIP7hM2xY8fSIRpXQS/t1TOeaI45nFKS IYWFheTT0jtnhZJSgL0kvmJjYykEjSN1K8Bxoq4TZJUSRBNAcdCwkkeqsQAC dBw5tHCWaN4QBBqlrF6jAzWGt7e3LKYaJKPgYdLL/Kv11yekN//wNjWO3bJl C6kJ5Ap5Rk5QzgiBopGbJxrdJ+sUi1hUUoCELa4FJUlKTfSMQA7QW6bQ0FDE gSnOnz+fjrKmpFBEQ4YMQcif//xnqmBReULLI2T27NkUB6ZOh1BHBuTqO++8 g5BBgwZRBNwOWDgiUF8D0sSxiqpVgTULJEEEYMzCIReLRpId4imLj48npSP0 sjjw0KFDcMlQ+AcOHKAQ8exQlwpsCaIDcWCxZBuykqKRePQdVZPZhkNCQkjB yf07euxW4EAlZdE+ccsQ4u7uHhAQAIkKQUGXaXHkJOJgb0REhByIJ50uh7pQ UWIG84hBhSYKDAykaPSKku6Lp6cnjfGTOXr0qMX863mONGAl1XxhJcVKimky rCmp5ORkEQdOFEKefPJJ2iQfA/6JPKYoLi7OzdyZRZuspBiHQMY2ZcoUEeLr 64uQXr16iRB4ti3MkHyAAwaXQx59h5AePXrgKLGgsYaSmjhxoiIPCh/bVOfD DxgwYOPGjfgxfPhw7Ske1E+kHpvk5+eHcFS8IgQPIJx2+tyPtpIiFSb3N5ED CQ9H7h6i19pbt24VRYGqG3W4xtQJml5hqD/Lfu/evQhBhtXxhZiCD9wgGSV6 0OCEK3YVFRVRHjSae1pKAo6lHPjaa68pOijtUVITJkwQEZAT6n+PiYmhkHHj xmFz1apVcjp/+ctfNJQUPHOyXrl8cDvGjBkjXlXRdeH+igi4Wffeey8CSTrh DlJlKz4Lhbzh7msMfrNmgSSIYM9yIFIrLS2Vx+nhRCTJhUqiA+UeKJgWnUWI HcglQ/0ZcACmKCspegTkoYk4O/VbyUtW6rFbgaOUlDX7hN4R/WgCiE2LlQA9 gHSNlHnYNglbIEaTUhfVunXraJwnYiYkJIjE6UkUcpJi4owoW9HrJzfZAj3P kQaspJovrKRYSTFNhkUl1bFjR7nBQhVNvVS0ST7G+++/r0gKRyH8qvndHSsp xiGQscHxECG0rLRiVh0Nl1LM0Th16lRAQABcDg8Pj7vvvls4oiZNJaXwzE2W lBSYNGkSvXNo166dRtVqMvcpkDOj/sbu0aNHZb8oNzdXfu40lBR5WYqFuej1 PgLjJGiWh8aoNjV4TnGIYrE76nGwNkwIJf/NN9/QteiUUagoqJ8LVZDaPUZD T6nJ8/oV9OvXD+Wv6AjYuXMnAsePHy9C7FFSsBw5DoQVAuFF0ya0jGxUBBpH DSVF7wFefvlljZzQdaEYfSToe1jUMwsGDRqEzT59+syfPx932eJnIgUaFkiC SDEHioCjlZmZmZiYGBsbCx1Ng0WFFqMDhagkIKwQSFO6cE9Jhijm7xw+fFhW UpQs0peNlpSUtX4WmzhESVmzTxQm6SBon7CwMBQOTE6s+KdOB8WIcCp/mJNY Pp1G/YlFHc+dO0fF5e7ujlu/dOlSepYpMt1fGgRILzRkEU0vVWAk6rPreY40 YCXVfGElxUqKaTIsKqnevXvLcWAeaiXl5eWlSIraehpPzkqKcQhkbPKca4tK ChYrKylvb28aQ6VAj5KSz0VYVFLwD1u1aoXw6dOna1+C8GPVzgzcoejoaPKa iK+//lpU1NaUFLQD3C34XQp1IAZcWUS4Xjah9+2K9/DUlaCYOC8TERFBJ7Ko jBSgKMhLhLKwmDEEWnP+CaqpWrdurQin1z4DBw4UIfYoKYUx/OMf/xBKSmRA cbHUO29NSc2cOVN7LKgYFWARFAhFgx75wx/+IMK7deu2bNkya98z0rBAshn1 l9nhDNBSeAoUSkpxILn0pKRwLlkFCGjGEykpEccijV463n4lpWGfyJXBvAS6 bJk5OTmkg9LS0tSpVVdXQyqKfihoKDhR9IpD9qMgpiBFxTxHFCaNzROdnnjw KQXF2wMa9YeKQv0o2XyOtGEl1XxhJcVKimkyGq2kPv30U0VSd911l1vdasOk pIYOHSpHiI2NZSXFNIhGKKmjR4+2aNGiXbt2U6ZMgWcSExOD1oTcaQcqqRUr VpAT2717d5sfIaK59taiwdGFA7Z79+7ly5eT21NUVGSyoqSKi4spmnowD80u h3tzXoX2+hIKLM6podFHNLVHDc2IQRHRR3DgJWqIKVwvdWBB8Fr72A0cP5sC ECWvlkh79uxRCBknKSlAg/0UKZNpWVNSXl5ebqplKxR07doVcWAPcSoUa7mf OHFi5cqVzzzzDJnixx9/bC1NaxZoURCR2IHh4Urx7MDDh4CiDhr9SgrQgEAx pJagBcBJSYl+q9TUVLXR6unZtIidSkrbPmktPjHxUEDXLi/KocZoNAq1S5Mc 1T4e/Ft4WeTlpqSkGKTeYYTTQyEPXTZJSvlq/cmGJn3PkQaspJovrKRYSTFN RqOV1BNPPKGOAx+ANtHgupmnfsuvSVH3WlRSU6dOdcKVMa5AI5QUBJSbtPiJ ySxJaAU/hZJSGJ5+JQXHr23btrB2WkJNnk1jEfJCFQvfmcyOitycwduhRfNo 0JRaSVVXV1MEi5/koLWUPT09ZRWD33C0tEd/yQivTB4qSbMtkBOLM0GEjILr K+ZMWRNTuEZaDO3rr79WjMaUgTwxmBd808jqs88+i8JXDDhcsGCBQlM4T0k9 9dRT8iYBw9NQUocPH8bmb3/7W/m+wzGGehUOM2pIxJGXc7dJZGQkiXpr3pE1 C7QoiLZv326oP5ITd4o+/9ogJUWL5MtdS8KAxeg+Eibyd5RM5jFpsMNGe3r2 KCmb9kkZFsu8CKjTVq2kjh49ipJUjFSkHAJqH6GwECckJETRaUhrWYjVKlAg 9EIDClqORsrX4rISep4jDVhJNV9IScFR/6kOeS8rKYKVFOMQGq2kgJh8gaeS 1v6VpyfQHP8NGzbQZm5u7q9//WuFkkLz4WZevMJ5F8g0axqhpOiQ559/Xuz9 6KOPyGKFkrJoeDqVFLzBhx56CCFwFOEC0WQZeeq9GppWr/B/4LPR+tJihr5Q UjSITq2kaG1qeHoWG0FcPr3hF+4ctAx9jxVZpRB4bgjB5ctf9pQ5c+YMyTE5 kN6NI2/q+EJGUT+aqf4CFAoxhWyTw+nl5aWxdLOp7lOnFs8oQOm5mVdvECoJ /mq7du0QKK8x6DwlRYf07NlT6IuoqKiWLVtqKClUldST9cknn4gINOdu8uTJ tElr+A8aNEis+JGdnd2vX7/27dvTygyQXUgE8UXx0ueAUcFauzqLFmiyIogo EJKB0odZkkRqqJKiKVEG8zwseA55eXkiHaGkaP3w5cuXi4tFNBrvKus+m3Yr Y1FJqVNQh+ixT2gog3niobx8R2pqKk2VojXkZVDmBvPsJ+HO4UmhT+6K5fpR zhSCs4tHm06k6NSjuZAeHh7CnlGwNK1MLF4ho+c50kChpH7++WfhlqN4WUnd yJCSkoHFir2spAhWUoxDaLSSoiEov/nNb5577jn6oAk8CrnOf+ONN8h9HTx4 MNRTx44daTSOrKTg2bZo0QKB/fv3f+2115x/uUwzoxFKCh4IGdXQoUNfeukl mB9+d+7cWVZSFg1Pp5Kila6ffvpp2oS6ofStjXwz1a3hoOi8MNV99NNgXgMN jhBNQv/iiy/o1bRCSYlZEnDzNtZHvDyHKKO+gxUrViCcprrIXhzlRN0LICD/ DR6mHEgL9ylWcjaZP8+qkFGEvDS6HE6yy2CeBuJRH8U3c2nWlfrNvwwkLSof N/N3wXA7Ro4cSTPXFJ9Idp6Swg0aPnw4ZeDZZ5/FbxhV3759NZSUyWzAJPcg jlDx0qDoW2+9VXyPDMqFuqVQr06YMAGm3qlTJ7nrMz8/nz4y9cgjjyCfr776 KnW5olStXZ01C7QoiEhNk8CBniIrovGBDVJSJvMHsg31UfRJ4SZSHxCsFLu+ +eYbMmDFqwmbditjUUmpU1CH6LHP4uJiUj0Q8sgkLJx6qeh9hdo7xQWKGVK4 Okhaet2BNOW1MUlwGcwLwm/ZskUcouh9hm3QZ45RLaDcNmzYQG9jcLFXVUP7 TPqeIw0USsrf31/2zFlJ3cj88ssvaDLOSPzwww9iLyspgpUU4xACAgLczJ/Z pc2Kigo385JQcpzc3Fw3adKTWOJs2rRp9AIWjgGae8WXSVHnw48V30UdMWJE VFQUfjzzzDNyNLiF9FYfvqgzL5RpltAMffmdKpw3hIwdO1aOBotFoJjUgBaf HFQ382R8uBywT7f667ypDU99LuKzzz5zM6+lZjIvRAx3HX4sfWqHoGWr5Q5Z BSUlJXCfoG4UfTTwsvbt2yc+oUu+mVAltO6WGJmTlpZmsII8nTw5ORnXS34v eaeyxIMnjNPBi1NMtRDQ52/og6ECeU02GShBtYwiSEwpVk0PDw+3dgmKXjB4 kvAVrWVSUFZW9sYbb9Bid+D+++/HzVLEISltcy6bfJdNVoxh6tSpivGEKNuJ EyeKb0ajIqXuodGjR4s4kFcwQjkd+LeQWqSn2rRp89hjjylmveGGonYlqyYh Bv0uT3WB84+jSDm6mVdER2lrXJ01CyRPW72QCO41DSQzmKU9LopW2xaC1OKB 9P0j+Vpg4dhEiSF7sCJkm14IyMIZpoII1LFCp0NLoRhHatNuZWihBoVFqVNQ h+i0T1jdtm3bqBOKWLp06Z49e6y5pqiXxPJ9BMSXekV6eGhipiQ9+DExMeoh stDveEhJbxrMog+3xuIzaNL9HFlDoaSgImXPHCe9HhKBcQDNXUlt3749WAeo c/QoqS1btuhJDdH0KCl6wWITXAIrqZsZ+W0t2ojU1FRr08ZN5pYUEbRrcjQW UOjWFp5imMYBLzc7O1ujvWiE4SE19bQji4EKaMFzaws7w9EtLCzU860cncDd gnawOK1JMTnLHmgSlrW9KBDtz2xZgz6HSquA6iQ/P1+x2nZTgsvMyMho6ELT uEdQx9pLAUB3KFZskEEJo6wsdkao0bZAi8As4S032lpQIFBeimLZvXu3QRrY JoNmQmOJOfvtVp2CPWnSd+syMzN1Prm43efOnTt79qzcFaUGe5FmTk6O9jhG Sg2em0a104jnSIHGPCmmWdPclVSDsKmkGoRNJdUgWEndtFgbBMUwjAb+/v4e Hh6NXpHsJsFoNK5atWrTpk0O1JUM0cQWSIPlVq5cSVoP0gMageZAHT58uGny cNPikOeIlZSr0nyVVGVlZWkD0XihV1VV1dDUNN594UQNTU3PxE/GJWElxTCN AC1XREQE3Bubc3ZuWgoLC9esWRMcHNy4RZsZbZrYAuEniI9SQU/R/CAa2qex bCNjP456jlhJuSrNV0kxjGuwe/fu2bNnK2ZFMQyjh+PHj+uZNX9zkpycrPHx X8YhNKUFGo3GkJAQmq0DJQUNFRUVxTLZ2TjqOWIl5aqwkmIYhmEYhmku1Ji5 3rlgGgYrKVeFlRTDMAzDMAzDOA9WUq4KKymGYRiGYRiGcR6spFwVVlIMwzAM wzAM4zxYSbkqrKQYhmEYhmEYxnmwknJVWEkxDMMwDMMwjPNgJeWqsJJiGIZh GIZhGOfBSspVYSXFMAzDMAzDMM6DlZSrwkqK0YaefecZCRKvrq7WjoMIiCZ/ PqOoqKikpMRJWWKaHhhY8/2+pNo+7QHlYPOJaO5cvnzZUUk1yHK43rCHm8Ey GcZ5sJJyVVhJMdrExsYaDIZ9+/Y5I/Fz584hcR8fH+1omzZtQrSzZ8/SZn5+ PjYXLlx49epVZ+TKHsrz8nL9/S9HR1vceykyEnurSksddbqa6urcHTtOzZqV NHly8qef5gUGIgOOSlwn4X36hHbrZk8K8G9hAwkJCYrwnJycxMTEkydPqg+5 cOFCeHi4n5/f9u3b4+LiKisr7cmAnSjs006OHTvm7e1dWFjokNRuQN599103 N7f58+fbn5Q1y7HIjVxvNAKj0ZhonYsXL8qRS0tLo6Oj8bBs3rw5KioqIyPD YppZWVmItmXLlh07dsTHx1+6dEne6/KWyTBOhZWUq8JKitHmwIEDcD/Q+Doj cVJS3377rXa0jRs3ItqZM2doEy6rwQxcI2fkyh6gawLc3AJbtiw8eFC9N/KB B7C3ODlZf4Ilqampn38O2aDeVZqREX7PPUhQ/tvZtm32unWNv4CGE9yxY2Dr 1o0+HG6th4cHPDe5IsJveHTu7u64y6tXr1YckpaWZqjPmjVrrqOYUtinndTU 1ISEhCxdutSBHTfOxmykn4dbslI1o0ePhpJ68cUX7TypRcvR4EauNxpBTk6O wToHpfoHlrlkyRJFhLCwMLmDCVYXGRmpiLNo0aJTp07JcZqdZTLMjQMrKVeF lRSjzQ2opNCgp6enO8pxdSykpPC3e8CAKqNRsbcRSipn2zYccuyNNxThlQUF OAV2RfTte/ydd/KCgrLXrz80cSKd/dTHH9t5IfqxR0mh/vnGjFwLwUP28fER 7pxCSRmNxuXLlyPcz88vJSUlISFh1apV2Ny1a5ddl2EHjlVSJrOF+/r6enp6 VlRUOCpNp7Jt2zaIozdUVmqRvLy89evX2+mNW7QcbW7keqMRlJaWxqpAdb1w 4UJY47FjxyganqbFixcjxNvb+9ChQ9nZ2ajM6R3FkSNHRGp4lOhxw17cIESD RKWQ4uJiEa3ZWSbD3DiwknJVWEm5JHpGs6vjoJVUG4NCSTUuZWsRNJTUNTP0 u0GeqsWrsJaywxFKCn9HX3tNsddRSupaVVX0iBEIjxwyRKHXzm3cWNsz1aZN Rf3hPc7DHiW1c+dO3NmsrCw50MPDg7qZ4PiplRQdsnbtWnETz58/Ty4fzEnn efWbqDV02mejnxd4s0hz69atjpp+5VQapKQcgkXLkXH4jB4HJmgtqbS0tPDw cGt3PCYm5vDhw9opQyqiWJYvX46WnUKCg4MR8uWXX8qdtvv370cgdKgICQoK Qojcq4hsQHwhUHHS5mWZDHPjwErKVWEl5UqggY6Li4Nft2jRohUrVgQGBspz NyoqKuCdbt68OTs7e8OGDQsXLly6dCnaWYTn5+cjZJEZHx8f2SMVSio5OdnX 1xcRVq5cGRIScuHCBcXZc3JyEI4m293dHeIIjbVozYny8nKkg+ab2vrdu3dn ZGSolRQCd+zYgbwhh+vXr8clKDzVqqoqXMjXX3+tuC5kAP8vWbIEB2Lv6dOn FTmEm+Hv74+UcRVIGQkiDzgWB9pT7DKkpCLvv39n27b4kR8SIu+1qKTO+/kl vfxyeK9e+4YNOzVzZnHdtCDUudEjR+4eOBCHhHXvDukUP2YM7So8eBCBIXfc YbTkvSMa9qYtWECbZ728cGxBbCxC9g8fDuFzJSmJdpWdO5exYkXcmDGhXbti 17G33qqUJkFQBo5Mm4bA4+++u+fee8P79Dny6qsXIyLk05GSqiopOTl9Oi4h tFu3mMcftzgcUQHNW5HdOQJ3B34yHD+SSAolBcNGoGKix/bt2xEIV9PaubAL NxomERkZ+dVXX5EFwlxhk3I0WEJoaChiUsZgHleuXFEkZdM+TbaeRCI3N9fP z++LL75AOnj6EL+0/gQ6BBocN/2K8PLyGgFDio9ftmzZo48+2q1bt9GjR1P/ BR78hx9++M4770Q49ioORFm9/fbbgwYN6tOnz7hx41DgFI5nfOTIkQMHDoSS 6t69OxIfU2eldK7Y2NgFCxYMHz68Y8eOSWbDw/OOcJQJfiPkqaeeevbZZ+Vy XrduHQJnzZpl7SqsWY5Js1QV9QYKdo0V5A4Xm9UaMnPR1luL6OhoJHvkyBFI FdgzksI1pqamKqKRIUHUqEXKvn376PWCxllwFNWu8sQx6rHNzMyUY+ISUIEX FBSIENScatFEchVNgOJEzrBMhnF5WEm5KqykXAl6rwjgCMGFo4HucEdpr9Fo RMjixYuxF025GDm/adMmuIV0FIVgl3BsSEl5enqKXfQDjqg86gPtMu2CAwM3 huJs27ZNuAT4QY01xaEhKDRMS1ZSUHk08oSySpEpY8JThY9NlyZfF85O5xVX AWSJBF9CpExxsEluuQMH/JCSOjRxYuaqVaSAKqRZ22ollTJvHnVgISaJr9Au XUrNMuFaZSWUC+QSAoPatw/v3RuH//daVq9GYNIrr1jMQ5XRWJaT813dtPqT M2YgctTQoWIi1ZXERIQjY7t69qyd1dWqFXQc7d0zaFBl3bCr6vLy2mGKAwfu /Z//qRVunTohZm20Fi3yg4LE6aCkEBL7xz/WZv7OO0WX3AVbw+2gUywOHBW3 Q62k6NYDhTdLvVdw/Kyda9euXfRmnkyFzA/AJkUcuMTCvIUN4+yywNFjnyZb T6LJ7MZTOBxdKA6KvHbtWrm3AioAgcHBwdrF2CBmzJgByTNq1Cj8f8cdd7iZ 6dSp0/Tp0/GjQ4cObdu2pUBcpjgK10iBUF6IQ79RmCbzHYG2oqTat2/fu3fv B+qslM41dOhQio+UE82GN2/ePGyK20rTpl599VXaRCndeuutrVq1kseeKbBm Odqlqqg3cL++rA/dUCD0tc1qjd4FKaoaNVDcolISFRTSVPSiQsFBYdFNl8UU dSFBRslVrhpIM0P9Dim6ZJwIrTxyCIvas2dPSkqKUTXwGBoKMX19fcWxRUVF VD2qp5U5wzIZxuVhJeWqsJJyGciZRDNNDhvubEhICEI8PDzIkSDFAdCsl5WV IfDgwYMUghaTlmlCo0kOg3BjSEmBDRs2oG01mVcwppf2Ys09nGvlypXUtlJD nJubSw6DmPhMawCiTUdzD5ODzwBHiFIWSgrtPo3sgn9LixUjP5SyTSUF4DtR DuEzkB8VEBBAceAaUX62b99OjjF8JPKrnaSk4AiRuEgYN07sVSipXH9/Uh/U SVRdUXH8nXdIzohDLI7uO/z3vyMww8NDT5ZISVEvVaX0FpoSOTh+fLXZbzRm ZZFiyq67raSkanVchw6X9u6tuXatqqQkefZshATfdltZnRNYq6Tc3CD06LqM Z87gkhGy/3e/084Y9Q1pFL5aSVFnhIfqwsmHXGd9qQ1SUgCPCcwP9iAMm97M w27Jb9yyZQvZHsLJioSd67RPm08ioEkoyBU5zIWFhWSfct8BHiJ6NrWLsUGQ umnZsqW/v7/JvFabUDpz5sypqsIdLpk1axY2Bw8eTIcgS7fcckubNm2izStS ohBwRxChb9++IlmLo/voXGDBggVy94dCSeEySYjRAqFjx46lQzSuwprlaJeq ot5QgL10u1FTUYieau3kyZN099U94ALcdNJ3qDZpdhgqH0pZvUYHLJPG1Akx pVNGITI0o6F+hxQ9L56enjTGT+bo0aPy4WgRaHIiyhYFGBoaKmYjqs/lDMtk GJeHlZSrwkrKZUCTh9bt+PHjIgR3loZ20EtFoTjE2raIQN4jfD9xFLW5EXUj uMjhhLySX2PCE6CkaOw9ebxo6+X3qNBiBvNwetqkASHCSzGZm37Ks1BS2dnZ 1EDLY67ELBibSgrtuziKhBt8EtrEsQbzm2p5sgAN+HeSksLv8tzckM6dsZlV lw2Fkjr0wgvYxCFymey5914EGutGzlhUUpFDhiDw4p49IqTy8uWzXl7ynxhY SEqKsiTzXVFRRX7+Nalz58yaNbXTu15/nTaFkjrn6ysfGPv007WCq+6ukZK6 uHu3iFAQG1urtm6/Xbus6G2/xsgotZKiefHqYV3k2sFKrSVFSkrugYJ/Sxkg H5jOBduTR3bB/5ftXKd92nwSRRw4ySIO0oSrLH8aqaioiFJ2YBVN6mbKlCki xNfXFyG9evUSIbgjLcyQfEBBIdvy6DuE9OjRA0eJIb4aSmqiyvAUSgpAvSJk wIABGzduxI/hw4drT0qyZjnapaqtpEiFyf1Neqo15PPYsWO41xqThkQ9I68f vnfvXms6RYgpCHCdMgqcOnWKjFPurhVCz2B+zxAfH4+qnqpikFx/mHFGRobc oU+qyuKa586wTIZxeVhJuSqspFwG6ktCAx0nQf4bvX4Uo+Dko+g9pDxxm96o i5XQSEnJLihB71TprTsd4unpKZ86NDQUgYhG8amNVjg/iYmJspKiPjIxBUNx adpKCpuy90WutZhWQMNRRBeVwHmj+2gzZ/v22j6dW2+lCU0KJRXRrx82oV+y fXzE375hw2rllbnLwGRFSUWPHFkbR7qcK4mJiuXQY0aNol2kpM54elrMcFVp 6YXw8PN+fohweMqU2kGDL79Mu4SS+s7c0yfIWL4cgSfee482SUnJH8mquXYN l4xADfcSgoWcMY0vpaqVVGZmJlmaIiZpHBpvZhFSUoqJVDTclL5XRTbsX1fs Aup+Rfom3fZp80kE+CH7tzBXdVmhcqY4pY77ABmpG3jIIiQhIQEhYnITAWGF QMUkMvjqeIKQYQ8Pj7vvvhsRxEwZDSWlvllqJQUmTZpEHVjt2rVLT0/XuAQN y9EuVQ0lhUIgu5J1tJ5qTQ9JSUkG8yBqOZDqVWsD5FDyNONJp4xCI079dAfr f3wBKo8SgWQTIh2RSTbKH/IjhUj5jI6OjoyM/Prrr6m48lSfqHOGZTKMy8NK ylVhJeUaoEUzWAfNoqlOcSje25OSIkeRsKikIuovMmAyv8o21L14F0OnLIIW XGRPfuVuqptlIJQUpUO5laFZBtpKSnFdly9flpUUpQznVpEyzUpwnpICECYI iX7sMegLWUkJnWLxL81goMMtKimoGASmzJsnQioLCrLXr6e/42+/jb1xzzxD u0hJqT8yVXb+fNLkyUEdOihOrVBSYT16KA7MDwlBeOzo0bRpce2+sO7ddSop DWdMraSuXr1qME9QUsQ8ceKE2lmVIQOQ+15Ndd0QpKSs2R51c9BMfD32qedJ NJlXP4C/SvMTCTiuCgWBOLRLQ2w2FFI38jBIi0qqd+/espLy9vYeMmSImwo9 Sko95NKikiooKGjVqhXCp0+frn0JGpajXarWlFRWVhbNG1V8iNZmtaadTwFN mlOs27B161YEJtWt/aIGVS6dCFZqc5U8sn9FhxRdGiWiWOSQRv2hKOgqUKQ0 c1AWYtXV1YGBgQZLXcDOsEyGcXlYSbkqrKRcAzS1NOIlNTX1vAp6pWmPklK/ h4cjJBpo6vGBH6s+NU3ERvZoNrc8XcJU9yJUKCmdPUeNUFJ0UeqrUK8VYCdq JfVdUVF4nz614sjdXdEnFdq1K42LK4iLU/yV141UtKikaKnz+LFjLeYhZe7c Wp01Zw5tWlNSB8ePr12J4sEHkbGLERFXkpLOeHqqlVRgy5bV9b8dc3btWoQf mTqVNhunpEzmdQxQ+BrfFVIrKSRIU04UXjRZqcYnpWwqqfj4ePymBeVkaO4M WYge+9TzJArgwaalpe3evZvmpIAiqfsPbmpDnXabNEJJHT16tEWLFu3atZsy ZcrmzZtjYmJQYvfdd59jlRTKkNRZ9+7dbX5qSttyrJWqRSWFO0LRklXfJrBZ renE4qwushl1dw8BPUiWT1OfQkNDNR4ltODUc6owb1PdmwegWIJSqNGr5nVp aPAzqlDFWagWNdSNbhU4wzIZxuVhJeWqsJJyGUgTyZ+kN5mHYaDRpFtsj5JS vO0U4+TJ14JDZTAPg5EbYvzGqcVR9N5eMc0ZTqmspGjsFvwHeZyemKtij5Ki D1chjuzK0rdXnK2kajMTHR3QogXkBg2EE0oq7k9/Ukx3UkNKSsgWoiQtjfqP 8qQ19IhrlZUQR9h1qa4DzqKSMmZn16qk1q3lpQXTIBMsje67LE08AYf/8Q8a lEibjVZSNEBU4yNQFldBJ2lDqx/8N5/V1aTr1Z6kwKaSorf3CtuDXiOnvays zKTbPm0+iSbVR83gkdIDIo8/vHTpkkE1HNdOGqGkIKDw+726wZwm8wNIK/gp lNTU+laqX0lBdbZt27Zr167Tpk3DrgkTJmhfhTXL0S5VtZLCfaQIu6VZfgI9 1ZpNhGaRh0pS/akYkCwQMgqVlZgzpSGmaEwjqmi1rhEzYU+cOCGH02AAsWQE KSlYmiI/1HXl7u6u+A6vMyyTYVweVlKuCispl4E+OLJ8+XIxRzgvL48GupDL YY+SolfxYjFhcj/Wrl1LcdDc03t4MXwOjT59i2Tjxo0UQjOskR8xVUpMahBK Cjmkriu4DWSWSJneytqppGDn5GzDr4ODVFJSAv9ZjAJytpICp2bOFMPnhJJK X7jwvwuP190yCJyIfv2C2rcvqluvID84uDbOffdZTDD4ttsKpaW6qozG2NGj a9cD7Nat2uz8m6woqariYlrSvLjO4S9JTaVF0dVKKuLXvxY5zA8NhSrEX5H5 I0QmO5QUDZxT+HgyFpUUmQ3sjey8do1E8+oiHh4e6rWdBTaVFHxFsuGwsDCy PdgMDWEVdq7TPm0+ifB4KR2x4Jvw+eXhXqT0Ea5Rhg2lEUqKDnn++efF3o8+ +kgxui84OBib99W3Up1KCrXKQw89hBDUFSjhe+65B7/VEzNlLFqOzVJVKync a4N56W+LjaCeag0SGyG4fEWvjYBEimKyGC2cYvHOChkleiflBSjUDxRKj9Yg tfYagepzPB1i7OLVq1dp1p7ogRWrr0BRilPgidi0aZPBvGqrIk1nWCbDuDys pFwVVlIuA1pDcvzc3d3Rxn3zzTf04Rvhk9jZJ0XuK/wiGl6F3/LyEfBe6HSI CTeDFotGiPiGDixNzKGG80lvlcm9lL8nJRabwi6ci3oEKJo9Sspkfo8qT6Ag X5fkVRMoKVSgNLRPVlLXqqqoWyq0S5eDEybEjxkT0qlT7crk0jt549mztcrF zW13//5HX3tNhMPhOfT885RgeK9eSZMnx4waBQmGzV133VUiffTT2ug+WqQd Zzw4fjxOHXTrrfQ1KIWSCmrXLqhDB/whq/seeogyI6ZxmexQUjQTP0jVrSaw qKRMdStFGMxT5mlcE4B3qnEum0rKZO5yEvYGWyU7h13JQ1L12KfNJxHQwgUU CIeWnGGcSx61SDNl5OUu7acRSgoZaNGiBX0Z6qWXXho8eDB+d+7cWVZS+EFx +vfv/1qdlepUUvPnz8fm008/TZtQN5S+tZFvJuuWo12qinpDzCHy8vLaWB/R f2SzWqOcGFRdkAKS+YovndFrJfXk05iYGIWMIuSl0S0WBapTawPtEC7WjoBZ QhaR3sRZrtZ9ck7k02Bery8gIACPBtX5iKz4CrbJOZbJMC4PKylXhZWUK4EG F00tvW8kFyIqKkoM2CgrKzOovsVDLp88ToaaZtHKUwsLFxStPLWt8B7RHMsf viGSk5ORGvmW5EMqfCE03GigxUK7W7duJWdGXkLKZF7QT2guiJ1jx47RIEDh ttHKUWJsicXrKiwsJK9ADoS8wuXAI4J2Cw8PLy4upst3pJJCXt3cEidNUu+C gNrZpk2tkpJ8/qqSkiPTptFEqgDzt6VS58+/Vt8pOuvlFX7PPbWqp3NnOby6 ouLYm2+SeqK/kDvugCYqre/5UO9VtuoF8ndFRQnjxkEE0bEJf/7zhbCwAOmD v6Skwnv3vrR37+4BAyja7oEDU+bOldOBFsN1KRIP69HDppIqKSnBTYRRWYtG CzCqF3/DgfQGgICLuL/++EM15PspJvjThHp5gkxqairMgzQU9BFMVF5an7Bp nyZbT6LJrLb27dsnPvYKfxXOv8J/hvUiJ4oZLnYyc+ZM6BS5NwGPNkLG1p9z 16dPHwSKMV3+/v533XUX9UN169YNTw2UF37L6xhAj1B3Uuc6K1Wfi/jss88Q Tm85oExbtWrVtm1buT554YUXEGH8+PHWrsKa5WiXqqLeSEtLM1hBXkhBu1q7 ePEiTofKx9ptItugTxIL5EUjZaAE1TKKIDGlXjWd3gUp0lcACYlsiC9KUwWu PguuVHwcjYAEU3+W1+Qcy2QYl4eVlKvCSsolQRvnpFWVoD4UY+YVoNWGYNH4 HAzsDdmzNhhGAH2Ett7molU6QTqXzMgJlpaWks+s8UmjJsOYlVVR93UeNcj2 d1evigF7MtBiVxIT84ODodQaUVzXKitL0tOrLC2zLJQUbZbn5RmlvkuHQEP1 FLPndAIzg/sNZ77MUrHYQ1VVFVS/dmHqtE+bTyL2QvWr06FvDctfWb3uQEFk Z2drtBe4CpRbg24HfaRbMe3IYqACbcuxVqqNRqNaU0zOsgeahGVtLwpE+zNb 2uASzp07d/78ee2CRa2IZwoxrVXRN6BlMkyzgJWUq8JKirkZgFdAEwGio6PJ 4OHy0TeIxbRrRoFCSTkJf39/Dw8Pm1/MuakwGo2rVq3atGmTA7WA68GW0/Sw ZTJMo2El5aqwkmJuEmgqNw0GowkUNNDlWN3KCYyCplFSqH8iIiLgnim+5nPT UlhYuGbNGsh8XmJaG7acJoYtk2HsgZWUq8JKirl5SElJ8fHx8fDwcHd3X716 9fbt2zWW4GZqqquTZ88+vWxZE5zr+PHj1ubs32wkJydrfLCVUcCW02SwZTKM PbCSclVYSTE3IWzzDMMwDMM0GaykXBVWUgzDMAzDMAzjPFhJuSqspBiGYRiG YRjGebCSclVYSTEMwzAMwzCM82Al5aqwkmIYhmEYhmEY58FKylVhJcUwDMMw DMMwzoOVlKvCSophGIZhGIZhnAcrKVeFlRTDMAzDMAzDOA9WUq4KKymmKYG9 VVdXX+9cMDccqIWqqqqudy4aCUwahl1TU+OQ1FAOLv+MXL582VFJNchyioqK SkpKHHXqm42bwTIZxnmwknJVWEkxTca5c+cMBoOPj8/1zogTKc/Ly/X3vxwd bXHvpchI7K0qLXXU6Wqqq3N37Dg1a1bS5MnJn36aFxiIDDgqcZ2E9+kT2q2b PSnAv4VVJCQkKMJzcnISExNPnjypCEeVlZ6eHhkZuWnTpl27diHC9VVhyAYM ++zZsw5J7dixY97e3oWFhQ5J7Qbk3XffdXNzmz9/vv1JWbMci+Tn5+M2LVy4 8OrVq/af+rpjNBoTrXPx4kU5cmlpaXR09Pbt2zdv3hwVFZWRkWExzaysLETb smXLjh074uPjL126JO91ectkGKfCSspVYSXFNBmkpL799tvrnREnAl0T4OYW 2LJl4cGD6r2RDzyAvcXJyfoTLElNTf388wvh4epdpRkZ4ffcgwTlv51t22av W9f4C2g4wR07BrZu3ejD4dZ6eHjAc5MrIvyGR+fu7g6DWb16tRwfogkOoaE+ vr6+5eXljb8G+9i4cSPycObMGYekVlNTExISsnTpUgd23Dgbs5F+Hm7JStWM Hj0aSurFF1+086QWLUcDSF2yFkgqO099I5CTk2OwzkGp/oFlLlmyRBEhLCxM 7mCC1UVGRiriLFq06NSpU3KcZmeZDHPjwErKVWElxTQZN4+Swt/uAQOqjEbF 3kYoqZxt23DIsTfeUIRXFhTgFNgV0bfv8XfeyQsKyl6//tDEiXT2Ux9/bOeF 6MceJYX65xszci0ED9nHx0e4cwolBQ8QgcuWLTty5AicyaSkJPISd+7caddl 2IFjlZTJ7LJCG3p6elZUVDgqTaeybds2iKM3VFZqkby8vPXr19vpjVu0HG1Q qunp6Q68TdeX0tLSWBUHDhxYuHAhrPHYsWMUDU/T4sWLEeLt7X3o0KHs7Oyo qCh6R4EnSKSWkpJCjxv24gYhGiQqhRQXF4tozc4yGebGgZWUq8JKyiXRM5od baLNW28zHT0nEnG0lZSepGCuNuNcX4SSwt/R115T7HWUkrpWVRU9YgTCI4cM Uei1cxs31vZMtWlTUX94j/OwR0lB/sAksrKy5EAPDw8ErlmzBo6fWkmtWLFC ccipU6cQgqP0z1Sy37CvmaHfGkqqQQ+IDLxZpLl161ZHTb9yKg1SUg7BouXI OHxGjwMTtJZUWlpaeHi4tTseExNz+PBh7ZQhFVEsy5cvF1VlcHAwQr788svK ykoRbf/+/QiEDhUhQUFBCJF7FZENiC8EKk7avCyTYW4cWEm5KqykXAk00HFx cfDrFi1aBIczMDBQnrtRUVEB73Tz5s05OTn4f8mSJQsXLvz6669Pnz6tSCc3 NzcsLAyR0WJ6enqiLS6tP7UHKYSGhlIENMe7d+++cuWKIpHy8vKoqCjspZYd cTIyMtRKCkmFhISgoXd3d8cuNPGyXILzgLOkpqbidCtXrnTsm39nQEoq8v77 d7Ztix/5ISHyXotK6ryfX9LLL4f36rVv2LBTM2cW100LQp0bPXLk7oEDcUhY 9+6QTvFjxtCuwoMHERhyxx1GS6WBaNibtmABbZ718sKxBbGxCNk/fDiEz5Wk JNpVdu5cxooVcWPGhHbtil3H3nqrUpoEQRk4Mm0aAo+/++6ee+8N79PnyKuv XoyIkE9HSqqqpOTk9Om4hNBu3WIef9zicEQFNG9FducIWC/8ZDh+58+fVygp o9G4atWqr776SvbiSkpK6OW5xmICZEgw9cjISBxONgnDU4wJ1GPYMOMdO3Ys XboUj8/69evxiKmVlPaTSOAp8/Pz++KLL5DOhg0bEF/xlCHQgdOvCC8vrxEw pPj4ZcuWPfroo926dRs9ejT1X/j4+Dz88MN33nknwrFXcSDK6u233x40aFCf Pn3GjRu3fft2CsfTOnLkyIEDB0JJde/eHYmPqbNSOldsbOyCBQuGDx/esWPH JLPh4TFHOMoEvxHy1FNPPfvss3I5r1u3DoGzZs2ydhXWLMekWapVVVW4s6jx aBMFu8YKcoeLdgVFmblo661FdHQ0kj1y5AikCuwZSeEaUa0popEhQdSoRcq+ ffvo9YLGWXAU1bfyxDE8LwjJzMyUY+ISLly4UFBQIELQIqhFE8nVAwcOKE7k DMtkGJeHlZSrwkrKlaD3ijT2CS4cDXSHO0p74YUiBAIKbgbFEQOo4C2IRNDC wksU6dAPNMdiojE8BzHqnpIid1f2A9GmU9NsME/xpgEn8F0VSgrnoqQQQSS1 bds24Ujs2rVLzgYuJzs72+nlaAekpA5NnJi5ahUpoApp1rZaSaXMm0cdWIhJ 4iu0S5dS83zwa5WVUC6QSwgMat8+vHdvHE5HZa5ejcCkV16xmIcqo7EsJ+e7 umn1J2fMQOSooUPFRKoriYkIR8Z29exZO6urVSvoONq7Z9CgyrphV9Xl5bXD FAcO3Ps//1Mr3Dp1QszaaC1a5AcFidNBSSEk9o9/rM38nXeKLrkLu3ZplxV0 Co0mUoQLSaJWUhbJyspCNBitxktyMiT4w/QIkEECWKmIo8ewYX40MgrQoCkk RfYpKyntJ9FkduMpHE8WFAdFXrt2rdxbARWAwODgYO3LbxAzZsyA5Bk1ahT+ v+OOO9zMdOrUafr06fjRoUOHtm3bUiAuUxyFa6RAKC/Eod8oTOyC4IW2oqTa t2/fu3fvB+qslM41dOhQio+UE82GN2/ePGyK20rTpl599VXaRCndeuutrVq1 kseeKbBmOdqliqzSjaDIuF9f1oduKBD62mYFRW+HFFWoGihuUdmK2gxpnjt3 To4GBQeFRTddNmbqQoKMkgfaqYE0M9TvkKJLxonQyiOHsKg9e/akpKQYVQOP oaEM5vmG4tiioiLq/1VPK3OGZTKMy8NKylVhJeUy0FAoNNPksOHOhoSEGMyj nsiRICUF4GOglTSZX7eSvxEQEECJ4ChqPf38/KjVRjNK7/BDQ0PlCFu2bKEu gIKCAkpEXpQvNjaWWnA07jAweAhwe+jsQkkhKepmQotMzXdubi65GWK6NDnA YN++ferW/wZEKCk4QiQuEsaNE3sVSirX35/UB3USVVdUHH/nHZIz4hCLo/sO //3vCMzw8NCTJVJS1EtVKb2FpkQOjh9fbfYbjVlZpJiy6+4jKalaHdehw6W9 e2uuXasqKUmePRshwbfdVlbnBNYqKTc3CD26LuOZM7hkhOz/3e+0M0Z2pdHJ qEdJwbZpDFJYWJhGNGFIeExgkHCVDxw4QCH0Zl6PYcMvpZGH0F8UB08H2bB8 ITafRECTUJArcpgLCwvJ8uW+AzwOCEGutIuxQZC6admypb+/v8ksQoXSmTNn TlUV7nDJrFmzsDl48GA6BFm65ZZb2rRpE21ekRKFgDuCCH379hXJWhzdR+cC CxYskLs/FEoKl0lCDM84NseOHUuHaFyFNcvRLlWFklKAvXS7UXdRiJ4K6uTJ k3T31T37Atx00neQQjQ7DAKNUlav0QHLJHsWYkqnjEJkaEZD/Q4p6ryjcQWG +hw9elQ+vKysjCYnomxRgKjt6cUXGgL1uZxhmQzj8rCSclVYSbkMaPLQuh0/ flyE4M7S0A56qSiUFNpBEYckD9pu2iTfFW0oGlYRB/G3bt0aYh6oRhHQhsoD YOAmUcpiKD4N/xA+icnc0FMOhZKipOAhyG9fjxw5YjAPwqdNcoDh3DqmjJyP UFL4XZ6bG9K5Mzaz6opXoaQOvfACNnGIOBxFsefeexForBs5Y1FJRQ4ZgsCL e/aIkMrLl896ecl/YmAhKSnKksx3RUUV+fnXpKFKZ9asqZ3e9frrtCmU1Dlf X/nA2KefrhVcdfeRlNTF3btFhILY2Fq1dfvt2mVFb/s1RkbpUVLU+6OwWDVk SHIPFPxbygD5wHoMOzs7m+LIYwLpQNmxt/kkijhwkkUcpAlXWV7OvaioiFJ2 YBVN6mbKlCkixNfXFyG9evUSIbgjLcyQfEBBIdvy6DuE9OjRA0dBEVCIhpKa qDI8hZICeMARMmDAgI0bN+LH8OHDtSclWbMc7VLVVlKkwuT+Jj0VFPJ57Ngx 3GuN/lCaWATk9cP37t1rTacIMYUqV6eMMtXNFoRxyoMPhdAD69ati4+Ph8yn yhkk1x9mnJGRIQ9UIFVlcc1zZ1gmw7g8rKRcFVZSLgMNPkEDHSdB/hu9fiQl BUdC9lLo7aIYfk+v08UkCDUUgV5oy9DUEjH0jlpkhauTmJgoKylKytPTU85w aGgoAleuXElxyAGGA2BX0TQhspICOdu31/bp3HorTWhSKKmIfv2wCf2S7eMj /vYNG1Yrr+pK2KKSih45sjZOXU8iuJKYqFgOPWbUKNpFSuqMp6fFDFeVll4I Dz/v54cIh6dMqR00+PLLtEsoqe/MPZiCjOXLEXjivfdok5SU/JGsmmvXcMkI 1HAvIVjIGdOY3GRTSdHkkcWLF2uPrTLVGVJMTIwcSANQ6XtVegz74MGDFp8O evSEkrL5JAL8kP1bPIbqskLlTHFKHfcBMlI38JBFSEJCAkLE5CYCwgqBiklk 8NUDAgKQYQ8Pj7vvvhsRxEwZDSXlqTI8tZICkyZNog6sdu3apaena1yChuVo l6qGkkIhUF0k62g9FZQekpKScMimTZvkQOoStTZADiVPM550yig04tRPd7D+ xxeg8igRSDYh0hGZZKM8ioAUIuUzOjo6MjLy66+/puLKU32izhmWyTAuDysp V4WVlGuAFs1gHTSLpjolBTdPPvDy5cuykiKHE06gtRNRBEpQht4G04RlkRnF 91IVK06IAVcWoWMpDlwaO8unyVAoKQBhgpDoxx6DvpCVlNApFv/SDAY63KKS gopBYMq8eSKksqAge/16+jv+9tvYG/fMM7SLlJT6I1Nl588nTZ4c1KGD4tQK JRXWo4fiwPyQEITHjh5NmxbX7gvr3l2nktJwxrSVFOkad3d3jYFVAouGRP4k KSk9hm0tDs2CISWl50k0mVc/gL8qJiQCOK4KBYE4tEtDbDYUUjfrJGOwqKR6 9+4tKylvb+8hQ4a4qdCjpNapDM+ikiooKGjVqhXCp0+frn0JGpajXarWlFRW VhasaMmSJYoP0eqpoPRA3aaKdRu2bt2KwKS6tV/URERE0IlgpTZXyTtx4oRB 1SFlqptCaFAtckij/lAUdBUoUpo5KAux6urqwMBAg6WVPZxhmQzj8rCSclVY SbkGaGppxEtqaup5FfRKU4+SonezO6TxZgri4+MRgdbdkqEpBuRMIjM0d1ue HGGqe+0plBRNW960aZM6w6KLwQWU1HdFReF9+tSKI3d3RZ9UaNeuNC6uIC5O 8VdeNwLTopKipc7jx461mIeUuXNrddacObRpTUkdHD++diWKBx9Exi5GRFxJ Sjrj6alWUoEtW1bX/3bM2bVrEX5k6lTabJySMpnXMcDN1fiukIaSItcR4IfG KQQ2lZQewyaLDZC6AgmaYCWM3+aTKIAHm5aWtnv3bpqTAoqk7j+xJqF+p90m jVBSR48ebdGiRbt27aZMmbJ58+aYmBiU2H333edYJYUyJHXWvXt3m5+a0rYc a6VqUUnhjlC0ZNW3CfRUUHqwOKuLbEbd3UNAD5Ll09Sn0NBQjUcJLTj1nKrr yatXr1IJKJagFGr0qnldGuSNmgbFWah1MEjDtglnWCbDuDyspFwVVlIuA80X lj9JbzIPw0CjSbdYj5KiTz6tXLlSfreJ1nb//v00kZlecqKJl4cIlpaWkm8j 5qrQW3rFpGa4oLKSghtmMA+ekZtv/EaGxdldQEmBy9HRAS1aQG7QQDihpOL+ 9CfFdCc1pKSEbCFK0tKo/yhPWkOPuFZZCXGEXZfqOhYtKiljdnatSmrdWl5a MA0ywdLovsvSxBNw+B//oEGJtNloJUWT7hXLl8lYU1Lp6em0gJ7CwDSwqaT0 GHZmZqY6jphLJVxlm0+iqf7nqEzml/z0yMjjDy9dumQwrzSo8xr10AglBQGF 3+/VDeY0mSUJreCnUFJT61upfiUF1dm2bduuXbtOmzYNuyZMmKB9FdYsR7tU 1UoK95Ei7JZm+Qn0VFA2EZpFHipJ84wUA60FQkZB5Yk5UxpiisY0QpqpdQ1K gySb4oUDDQ8QS0aQkoKlKfJDXVd41hTf4XWGZTKMy8NKylVhJeUy0JyR5cuX iznCeXl5NNCFXA49SgrOBk1xCgwMpFYVFuLr62uomx6CJpVeuYeFhZHliAhr 164VydJ8apxdTJUSUxiEkoKTQEmJwYRwFegLJhs3bqQQ11BS4NTMmWL4nFBS 6QsX/nfh8bpbBoET0a9fUPv2RXXrFeQHB9fGue8+iwkG33ZbobRUV5XRGDt6 dO16gN26VdepWotKqqq4mJY0L65z+EtSU2lRdLWSivj1r0UO80NDoQrxV2T+ CJHJDiVFA+c0OpUsKqns7GxaCY2W1NaJTSWlx7DxBFFnK9xaigMbpl4DWUnZ fBLh8VI6Ylyi8Pnl4V70lVWE679MmzRCSdEhzz//vNj70UcfKUb3BQcHY/O+ +laqU0mhnnnooYcQgqceJXzPPffgt7w2iBqLlmOzVNVKCvfaYF7622IjqKeC gsRGCC5f0WsjIJGimCyWkpJi7c4KGSV6J+UFKNQPFEqP1va3VknSbC8PDw8x dvHq1as0a0/0wIrVV6AoxSnwRGzatAmBGzZsUKTpDMtkGJeHlZSrwkrKZUBr SI6fu7s72rhvvvmG3tsLn0SPkjKZX87Te3i0rfCCyGuFiyK+hpOZmUkR4Bzi LBQBycpj+WBXYsY0XE16h0zOpPw9Kfg8lMkVK1bAOaElphEizuUySgoVKA3t k5XUtaoq6pYK7dLl4IQJ8WPGhHTqVLsyufRO3nj2bK1ycXPb3b//0ddeE+Fw eA49/zwlGN6rV9LkyTGjRkGCYXPXXXeVSB/9tDa6jxZpxxkPjh+PUwfdeit9 DUqhpILatQvq0AF/yOq+hx6izIhpXCY7lBTNxA9SdasJ1EoK3p2YBeOhIi0t zVpSNpWUSZ9hi8XQEAdPh4gvKymbTyKghQsoEA4tOcM4lzz3h2bKyAtg2k8j lBQy0KJFC/oy1EsvvTR48GD87ty5s6yk8IPi9O/f/7U6K9WppObPn4/Np59+ mjahbih9ayPfTNYtR7tUFUpKzCHy8vLaWB/Rf2SzgqKcGFRdkAJaHBVqSw6k F00R9b9wbTJ/P1ohowh5aXSLRYEK1tpAO4SLtSNglpBFpDdxlqt1n5wT+TSY 1+sLCAjAo0GdWYicYf68nYwzLJNhXB5WUq4KKylXAg0umlp630guRFRUlBiw UVZWRv6nfEhhYSG1nnJgenr6pk2bxIdH4Q0qPs6YmpoKv4JcTbiRW7dulVdW J9BMozkWy+oiDrku8oJRIDk5Ge07eaTkecoeFDXZGvOybzRy4YW4uSVOmqTe BQG1s02bWiWVkiICq0pKjkybRhOpAszflkqdP/9afaforJdX+D331Kqezp3l 8OqKimNvvknqif5C7rgDmqi0vudDvVfZqhfI3xUVJYwbBxFExyb8+c8XwsIC pA/+kpIK79370t69uwcMoGi7Bw5MmTtXTgdaDNelSDysRw+bSqqkpARqHQ6b tWi0sKT8Pl+s5G8Ra96syYoh0YR6eYKMHsNOTEwUbwmWL19+7NgxGrYqZIXJ 1pNoMqutffv2iY+94lmD86/wn/FUIieKGS52MnPmTOgUuTcB6hIhY+vPuevT pw8CxZguf3//u+66i/qhunXrhgcWygu/5XUMoEeoO6lznZWqz0V89tlnCKe3 N1CmrVq1atu2rfwhrRdeeAERxo8fb+0qrFmOdqnSinNiTBp0tzVDkhdS0K6g Ll68iNOhUrV2m8g2FP2n8qKRMlCCahlFkJhSr5qO+Or0FUBCIhvii9K4Fugp 9VlwpeLjaAQkmPqzvCbnWCbDuDyspFwVVlIuCdo4+1dVqqqqunz5ssaHXRAB cknbW4Z1ITPWhr4IEEH7XC6PMSurou7rPGpQyN9dvVpt6ZNJ0GJXEhPzg4Oh 1Gyu8aXmWmVlSXp6laVlloWSos3yvDxj3Sr3joLGfOqf7tQ06DHssrIy+KI2 C9zmk4i9hYWF6nSg6Qz1v7J63YGCyM7O1mgvcBUoN+0Peymgz3Yrph1ZDFSg bTnWSrXRaFRQislZ9kCTsKztRYHYU0PiEs6dO3f+/Hntgi0tLYVARkxrlfYN aJkM0yxgJeWqsJJiGMYiCiXlJPz9/T08PGx+Meemwmg0rlq1atOmTQ7UAq4H W07Tw5bJMI2GlZSrwkqKYRiLNI2SQv0TEREB90zxNZ+blsLCwjVr1gQHB/MS 09qw5TQxbJkMYw+spFwVVlIMw1ikpro6efbs08uWNcG5jh8//v/bu/OgKqv/ D+CQjEA1Q5tpmZRji9k0jZlOYWNNjW1ONZUtZuOUTTpNNWbZtBqbGDfRIPKH BoK4oCaLgAKxiJEhKKuYGJIhEsiSxR3vXzXN7z33M98zj+cuXOTe0If36w/n 3vOce56FB+958zznPG5GOY0otbW1F9HAwGHHM+c/wzOTaCiYpMyKSYqIiIiI yHeYpMyKSYqIiIiIyHeYpMyKSYqIiIiIyHeYpMyKSYqIiIiIyHeYpMyKSYqI iIiIyHeYpMyKSYqIiIiIyHeYpMyKSYqIiIiIyHeYpMyKSYqIiIiIyHeYpMyK SYqIiP5L7e3t3moK31/9/f0eVu7q6uru7vbWqkcaHGer1TrcW0F0sWKSMism KSJfQK9j27ZtH3zwwfz58xcvXmyxWFpbW4d7o4iG31tvveXn57d8+fKhN4Vk lJycvG/fPk8qnzhxIiIiIioqqrOzc+irHna9vb0/utbW1mas3NPTU1xcvHXr 1vT09MLCwqamJqdtNjc3o9qmTZvwf1d5efnJkyeNS6uqqpKSkjo6Ony4V0Tm xSRlVkxSRF7X2Ng4YcIEv3ONGjUqPDz8PFqrr6///PPPc3Nzvb6dIjExEe3/ 8ccfPmqfTG9Qp+js2bPx6/DSSy8NcaUIRBaLBX1+D7/Cjh49GmGHSDXEVV8I jh8/HuHaDz/8oGr+/PPPMTExWoWcnBzjBaazZ88WFBRodaKjo2tqaox1srKy Vq5c6cVLikQjB5OUWTFJEXlXW1vbxIkT0VcMDQ1dvHjxli1b0Pd48MEHL7nk EhQmJSUNtkG0gA8uWrTIF1sLt99+O9rX/v5M5LlBnaK//vrr+vXrh9gbxzfX /9l5/v2FINDY2IhYMZT1Xjh6enrKHJSWlkZFRSEEVVVVSTXkzRUrVqAE//NU VFQcO3assLAwMjISJZWVlaq1uro6SU9Yih8QqiGiSsnp06dVNRzDlJSUhISE M2fO/Nc7THSRY5IyKyYpIi/6888/77nnHvQq77jjDnR1jItWrVqF8oCAgIKC gkG1ySRFFzhfn6KOvvvuO3Tym5ubXVXw+ogeLzboqqmGhobc3FykFadLS0pK fvrpJ/ctIyrisHz55Zf4ZpeSzMxMlKxZswb/Nalq33//PQqRQ1XJzp07UWK8 qojNQPhCobZS5CwUbt682dV2EpFTTFJmxSRF5EXl5eXoUl555ZVO//T98ssv Y+ncuXPl7YkTJ8LCwt544w1jncOHD6Nw6dKlNvsf3mfOnHnLLbfgU2PHjkX5 448/LtUSExPxFqvLycmZNWsWlt5///3R0dHGQDRg+9JIYGAg2kcAxOvt27d7 83DQBUadNl988cW999577bXXzp49W65fJCcn4xy45pprUI6l2gezsrLefPPN 2267LTQ09Mknn9y6dauUD3iKlpWVhYeHz5gxIyQkZP/+/Sj/9ttvUb5jxw68 RslDDz302GOP/f7772pd69atQ+GyZctc7YWMeDIGAaW1tXXjxo2xsbFRUVEb NmzYu3ev+oNGf39/fHz82rVr5e3Ro0fjXTBecDl+/Dj2HUkkMjISW44MokKK 2hhtUJKj4uJiNFtZWYmo8tVXX6Ep7GN9fb1WLTU1FfuFUOMYUoqKirAIjbhZ Cz6FY4JqxoFjq1evRsmRI0eMNbELv/3226lTp1RJenq6Y2iSuFpaWqqtCAcW 5TiA7veaiIyYpMyKSYrIi9DjQpdy/vz5Tpei34ilkyZNkrdIW3g7ffp0xzqP PPKIzX6FCx1X5DKUXHrppRMmTLjrrruk2nvvvYfCZ555Bv/6+/ujAyyjsdAN Vn98HrD9FStWoE257XDcuHF4je60F48GXWjktHnggQck78s5c8UVVyxZsgQv LrvssqCgIClEb199CqlECpG8UEde41S3eXCKTp06Veqj5R9//BHln376Kd4i UEg1GTb16quvytuWlpbLL788ICDAeO+ZJj8/X+5D08rRt4+OjsYixAfkOLk5 7ZtvvpFrQNhUGfsjlfHbseZcchccqDGDiBsywghHAOlMlm7ZskUlnaamJilE 4HJz2NevX68GHyGlygu0+csvvxirIcEhYWFRZmamMUzJJSTEKOONdo4QzbQL UrLLWBG+5bGFyJW7d++uq6vr7e3VPosMhZopKSnqs11dXatWrXI6rAztyEa6 2Rgi0jBJmRWTFJEXvfLKK+gWWiwWp0vRUxo1ahQqoJdi8yDpCKe3Tkk3FebN myfXoaqrq8ePH4+St99+W+p42D7v7hs55LTBSZiRkWGzz9Wmks7HH3/c39/f 3d29bNkyvJ0yZYp85MiRI8jagYGBxcXFNnvnHCEIFW666SbVrPtTNDw83Hj5 Q0tSra2tEsSKiorwds6cOfIRN3vx9ddfoyfveNlXhvbs2rVLYkhHR4fEFrki oyUpDZZK+CorK5MSfDnGxcVJZJB8gU2VBtV8DocOHZJYdPjwYVdbixwn+Q5R SEaHIaBJy45zdCDEyT11Kkx5GKNQGZlRuyAlF+8SEhLkHj+jAwcOGD/e19eX nJyMchxbHMDs7GwkMrzduHGj47pwHLAIOcvN9hCRhknKrJikiLzozjvvRD9w 9+7dripMnjwZFdA7snkjSSEEGe9E2rt3LwpvvPFGecskRRo5bRYsWKBKUlJS UHLDDTeokra2Nn87iQ8IAuiQG+++Q8m4cePwKSQCKXFzij7//PPaNmhJCjZt 2oSSm2++OTU1FS9mzJjhflCSXCdyvKcO3X6Uyy+XQDBBAJEHTrlPUpLCjNeb WlpaJMIYLw9VVlbKKCF1KKqqqqqrq90MGpKBRWCcP3zPnj2ucooKU1lZWR7G KKipqZF0Y7z5UAU9WLduXXl5eUVFhdybB7W1tcYWmpqa1PUygVTldM7zrq4u qcDOA5HnmKTMikmKyItmzpyJrqCb0UbXX389KsgfhIeepN555x2t/ZCQEJTL E3OYpEgjpw16yKpk3759KFGDmwSCFQq1ifHRV8eJjQ65xWKR01iNlHFziiYk JGjb4Jik4IUXXpALWMHBwY2NjW524cyZM9KNd3zGLn6tjKmhtbXVGHDcJCkc BLl2Y/y7BEKHFO41yM7ORmFcXJybLdTgNw4fSUtLMxaWlpa6uUEOR15GPHkY o/AlLtfpjJOf2+xPgJJGENnU84tRWWKj8VZeSYiyncXFxQUFBWvXrpXDhSSo rQ7dBqmsTapDRG4wSZkVkxSRF8lTRz/55BOnS0+dOoWlgYGB0qsZepJKTEzU VnH33XejXO7wYZIijZw2CBqqxGmSkqehqSSVlJQkF1s1niQp47qE0ySFX42A gACUL1myxP0uqCTl2I3HrxVSwMqVK9VVFcQBlctcJanm5ubIyMiYmBjtV2DX rl0RrqlgMiCZFk+bt2Hz5s0olCk4nMrLy5MVIfUMOEvewYMHHS9Iya5JI9ok h3LXHw6F7AUOqcydbgxiVqt1x44dEc5m9sCnXIVZInKFScqsmKSIvEjuldL6 pUpubi6WTps2Td5K0pk6daqxTllZmedJ6qOPPtJWcd1116G8tbXV8/aZpEaO 80hSBw4c8Pf3Dw4OXrBgQXp6eklJyaFDh+Sc8WKSkmcE+NknABzwUVPS7XdV DbmgoaEhPz9fRvqADEt0mqROnz4t1bRb3Wz/m1chLS2txYH7+SU0Tkd1yWQO jpd7BPIgluIQydCn7OxsN2EK3+Dx8fGoVlFRoS3q7OyUI2C8OdNmSKPq4jVe x8bGamvBEZZqxhnUAQFqsHGSiJikzIpJisiL6urqpEMoI/qN0J279dZbsUhN S97W1oa3o0eP7uvrU9XQKXKapF577TVja9JNnTVrlrEQfTwUjhkzZlDtS69Y wheZ23kkKQQoP8M0JjZ7JJEZ/LQk5fQU9SRJ1dfXBwUF4bxduHAhFj377LPu 90Kma9AmvrPZOyrGrzP082XSPKQ/m7MkZbVapQJil+NasHdyd58xX+A1Yoh2 6ccNlVmMt0rKOCNsidPhYCpGIeWpMVNuwpTc04ho5phrcDQksh08eNBYLlMO qikjJEnFxMRo2yOXriIjI7Xn8J48eVLqe3gQiMjGJGVeTFJE3vXuu+/62SeU lv6baG9vR2dVpoNQ4/RBRu5v2LBB3iLOTJw4UUs6mZmZMrmEcS1qYjQ1ZAO/ y2FhYSh5+umnB9W+zN6GzrDXDgFdqM4jSclHnnvuObV06dKl2t19bk7RAZMU eu/Tpk1DSWpqam9vL35B8Do9Pd3NXsjMElo6QI6QaczVNHoqSclNdI5JKicn J8I+9bfTL0HsvkxtsWfPHilBlpFHLGFTpaSvrw8l2H3tqo0iIUUbLFZXV4dC bJtjfRWj5Dqa7dwJKBzDFI7emjVrnF6QEjLay2KxqIvOnZ2d8pApeaSXNCJ7 ikSpVoH0lJaWhkL1v4ciz/91uv1E5AqTlFkxSRF5F7oi8pgnP/tDml588cXp 06f7+/v72R/co91EtGjRIqk5ZcoUpJuQkJCxY8dqSQf9Vfn4pEmTXn/9dSmU buqYMWPw7+TJk5944omrr74ar8ePH29Map60LxcdRo8e/eijjzr94zyZxnkk qbKyMjn9kLjnzZuHEwmvr7rqKmOScnOKDpikli9fjrcPP/ywvEW6kfZd3flm +98cDjt37tTKZToIQBBDTJCIERsbKyOqtCSlxhAlJiamnktdP0Ioi4yMlMs3 KEcekWs0LS0txi2Bmpoap5uKo4elSFvGQpm4Ly8vT6tcUlKixShhnBrd6aGI i4tzdaMdytXcEciMiEWSN7EWubXPuJ0R9vn6tm/fvm3bNrmYhcpNTU1amzKG S00XT0SeYJIyKyYpIq9DzwcRRj3DVNx3333l5eVaTfRz0DtVNcPCwgoLC/EC ocZYDZ09+Vs9ephSoiZGW7hwoTyjKjg4GP1h7XmmnrSPbuFTTz0ljaxevdo3 h4QuCO+//z5+ysarCRUVFSiZM2eOsVpoaCgK1T1dGRkZMv7Oz/5wXnTI5QKr cR4Dx1PUcV3is88+Q3l8fLzNPk13QEBAUFCQPPJJzJ07V7u0qunu7o6JiUFX X7tGY7Vai4qK1CN0kQIQqVQqkRnn1D1pDQ0NES4YJ1Kora3F/srILGQo7I4x 4rW1tWF1SFjaQCQFqQQflEcSK9gqFGLftcpIgo4xSkiYcpw1HfUd29cgQmIz JBJG2B/UizzluBbsqdw2qSCCOT6W12Yf+YVc5mqXicgpJimzYpIi8hH0x/bv 35+ZmYnw4jimwwg9wPr6evc9E3QaOzs71Ygn4x/80ePFx7WxDINtHx/v6OgY cJYwGrGQII4dO+bm+0I7RT2B1hyHHTkt1MjgIO3xsgp+9bx7MiOMtLe3Ox3W pA3OGgoZhOVqKQ6I+8dsuYddwP9CLS0t7g9sT08PAjJqurpfEf+TRJz7/F8i 8gSTlFkxSRFdjFzdOkU0QmRkZFgslgGftURe1Nvbu3r16rS0NP7JhWiwmKTM ikmK6GLEJEUjHL658vLy0LHnBP7/jY6Ojvj4+MzMTE5+TnQemKTMikmK6GKU n5//4YcfaqOiiEaa6upqV7M9kHfV1ta6eZQwEbnHJGVWTFJERERERL7DJGVW TFJERERERL7DJGVWTFJERERERL7DJGVWTFJERERERL7DJGVWTFJERERERL7D JGVWTFJERERERL7DJGVWTFJERERERL7DJGVWTFJERERERL7DJGVWTFJERERE RL7DJGVWTFJERERERL7DJGVWTFJERERERL7DJGVWTFJERERERL7DJGVWTFJE RERERL7DJGVWVquVSYqIiIiIyEfQ3+7v7x/uXj9539mzZ/GTxb/DfYoRERER EZmN6mwPd6+fvO/vv//mDX5ERERERL4gt/ahyz3cvX7yCUnKDFNERERERF6E DjYvSJnbv//+Kz9l/Mvb/IiIiIiIhgidaulgW61WdLaHu79PPoSfL37if/31 l0xAQURERERE50emmEDXGnmKMWqE+Oeff5Cn5OdORERERETnAUkKnWp0rYe7 d09ERERERERERERERERERERERERERERERERENJz+H37SApQ= "], {{0, 89.}, {567., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{567., 89.}, PlotRange->{{0, 567.}, {0, 89.}}]], "Output", CellChangeTimes->{ 3.7638130624319277`*^9, 3.763813920680509*^9, 3.763814449954289*^9, { 3.763814988874877*^9, 3.763815107405212*^9}, {3.763815457582891*^9, 3.763815460028717*^9}, 3.763815812674694*^9, 3.763914842123302*^9, 3.7639163377856817`*^9, 3.764005691879901*^9, 3.764081108889886*^9, 3.7640822866079817`*^9, 3.7645950961520243`*^9, 3.768659786402215*^9, 3.7978486724294977`*^9, 3.7979224745491533`*^9, 3.7981787648593483`*^9, 3.7981861857670383`*^9}, CellLabel->"Out[14]="] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell["Basic usage", "Subsection", CellChangeTimes->{{3.728626116968351*^9, 3.728626120224564*^9}, { 3.728626607450799*^9, 3.728626618722734*^9}}], Cell["\<\ Given a piece of text, the BERT net produces a sequence of feature vectors of \ size 768, which corresponds to the sequence of input words or subwords:\ \>", "Text", CellChangeTimes->{{3.731749644794014*^9, 3.731749663143916*^9}, { 3.731756512052701*^9, 3.731756553969533*^9}, {3.731822329478395*^9, 3.731822396947947*^9}, {3.763318344558814*^9, 3.7633183539015007`*^9}, { 3.7633185584007397`*^9, 3.763318559496709*^9}, {3.76338285182791*^9, 3.7633828638348713`*^9}, {3.764585278110647*^9, 3.7645852788868113`*^9}, { 3.764587403242468*^9, 3.7645874622522373`*^9}, {3.764587501349124*^9, 3.764587558206744*^9}, {3.764588055015943*^9, 3.764588061615952*^9}, { 3.773419593728712*^9, 3.773419599384046*^9}, 3.799769771235303*^9}], Cell[BoxData[{ RowBox[{ RowBox[{"input", "=", "\"\\""}], ";"}], "\[IndentingNewLine]", RowBox[{ RowBox[{"embeddings", "=", RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], "[", "input", "]"}]}], ";"}]}], "Input", CellChangeTimes->{{3.731749671350584*^9, 3.731749676132331*^9}, { 3.731822587033185*^9, 3.731822589286665*^9}, {3.732555591432569*^9, 3.732555592903824*^9}, {3.751382248606879*^9, 3.7513822514277153`*^9}, { 3.751634162110808*^9, 3.7516341935245333`*^9}, 3.763318370778699*^9, 3.7633829301109962`*^9, 3.763404216491029*^9, 3.763417037605917*^9, { 3.764082354064518*^9, 3.7640823638406763`*^9}, {3.764588085254237*^9, 3.764588086216671*^9}, {3.764588852502605*^9, 3.764588852813987*^9}, 3.797848642296974*^9}, CellLabel->"In[15]:="], Cell["Obtain dimensions of the embeddings:", "Text", CellChangeTimes->{{3.763737577074662*^9, 3.763737584066257*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"Dimensions", "@", "embeddings"}]], "Input", CellChangeTimes->{{3.763382957642108*^9, 3.763382966931374*^9}}, CellLabel->"In[17]:="], Cell[BoxData[ RowBox[{"{", RowBox[{"8", ",", "768"}], "}"}]], "Output", CellChangeTimes->{3.763382967639161*^9, 3.763404155237269*^9, 3.763417105269256*^9, 3.7637439981686897`*^9, 3.763744187709096*^9, 3.763744897876059*^9, 3.7638099232009478`*^9, 3.763815804971939*^9, 3.763914841856917*^9, 3.764005675836177*^9, 3.7640811016706038`*^9, 3.764588315908642*^9, 3.76458885739703*^9, 3.797848678022746*^9, 3.797922478120282*^9, 3.79817876946235*^9, 3.798186191581502*^9}, CellLabel->"Out[17]="] }, Open ]], Cell["Visualize the embeddings:", "Text", CellChangeTimes->{{3.731749644794014*^9, 3.731749663143916*^9}, { 3.731756512052701*^9, 3.731756553969533*^9}, {3.731822329478395*^9, 3.731822396947947*^9}, {3.763318344558814*^9, 3.7633183539015007`*^9}, { 3.7633184553917913`*^9, 3.763318461791045*^9}, 3.763382867930786*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"MatrixPlot", "@", "embeddings"}]], "Input", CellChangeTimes->{{3.763318523002062*^9, 3.763318551673849*^9}, { 3.763320123430731*^9, 3.763320123741251*^9}, {3.763382939835025*^9, 3.7633829419509077`*^9}}, CellLabel->"In[18]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzt3Xlwleee4PfOUqn5M/8kqSRVqcpUKpOkJpVMVSrpyXRPpnume7pneqb3 5fbt7d7pBsQOwhiDzGa40GAwNptBYLAxYAwGY+Aag8FgNmMkxCZACBASAgkJ 7QIhYaE847eu6lgCCc4jxOHo861fuXQePdar47r3faXz0Tnnf/zpuD8Y9p/+ 0i/9Uu4/CP/4g5+8+msTJ/7ktT/8L8ONPx6bmzN87N/97W+PnfR3w/9u4i// 9D8Li7/8n/zSL/1m2PofP+6SJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmS MrJ9+/YtXbq0tbX1RX8jkqSBr7Oz85tvvtm0adPy5cs//PDDw4cPP3z4sMee 4uLijRs3rlixYseOHZcvX+79RcI1Ys+ePatXr163bl34Ci0tLYPyvUuSnm9H jhzJz8+/cuVKj/V+rwv9bpAkvURVVVXt3bs3/LS/du3affv2tbW19djguiBp SHX69Om9Tyj8/Jy6s9/zZ2FhYfKAzObNm8OXHcQ7kc2dOnVq2PfV1dW96O9F kjTA3bt3b/bs2cN+WF5eXuo5f/fu3T027NmzJ/WLtLS05Obmpm6YNm1abW3t oN8bSdJAVlZWNmLEiHBW7/F7Wb/XhX43SJJeok6cODFq1KjUs/rkyZPr6+u7 N7guSBpqLVmyZNgTmjJlSve2vs+fnZ2dy5cv7/Gvh5Ww/oLuVjb06NGjQ4cO df9nJxqSlH2tWbMmnOHHjx//ySeflJSU7Nq1a9KkSWFl0aJFyYbKysqcnJzh w4d/8cUXFRUVYUNyUTh//nz3F8nPzw8rc+fOLS0tvXTp0syZM8PNn/3sZy/o PkmSBqD29va8vLzknJ8qGv1eF57mwiFJelmqqqoKp/RwGl+7dm34fWHfvn2v vPJKuLls2bJkg+uCpCFY+PF4yw/bsGFDcnJbs2ZNsqff8+dXX30VboY927dv DxvCP8PZMqwcOHDgxd2zl7uampqFCxemChHRkKQsq6OjY+TIkeEMv3///u7F 48ePJ6f95MUGk2vB5s2buzeEK3VYeeedd5Kbly9fDjdHjBjR1NSUrIQPkqtw WVnZ4N0ZSdKA1v1LWQ/R6Pe60O8GSdJL1Pr168M5/M033+z+m+HCwsKwEn6P CL9NdLkuSNL3ffTRR+HMNmvWrAcPHiQr/Z4/ly5dGm6uWLGi+4usWrUqrCxZ smTwv//sKHlKYPgvfPDgweSBKaIhSVlWVVXV9OnTJ0yY0I0RoXDxTR6/Ki8v b29vT/6i4MaNG90bKioqkr8iuH//fri5ffv2cDNciFO/cvLEya1btw7afZEk DWBnzpxJnjK/YMGCVNHo97rwNBcOSdLLUjirJ48I1dTUdC92dnYWFRWVlpY+ evTIdUGSQiUlJcmfet68eTNZ6ff8GW6+8847YcP69eu7N2zbti2shJ/AB/n7 z5q++uqrDRs2JK+CnvwFL9GQpKHQhQsXkt8vwvW3rKws+Ti52naXXJevX7/e 9YtL8O7du1M37NmzJyy+/fbbg/qtS5IGoqampokTJ4bT+OXLl5O/HOsWjX6v C09z4ZAkvSxVV1eHs/fUqVM7OjrOnj27bdu2rVu3FhQUdP+9seuCJD148CCc J3v8VWe/58/QgQMHwp5x48ZdvXq16/szavIy4N5paEDyHA1JGiLdu3dv+vTp 3c+5KC4uDh/n5ub22Ja8D3jyyrfz588PHx86dCh1Q7jprTQk6SUtUYyPP/64 ++Nu0ej3uvA0Fw5J0stS8gKzc+bM6fGy5AsWLGhubu5yXZCkrq79+/eHE9qY MWNSn3fW7/kz1NbWtm7dumR93LhxyQdhf/cGxUQ0JGkoFC6myauLjB07tr6+ PqycPXs23Hzttdd67EzUo7CwMHwcLtDh4+PHj6du+Pbbb8PizJkzB+2blyQN SAlJhxN46qv7dotGv9eFp7lwSJJelk6cONH9KNzatWtPnjy5Y8eO5HU88vPz u1wXJKmra8aMGeGEtnHjxtTFfs+focbGxuSH7dS2bNmS/ByuyIiGJGV94Uqa 2ES4whYXFyeLyStQTZ06tcfm8CtJWA+/noSP582bFz4OF+vUDcnbi4cvODjf vCRpQKqqqho9enT44b+ioiJZ6SEa/V4XnubCIUl6WQrn/+QRtm3btnUvJi+T kjxM5LogaYh39erV5JR469at1PV+z5/h5ltvvRU+njZtWnFxcUtLS/hnXl5e WFm8ePFg341sjGhIUnZXXV09ZcqU5HmOJSUl3euVlZXJYo/9ydMhk50rV64M H3/55ZepG5InXS5atGgQvnlJ0kCVvJDg7Nmzj/+imTNnhpX33nsvfFxfX9/v deFpLhySpJel8+fPJw++pb4ESlNTU/Jm3+GzrguShnjJy0YtXLiwx3q/5897 9+6NGDEi3Aznye4N1dXVyePwYecg3YHsjWhIUhZ38+bN5GVsp0yZ0uOPClpa WpJLcHt7e/figwcPksWamppwc9OmTT3+6iAUbibPrBycuyBJGpCSy8GTKiws 7Pe68DQXDknSy1LiEaG2trbU9eSviF0XJA3xwplt9OjR4WxWUFDQ41P9nj+L iorCB5MmTerxL86ePXtYr9fBUBoRDUnK1sJFdvz48cO+f4WoxsbG3huSR7e+ /fbb7pWTJ0+GlQkTJiQ3k1dcf/3111P/reQaffDgwef6zUuSBradO3du/mGv vvpq8ldn4eNEvfu9LvS7QZL0stTW1pa86nvqW3iHxeQRvIaGhi7XBUlDuBs3 boSz2fDhw3u/80W/58/S0tLwQU5OTir4dnZ2Tp48OawXFRUN0n3I3oiGJGVl jx49Sl5gZMaMGS0tLd/9sGTP7t27w4Y33njj3r174Wb4Z/IHA7t27Uo2hCty 8pzx7jcHP3bsWLg5ceLE8DVfyP2SJA1UPd5Ho+sprgv9bpAkvUQlz8h+7bXX khdOCb9BbNy4cVjKW2O4LkgasiVv/52Xl/fYz/Z9/mxraxszZky4uW7duvCp 5F/Zs2dPwhxedSo+oiFJWVlhYWEfry5y9erVru9feCoBiylTpuTn5ydvtzFz 5sxu8gh9+umnyZ8lvP19yctCnjx58sXdM0nSwNRbNPq9LjzNhUOS9LJUV1c3 ceLEcCYfP378u+++O3fu3PDxyJEju98Cw3VB0pBt+/bt4YS2cuXKx3623/Pn 119/nTz8kpeX9957782bNy+5uXfv3kG8E1lb6tMJJUlZ065du/oQjWvXriXb qqur33jjje712bNnp75xVdf3f2mwe/fuUaNGJRvCBXrHjh0v4g5Jkga45cuX hxP7sWPHUhf7vS70u0GS9BJVX1+/aNGi5M+WkrP6mTNnUje4Lkgamq1evTqc 03bu3PmkDf2eP0+dOpW8zFRSbm5ujx+8JUlS2rW0tJSXl9+9e7f76ZA96uzs vH37dvjdpMf7XkmSsrJ+rwv9bpAkvUS1t7dXVFT08UIorguS9Nj6PX82NDSE 02N9ff1gfleSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmS JEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmS JEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmSJEmZ1jBJkiRJ kiRJkqRn6UWJxgs5biYcXZKUObkiSJJSc12QJKXmuiBJvRuaj+0PzXstScq0 XBEkSam5LkiSUnNdkKTeDc3H9ofmvZYkZVquCJKk1FwXJEmpuS5IUu+G5mP7 Q/NeS5IyLVcESVJqrguSpNRcFySpd0Pzsf2hea8lSZmWK4IkKTXXBUlSaq4L ktS7ofnY/tC815KkTMsVQZKUmuuCJCk11wVJ6t3QfGx/aN5rSVKm5YogSUrN dUGSlJrrgiT1bmg+tj8077UkKdNyRZAkpea6IElKzXVBkno3NB/bd0WQJEmS JEmSJElPH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQ JEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZ H9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmS JEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmS JEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQ JEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZ H9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmS JEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmSJEmSJEmZH9GQJEmS JEmSJEmZ3xAUjdqz89OelrqzMVPT2h4zn1+qjpnjN+pi5l+/WZz2NNzYETN1 l1fHzJ2Tk2Lm6+u1MVPZ2Jb23D7wlzHTVH00Zv5szZmYqb+6KWYaK79Ie2rP zIuZfzazJGaaHzyMmfrSDTHTfPd02tPSeCVm6krWxMzd84tipvbsvJi5+vZ/ kfZUNrXFTE3h9Jj56eaCmPnicnXMNFz/OO2pL30/ZjafroiZv/2oIGZ+Y0Fx zLQ2XU17yhvux0zjzc9j5td+dilmWuqLYybmgthSfyFmpu05HzO3v/zzmLn/ sDNmKnf9VtrTXHMyZiZ/di5myurvxczfH7gcM0fL7qY9S49cjZnKz/51zFR9 9ZOYiTw9/nhdUdrTdOtAzNTd74iZC1VNMVN3aVXMXK+7l/a0tlTEzM/2X4qZ mJ/Y/+MP7ecWxsx7J8vSnj9fVxQzpysbYibyat7SUBwzrU2laU9NwdSYudPy IGbKN/2fMXOz8X7M1La2pz/nFsRM3b2OmPmwoDxmGto6YuaTs5VpT2tzWczc a70VM78+72LMVB3+u5hZ/+2NtKf5bmHMNJRtj5naojkxE/nDQMxvOtvP3YqZ u8VLY6b+6saY+fUfV6Y9TdVHYuZFPbbfNdCyUF9fv2zZsoMHDw7ycZ8pokE0 iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaGSlaOzbt2/p0qWt ra19b2tvb58zZ074guvXrx+Q4z6niAbRIBpEg2gQDaJBNIgG0SAaRINoEA2i QTSIBtEgGkSDaBANopF9onHq1Klh31dXV9fHttra2jfffDPZSTSIBtEgGkSD aBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iMbD1IQuPHj06dOjQqFGj +hWNAwcOjB49etgvIhpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBAN okE0iAbRGNieJAs1NTULFy4cllIfojF8+PCwYf78+Rs2bCAaRINoEA2iQTSI BtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0RjwniQLu3fvDp8aOXLkwYMH c3Jy+haN+fPnnzx5srOz8/PPPycaRINoEA2iQTSIBtEgGkSDaBANokE0iAbR IBpEg2gQDaJBNIgG0RjwniQLX3311YYNG2pra8PHI0eO7Pd9NJJ+/vOfEw2i QTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoDHhPIwv9Pkej O6JBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQjecR0SAa RINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0ejjuI/tqb7p uIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaKRsaLxJEHw HA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoZI5oPCmi QTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iEXPc5xTR IBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQ DaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iMaAN7CisXfv3rDzgw8+ GJDjPqeIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2ikZWi MXr06LCtoaFhkI/7nCIaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQ DaJBNIhGVopGNh23i2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0 iAbRIBpEI+OP20U0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSD aBANopHxx+0iGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSI BtHI+ON2EQ2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINo ZPxxQ5U7fyPt+Z23z8dM0+1DMTNq2+mY+b1l52LmSm1L2nP3wpKYufPNxJip +np4zEzYcSZmmu4cT3uqmh/ETN2VdTHz24svxExT1aGYKd/8T9Ked49fi5k/ yT8TM/O+vBQzdRdXxMzd84vTnqYHD2PmzreToyZOHuuvvB8zt/b8u7Sn+uio mLn1+e9Fzc//fczUnpkbM63t36U9kZTz80tVMRN5dv0Xcy/FzKmb9WnPiRt1 MdPY9jBmfmXqlZi5+em/jJlbTW1pz9azlTHT0nA5Zn6ysTBm5u6/FDP1V9an PX+65kzM/MX60zHzztelMVNf+kHM3L24LO2J/BOmxltfxkzk33W0NFyMmd9c WJz2rDp+LWaOXr8bM5G/rdQUzoiZhvKdac/J8vqYef3nF2Kmpf27mPnrDYUx s/dyddrzo7VFMbOxsDxm3jp0JWZuff77MVNa25r2xPyuEWZfyZ2YuX3gL2Km 6fbBmDlQWpP2LDp0JWYirwsxhhXm7rk3Y+bra7Vpz81PfjVmdhXfjpk/zT8T M813vomZ5Uevpj1/tOpszNy9uDxmIv8wo3L3v4mZn1+sSnta6s/HTOQDGo2V +2Lm/xtZlvaU5f93MfOiHtvvGiBZ6Ozs/OabbzZt2rR8+fIPP/zw8OHDDx8+ HITjphfRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0skk0 ampqPvvsszVr1qxaterTTz+tqKh47LZ79+7Nnj172A/Ly8urq6tL77jPO6JB NIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2hkjWgcPnw4Jycn FSmGDx++Z8+e3jvXrFkTPjt+/PhPPvmkpKRk165dkyZNCiuLFi1K47iDENEg GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSyQzRu3bqVcMai RYtOnjx5+vTpZcuWJa5RUlKSurOjo2PkyJFhff/+/d2Lx48fTza3trY+03EH J6JBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2hkh2h89tln YX3ixInfffddstLZ2Tl16tSw+NFHH6XurKqqmj59+oQJE5qamroXHzx4kIhG eXn5Mx13cCIaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhG dojGli1bJk+eHP6Zurh169awedmyZf1+zQsXLiSvUtXe3v5Mxx2ciAbRIBpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopEdovHY5s+fHzZv3769 72337t2bPn162Ll06dIBOe6ARzSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJB NIgG0SAaRINoEI1sFY1Dhw4lT7u4fv16H9va2toWLFgQdo4dO7a+vj7+uM8j okE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaGSlaBQWFg4f Pjzs7PE6VD1qbGycM2dO2DZy5Mji4uJ+j/vYnu27TyuiQTSIBtEgGkSDaBAN okE0iAbRIBpEg2gQDaJBNIgG0SAaRINoZKxoPEkQ+kWEQ4cOJZzxzjvvPHz4 8Enbqqurp0yZEraNGzeupKTkab6fZ74PAxTRIBpEg2gQDaJBNIgG0SAaRINo EA2iQTSIBtEgGkSDaBANokE0MlY0nlTfsrB9+/ZEPVavXt3R0fGkbTdv3szN zQ3bpkyZcuvWrfjjPteIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAa RINoEA2ikTWi8ejRo40bNyacsW3btnDzSV+hsrJy/PjxYducOXMaGxsjjzsI EQ2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRCNrROODDz5I 3gr88OHDffzrjx49mj9/ftg5Y8aMlpaW735YGscdhIgG0SAaRINoEA2iQTSI BtEgGkSDaBANokE0iAbRIBpEg2gQDaKRHaJx4cKF5NkZR48e7YEUnZ2dqTsL Cwv7eIeOq1evPtNxByeiQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG 0SAaRINoZIdozJo160lIsWTJktSdu3bt6kM0rl279kzHHZyIBtEgGkSDaBAN okE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2ikQWi0dnZOWLEiKcUjQE87qBF NIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQjSwQjew+bhfR IBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iEbGH7eLaBAN okE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkQj44/bRTSIBtEg GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2ikfHH7SIaRINoEA2i QTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIjGUBWNzs7OY8eOffDBBytX rvz888/LysoG57jpdXXJf572/OX7p2Mm8lrQ2lIeM7k7z8bM8I8L057Gmz+P mcgfOOvud8TMby26EDOTdp5Ne1rqzsVMw41PY2bk1tMxs/jQlZip3Pmbac+t vX8YM1N2nYuZV+OmubYwZqqP5KQ9dSVrYyZSHgtuNsRMU9XhmLm549fTnqqD fx0zv7f8XMxcX/Vfx8zZ240x8/cHLqc99de2xExT9dGYiXyktLX1ZszEPML8 O0vOx8xHpyti5lcnXY2Z90/diJlbX/xx2lN3OT9m/vDdszHzx6vOxMyxsrsx 8015Xdrz75eej5l/s/hCzET+9Li/5E7MVGz9p2lP+Yf/OGZWH78eMzd3/IuY mbPvUszcbm5LexrKd8XMH6w4GzORv2fVnJ4ZN7PSnvqrG2PmX867GDNrvymL mbvFy2Im5m+Qfrb/Usz86L2imDl8rTZmIv9C4FdmXU57WlsrY+Zm4/2Y+fT8 rZgZve10zFR99TdpT/XxsTFz58S4mPnyyp2YuVjdHDOtTdfSnrK1/0PMXM// b2Nm9hcXY6a5tiBm6i6tSnv+/L2imIn8/bRszX8fM0uPlMZM/bWP0p5//WZx zNxY9w9jprF8V8wsOVya9tw5+UrMvKjH9rv6lIXq6urt27evXLly48aNJ0+e vH///mO3tbW1zZ07d9gPO3DgQNrHfd4RDaJBNIgG0SAaRINoEA2iQTSIBtEg GkSDaBANokE0iAbRIBpEI2tE48SJEzk5OalIMWvWrMbGxt47N2zYED47YcKE vXv3lpSUbNq0KdlfWlqaxnEHIaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBAN okE0iAbRIBpEg2hkh2jU1NSMHDkyfGrp0qXnz58/fvz49OnTw83Fixf32NnR 0TFmzJjwqb1793YvJk/Z+OSTT571uIMT0SAaRINoEA2iQTSIBtEgGkSDaBAN okE0iAbRIBpEg2gQDaJBNLJDNPbs2ZM87aK9vT1ZOXfuXFjJyclpa2tL3Rlu 7tu37/333099TaoPP/wwbF6/fv2zHndwIhpEg2gQDaJBNIgG0SAaRINoEA2i QTSIBtEgGkSDaBANokE0iEZ2iMbly5e3bdt29OjR7pXm5ubktaTq6ur6/oKl paXjxo0LO0+dOvWsxx2ciAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEg GkSDaBANopEdotGj+/fvv//++2Hn9OnT+9i2Z8+eV199NYGP9evXP3z4MPK4 zymiQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoZJlo1NTU LFy4MHmL8Nzc3L7f7HvOnDndbyO+YcOG1tbWtI/7XCMaRINoEA2iQTSIBtEg GkSDaBANokE0iAbRIBpEg2gQDaJBNIhGlonGmTNnupFi0qRJ58+f72NzXV1d bW3t3r17J0+eHPbPmzev7+M+tjTvxrNENIgG0SAaRINoEA2iQTSIBtEgGkSD aBANokE0iAbRIBpEg2gQjYwVjScJQt+I0NHRUVtbW1xc/O677yabL1y40O+x CgoKks2NjY19fD/p3I2BiGgQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBAN okE0iAbRIBoZKxpP6ullYcaMGcnLSaUutre3X7t2rbi4OHXx4cOHY8aMCZtP njwZf9wBj2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBrZ IRq1tbXnz5+vr69PXfz888/D5ry8vNTFa9euhcWcnJz79+93L4aPk7feuHz5 8jMdd3AiGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIRnaI xpIlS8L6J598kroYbobFhQsXpi52Px3jww8//O6778JK+OfatWvDyrhx49rb 25/puIMT0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNLJD NJKnY4wdO/bWrVvJyo0bN0aPHh0W9+zZ02Pz0aNHk3fNyM3NXbNmzeuvv57c PHXq1LMed3AiGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSI RnaIxoMHD6ZPn57AxMKFC1esWJG8itTs2bM7Ojp67z9+/Hhubm73W41Pmzat xztrPOVxByeiQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINo ZIdohOrr61euXNmNFCNGjPjwww9T3yyjR52dnTU1NaWlpU1NTTHHHYSIBtEg GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2ikTWikXT//v3r169X VlY+fPhwMI/7XCMaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJB NIhGlolG9h23i2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbR IBpEI+OP20U0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBAN opHxx+0iGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtHI +OOGbu39o7Tnbz4sjJmqw38bM1fvtsbMH648GzN/taEw7an6eljMxPzQFabm 1NSY+f0V52Lm5ie/mva0NFyKmZqCvJgpW/cPY6b+6ocxU1Z/L+1572RZzGw/ Vxkzi74qiZnI38cr9/xO2nNjw/8WM5EAd/dee8xsKboZMzc++EdpT8XHvxwz W8/cjJnacwtiJvLH3dsH/yrt+fdLz8XMjfX/U8ycudUYM3fPL4qZ3112Lu35 7MLtmIl8zO2fTbsSM5E/RN3e/6O05+MzN2Pmrz44HTN/+1FBzGw7Wxkz9dc2 pz3/7p3zMRP5f/MVR6/GTGtT1Nw5MT7tabpzIma+uFwdM7VnF8RMfemGmIn7 bz4hZn5/+bmYufPt5JipK1kXM3+3pSDtqTr005iZ/vmFmLl2tzVmIv/3FnOC ijyx/8HKszGT9/MLMVO++Z/EzO+vOJv21BbNiZmmO8djZtmRqzET+fcwrc3X 056agtdjpvbM3Ji5dKc5ZprvFsXMhoLytCfyAbQbG/7XmGmq+jpmpu05HzM/ Xnc67Vlw8HLMRP4hd+R/9siTzI/XFaU956qaYqbx5t6YaSjfGTM/2VSQ9sQ8 2BvmRT223/UcZOHIkSP5+flXrlwZ5OM+fUSDaBANokE0iAbRIBpEg2gQDaJB NIgG0SAaRINoEA2iQTSIBtHIStFoaWlZu3btRx991O/OsrKyESNGhK955MiR +OM+p4gG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaKRlaKx YsWKsG3y5Ml9b2tvb8/Lyxv2fUSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAa RINoEA2iQTSIBtEgGgNYv7Jw5MiRBCn6FY0NGzYM+0VEg2gQDaJBNIgG0SAa RINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBoDWN+ycOfOndGjRz+NaJw5cybs mTJlyoIFC4gG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJB NAa2PmThu+++mzdvXtiwZs2avkWjqalp4sSJYc/ly5eXLl1KNIgG0SAaRINo EA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaIxsPUhCzt37gyfXbRoUVlZ Wd+ikSjGxx9/3P0x0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQ DaJBNIjGAPYkWbh69erw4cPHjRtXV1d348aNPkTj0KFD4bMzZ87s6OjoIhpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbReA49Vhba2tqm TZsWPvXNN9+Em32IRlVV1ejRo3NycioqKpIVokE0iAbRIBpEg2gQDaJBNIgG 0SAaRINoEA2iQTSIBtEgGkSDaBCNAe+xsrB+/fqwvmrVquRmH6Ixf/788KnZ s2cf/0UzZ84MK++99174uL6+vo/jPraBul99RDSIBtEgGkSDaBANokE0iAbR IBpEg2gQDaJBNIgG0SAaRINoEI2MFY0nCUJvRCguLk7W9+/fnyBF8oYaY8eO DR+fO3cudXNubm4fX7mw8Il3zXM0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2i QTSIBtEgGkSDaBANovH09ZaF5H0xntSsWbNSN+/cuXPzD3v11VfDtoULF4aP b9269fTHHbSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2i kQWice3atR5IsXLlyrBt5MiR4eMvvvii7y/ofTSIBtEgGkSDaBANokE0iAbR IBpEg2gQDaJBNIgG0SAaRINoEA2iMeA9jSz08T4avSMaRINoEA2iQTSIBtEg GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0RjwBlw0li9fHjYfO3Ys/rjPKaJB NIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2hkpWhUVFSEba+9 9togH/c5RTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEI2s FI1sOm4X0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhG xh+3i2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEI+OP 20U0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopHxx+0i GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtHI+OOGKrb+ ctrzx6vPxEz5xv89Zg6W1sTMgStRM/HTM2lPTeHrMRNpMc13T8fMby26EDMx P6zmn7geM3WXV8dMxZb/O2Ym7TwbM3Ul76U9jZX7Y2b9tzdi5vNLVTHzHzYX xEzl7n+b9tRdXBEzt/f9Scycu90UM2/suxgzMf9Tb6k7GzN1V9bFzMKDJTHz 4/WnY2bH+Vtpz28vvhAzN97/n2PmJxsLY6a26I2Yifmlct6Xl2Om6qufxEyk aMT+//Tj/yftaazYEzMfFd2Mmb94/3TM/HRTQcxUbPt/057qY6NjJvJvQloa L8fMzU/+eczc+uKP0p7qoyNjJvIEFfnj3+7iqpiJ+d/qx0U3YybmR8cw1cfG xMzKY9di5o/ePZv2VO78jZj56w2FMVN36d2Yqb3XHjPjd5xJe5YfvRozv7fs XMx8cKo8ZloaS2Lmp5sL0p7W1psxM2rb6ZiJhKS7xUtj5l7Hd2nP3Qtvx0xr c1nMNJR/FjO1Z34WM023v0p7yj/8xzFTdeg/xMysLy7GTHXLg5iJMYU/XXMm ZmIe+gsT+Wtaw/WtMfMX60+nPcXVzTFTV7I2Zu5eXB4zMX8We6zsbsy8qMf2 uwZIFs6cOfNFr7755pvnfdz0IhpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEg GkSDaBANokE0iEbWiEZLS0tvpAjV1tb23jxr1qxhvZo9e3Yaxx2EiAbRIBpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopE1onH69OneSBEqKirq sfPhw4c5OTnhUxs3btyS0v79+9M47iBENIgG0SAaRINoEA2iQTSIBtEgGkSD aBANokE0iAbRIBpEg2gQjawRjZ07d4ZPvfHGG1t+2O3bt3vsLC8vDztzc3MH 5LiDENEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSyRjSW LVsWPnXo0KF+v8LRo0fDzrfffntAjjsIEQ2iQTSIBtEgGkSDaBANokE0iAbR IBpEg2gQDaJBNIgG0SAaRCNrRGPKlCnhU9evX+/3K2zevDnsXLZs2fvvvz9v 3rzFixdv27atubk5veMOQkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINo EA2iQTSIBtHIDtFobW1N3jVj796977zzzpw5c9asWVNYWPjYr7BgwYLeb7cx ceLEsrKyZz3u4EQ0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSD aBCN7BCNS5cuPfZtwdetW9dj56NHj8aOHRs+lZeXV1BQUFVVdfLkyVdeeSWs TJ8+vaOj45mOOzgRDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbR IBpEIztEY9++fQlhfPzxx2VlZeXl5WvXrk1WTpw4kbrz0aNHZ8+ePXjwYFNT U/fi9evXhw8fHjZfunSpj+M+tgG8a0+KaBANokE0iAbRIBpEg2gQDaJBNIgG 0SAaRINoEA2iQTSIBtEgGhkrGk8ShMciQk1NzYkTJ4qKilIXly5dGjYvWLDg aQ43Y8aMsPnAgQN9fD/PehcGKqJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBAN okE0iAbRIBpEg2hkrGg8qaeXhYKCgrB5/PjxqYsdHR1VVVU3btzosXnOnDlE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBoD1WNlobGx sby8PPwzdfHs2bO9RaO0tDR5okdFRUX3Ymtr6+jRo8NicXHxMx13cCIaRINo EA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhGdojG4sWLw/ry5ctT F7dv3x4W58+fn7rY0dExatSosL5ixYqHDx+GlfDP/Pz8sDJlyhTvDE40iAbR IBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBCNAemxsrBr167kmRfX r19PVsIHiVwcOnSox+aDBw8mmydMmLB69erXX389fJyTk3PmzJlnPe7gRDSI BtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEI3sEI379+9PmzYt cYoFCxasWLEiJycnfLxw4cLvvvuu9/6vv/46Nze3+63GZ8yY0U0hz3TcwYlo EA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAa2SEaodra2hUr VnQjxZgxYzZt2vRYzkjq7OwM/0p5efm9e/dijjsIEQ2iQTSIBtEgGkSDaBAN okE0iAbRIBpEg2gQDaJBNIgG0SAaRCNrRCOpra2tsrKyqqqqs7NzMI/7XCMa RINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhGlolG9h23i2gQ DaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEI+OP20U0iAbR IBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopHxx+0iGkSDaBAN okE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtHI+OOGqr76m7Tnt94s jpnq42Njpun2wZipKZwRM7+5sDjtifzN7g/fPRsz+0ruxMywLQUxc/FOc9pT ueu3Y+b2gb+ImZs7fi1mIh+Zj/lR+e75t2Im5oeuMHcvvBUzX165EzNNVYfT nrvFy2Lm24r6mFl94nrMRMpj+ab/I+2pv/ZRzEQ+BlJ3eU3MNLR1xEzNqdfS nphrSpjb+/40Zuou58fMjQ/+Ucw03/km7fm3S87HzJ1TU2LmV1+5GjOtLTdi pv76lrQn9pfKD07HzK/97GLM3Pn21Zipu7Qy7Sm82RAzf/XB6ZiZsvtczLTU n4+ZmtOz0p76+x0x01C2LWYif6E+XdkQM3++rijteeWzszGz88LtmIn8qTvm joeJeeipsWJ3zPz+8nMxE/PYUZgFB0ti5u75xWlPS8OlmIn8/TQSLm/v+7OY Gbu9KO2paLgfM5H/c40xrDC1RbNjJuZPIxpubI+Zys9+K2aGfVwQM03Vx2Km 6vDfpT23v/zzqIn7fSESfGd/cTFm1n5TlvZcqG6KmcgTVPXXw6PmyIiY+dHa orQn8geJm5/+q5hpKNseMzM+L0577nyTGzMv6rH9roGThaqqqr17965evXrt 2rX79u1ra2sbnOOmEdEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINo EA2iQTSySTQ6OztPnjy5ZcuWFStWhH+WlpY+aeeJEydGjRo1LKXJkyfX19en d9znHdEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSyRjSa m5sXLVo07Id99tlnvXdWVVUNHz48fHbt2rUlJSX79u175ZVXws1ly5alcdxB iGgQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpZIxpr1qwJ n8rNzT1+/HhxcfHKlSsT1CgvL++xc/369WH9zTff7OzsTFYKCwvDysiRIzs6 Op71uIMQ0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNLJD NGpra5OnXVy8eLF7cebMmb2fptHe3p6TkxPWa2pquhc7OzuLiopKS0sfPXr0 TMcdnIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaKRHaKx Y8eO5FWkUhcrKyuLi4vr6upSF6urq8POqVOndnR0nD17dtu2bVu3bi0oKOh+ vsYzHXdwIhpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iEZ2 iEZ+fn5YP3HiRHV19cGDBzds2LBv377wce+dly9fDjvnzJmzcOHC1HfcWLBg QXNz87Med3AiGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSI RnaIRsITGzduTF5RKmnkyJFHjx7tsfPEiRPdG9auXXvy5MkdO3aEneFmfn7+ sx53cCIaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhGdojG 1KlTE6SYPHnygQMHjh49Om/evHAzJyensrIydeeRI0eSndu2beteDP9Kstjj Jar6Pe7gRDSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEI3s EI3JkyeH9XHjxjU2NiYr3333XW5ublhcv3596s7z588neJH6GlNNTU3JG4uH z/Zx3Mc2QHerr4gG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQ DaKRsaLxJEF4LCIkz8jYuHFj6mK4GRbDp1IXKysrky/S1taWup6XlxcWCwsL +/h+4u5Q+hENokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkQj Y0XjST1WFlauXNlbNE6ePBkWZ82albrY1taWvGtG6tMxwuLo0aPDYkNDwzMd d3AiGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIRnaIxo4d O3o/HWPnzp1h8YMPPuixedOmTWH9tddeS1546tGjR8mzOaZOnfqsxx2ciAbR IBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopEdotHQ0JA882Lf vn3JSnV19bhx48LKiRMnemyuq6ubOHFi+NT48ePffffduXPnho/Dv15SUvKs xx2ciAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopEdohH6 5JNPkjfImDt37urVq8eMGRM+XrZsWWdnZ+/N9fX1ixYtSt4NPDR79uwzZ86k d9xBiGgQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpZIxqh Q4cOTZgwIUGKcePGrVmzpr29vY8vFT5bUVHR1NQUedznHdEgGkSDaBANokE0 iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSySTSSamtrb9269dinZjzX4z6/ iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopF9opFlx+0i GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtHI+ON2EQ2i QTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoZPxxu4gG0SAa RINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNDL+uF1xovHXHxbG zN0Lb8dM5P+1Z+4tjpn5By6nPS0NF2Pm9c8vxEzDjR0x8yerz8RMzH/z21/+ OGYarm+NmZrTM2PmzjcTYqa55tu0p6X+Qszc3PYrMfPj9adj5lpda9S88w/S npbGKzHTUP5ZzFyuaYmZFceuxUzMI/OR9ldTkBczu4pvx8yfrS2KmRM36tKe fza9JGaqj4+LmQ0F5TFTdfCvY+bNr0rSnlMV9THT0lgSM7/yamnMrPv2Rszs vliV9jTfOREz7x6/FjORxF9TMC1mqo+NSXte230+ZiLPMF9drYmZMZ8Uxcyd byenPbVnfhYzMT8ChYkx+jCNlftipuKj/yvtaa4tiJmfboqa5rtFMRPpUDH/ zW9u/+cxs+diVcz8m7cuxMzd4ndiprnmZNrzr/7+Ysz8h48KYibyp6BbP//d mPn1eRfTnj9/ryhmIn+9/dP8MzGTs7UwZq6/+1+lPY239sfM+aqmmKm79G7M RP55amvHd2nPrc//IGaur/5vYqah7JOYGfFxYcz88zcupz3NtadiJvIRsMrd /zZmIv+zT/7sXNrzR6vOxkx96QcxU1LTEjMx+vbBqRsx86Ie2++KloWmpqa9 T+727dvP6bgxEQ2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAa RCM7ROP06dNPQoojR46k7rx27dqwJ3f06NFnOu7gRDSIBtEgGkSDaBANokE0 iAbRIBpEg2gQDaJBNIgG0SAaRINoEI3sEI0lS5Y8CSmmTJmSurO+vn5Lr8Ke sHPkyJE3b958puMOTkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2i QTSIBtHIDtE4cuRID6TYsGFDIhpr1qzp+wvW1dWNGTMm7Dxx4sSzHndwIhpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iEZ2iEbvPvroo7Bz 1qxZDx486GPbo0ePFi9eHHauWrVqQI77PCIaRINoEA2iQTSIBtEgGkSDaBAN okE0iAbRIBpEg2gQDaJBNIhGVopGSUlJ2DZixIg+XkUq6dChQ2Hn2LFjGxoa 4o/7nCIaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhG9onG gwcPpk6dGrZt3bq1750PHz6cOHFi2Llz58744z6/iAbRIBpEg2gQDaJBNIgG 0SAaRINoEA2iQTSIBtEgGkSDaBANopF9orF///6wZ8yYMffv3+97Z0FBQfJU jn6foPE0x31+EQ2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAa RCP7RGPGjBlhz8aNG/v9Um+99VbY+e677z7lcR/bU33TcRENokE0iAbRIBpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkQjY0XjSYLQNyJcvXo12XPr1q2+ v35tbW2y8/Lly0/5/TzDdz+gEQ2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQ DaJBNIgG0SAaRCNjReNJ9S0L69atCxsWLlzY79f57LPPws4ZM2YMyHGfa0SD aBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtHIJtFob28fPXp0 2FBQUNDv11m6dGnYuWPHjvjjPu+IBtEgGkSDaBANokE0iAbRIBpEg2gQDaJB NIgG0SAaRINoEA2ikU2icePGjfDZ4cOHd3R09Pt1pk6dGjafPHky/rjPO6JB NIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2hkk2icOHEifDYv L6/fL9Le3j58+PCwubKyMv64zzuiQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQ DaJBNIgG0SAaRINoZJNobN++PXx25cqV/X6RqqqqsHPEiBEPHz6MP+7zjmgQ DaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBrZJBqrV68On925 c+cgH/d5RzSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEI1s Eo2sPG4X0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIhG xh+3i2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEI+OP 20U0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANopHxx+0i GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIxhAWjcLCwk2b Ni1fvnzz5s2nT58etOOmUWvz9bRnzTdRU/X1sJiJ/I34et29mLlRn/7c3v+j mFly+ErM/GhtUczc+uKPY+Zu8dK0p+HGpzHTXHcmZiJ/AGis2B0zRZWNaU/j zb0xU1rbGjMxv8uHaW26EjXNZWlP0a3GmPn1v7gZM7+79FzMfHKuMmb+bM2Z tOdvNhbGzO+8fT5m7nz7asw0PngYMy11Z9OeyAdpb37yKzFTe3Z+zLS23IiZ mEdZ809cj5nbzW0xU3NqSszEUE6Y1uZraU/9lfdj5tDV2phpevAwZj49fytm fnfZubSn/uqmmPnLD07HTMP1LTGzqbAiZmKke+qe8zHzNx8WxsyND/6XmLnX 0RkzMY+yHrl+N2aaqo/FTPPdMzHzJ/lRE/PQ09tfl8bM5tMVMdNY8fOYyd15 NmaqDv007Yn8G6SW9u9iJvJ/7ZF/D/Oj94rSnhhFClN/bXPMRDp75B9QfVtR n/Y0VR2JmVl7i2Pm7oUlMVN/dWPM1JyenfZEYvGZW40xM+Pz4pip/Ow3Y+bP 1hSlPWO3R8035XUxc3P7v4iZ2nMLY+Y3FhSnPc21hTHTdPtQzET+4Fp79u/T nj9efSZmXtRj+119ykJjY+Pu3bvzvy98EG4+dltnZ+fy5cuH/bCwEtbTO+7z jmgQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpZIxqXLl0a O3ZsKlKMGzcuLPbe+dVXX4XPDh8+fPv27SUlJeGfOTk5YeXAgQNpHHcQIhpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iEZ2iMZ33333yiuv hE/NmTPn1KlTBQUFc+fODTfDYkdHR4/NS5cuDZ9asWJF98qqVavCypIlS571 uIMT0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNLJDNG7c uJE8L6OysjJZuXXrVrJy7dq1HpvfeeedsL5+/frulW3btoWVBQsWPOtxByei QTSIBtEgGkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoZIdolJWVJX7R /d4Z9+/fT15L6uLFiz02HzhwIHlNqqtXryb/7qRJk8LKnj17nvW4gxPRIBpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0skM0Ojs7J06cGD61 YcOG1tbWtra2LVu2hJtjxozp/apT4bPr1q3rfq+N5IOFCxc2Nzc/63EHJ6JB NIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbRIBpEg2hkh2iEiouLE9QY +X3hg9GjRx85cqT3zsbGxuStNFLbsmVLb/t4muMOQkSDaBANokE0iAbRIBpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtHIGtE4cuTIqFGjUpHilVde6f0mGqG3 3norfHbatGnFxcUtLS3hn3l5eWFl8eLFfR/3sQ3MveozokE0iAbRIBpEg2gQ DaJBNIgG0SAaRINoEA2iQTSIBtEgGkSDaGSsaDxJEB6LCKWlpd1Ptaj7vq1b t4abw4cPLykpSd157969ESNGhPXu9xAPVVdXJ2+60dTU1Mf3M4D37pkiGkSD aBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRINoEA2iQTSIRsaKxpN6rCwkfrFq 1arUxeTNMjZu3Ji6WFRUFBYnTZrU4yvMnj07rJ84ceKZjjs4EQ2iQTSIBtEg GkSDaBANokE0iAbRIBpEg2gQDaJBNIgG0SAaRCM7RCPxiMOHD6cuJngxbdq0 1MXk2Rw5OTnt7e3di52dnZMnTw7r4V95puMOTkSDaBANokE0iAbRIBpEg2gQ DaJBNIgG0SAaRINoEA2iQTSIBtHIDtFYvHhxWN+xY0fq4uHDh8Pi7NmzUxfb 2trGjBkT1tetW/fo0aNkcc+ePQlzeNUpokE0iAbRIBpEg2gQDaJBNIgG0SAa RINoEA2iQTSIBtEgGkSDaAxIj5WFzz77LKyPGzeuvLw8WWloaJgyZUpY/Oij j3ps/vrrr5M33cjLy3vvvffmzZuX3Ny7d++zHndwIhpEg2gQDaJBNIgG0SAa RINoEA2iQTSIBtEgGkSDaBANokE0iEZ2iEZ7e3teXl741KhRo95+++38/Pxx 48aFm5MnT25tbe29/9SpU8nLTCXl5uYeO3YsjeMOTkSDaBANokE0iAbRIBpE g2gQDaJBNIgG0SAaRINoEA2iQTSIBtHIDtEINTU15efnDx8+vNspVqxYUV9f 38eXamhoKC8v73tPv8cdhIgG0SAaRINoEA2iQTSIBtEgGkSDaBANokE0iAbR IBpEg2gQDaKRNaKR1NHRUfl9qW/8PQjHfa4RDaJBNIgG0SAaRINoEA2iQTSI BtEgGkSDaBANokE0iAbRIBpEI8tEI/uO20U0iAbRIBpEg2gQDaJBNIgG0SAa RINoEA2iQTSIBtEgGkSDaBANopHxx+0iGkSDaBANokE0iAbRIBpEg2gQDaJB NIgG0SAaRINoEA2iQTSIBtHI+ON2EQ2iQTSIBtEgGkSDaBANokE0iAbRIBpE g2gQDaJBNIgG0SAaRINoZPxxJUmSJEmSJEnSyxjRkCRJkiRJkiRJmR/RkCRJ kiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/R kCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJ mR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJ kiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJ kiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/R kCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJ mR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJ kiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJkiRJkiRJmR/RkCRJ kiRJkiRJmd/QFA2eIklKckWQJKXmuiBJSs11QZJ6NzQf2x+a91qSlGm5IkiS UnNdkCSl5rogSb0bmo/tD817LUnKtFwRJEmpuS5IklJzXZCk3g3Nx/aH5r2W JGVargiSpNRcFyRJqbkuSFLvhuZj+0PzXkuSMi1XBElSaq4LkqTUXBckqXdD 87H9oXmvJUmZliuCJCk11wVJUmquC5LUu6H52P7QvNeSpEzLFUGSlJrrgiQp NdcFSerd0Hxsf5gkSZIkSZIkSdKz9EJEQ5IkSZIkSZIkSZIkSZIkSZIkSZIk SZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIkSZIk SZIkSZIkSZIkSZIkSZIkSZKkftu3b9/SpUtbW1tf9DciSRr4Ojs7v/nmm02b Ni1fvvzDDz88fPjww4cPe+wpLi7euHHjihUrduzYcfny5d5fJFwj9uzZs3r1 6nXr1oWv0NLSMijfuyTp+XbkyJH8/PwrV670WO/3utDvBknSS1RVVdXevXvD T/tr167dt29fW1tbjw2uC5KGVKdPn977hMLPz6k7+z1/FhYWJg/IbN68OXzZ QbwT2dypU6eGfV9dXd2L/l4kSQPcvXv3Zs+ePeyH5eXlpZ7zd+/e3WPDnj17 Ur9IS0tLbm5u6oZp06bV1tYO+r2RJA1kZWVlI0aMCGf1Hr+X9Xtd6HeDJOkl 6sSJE6NGjUo9q0+ePLm+vr57g+uCpKHWkiVLhj2hKVOmdG/r+/zZ2dm5fPny Hv96WAnrL+huZUOPHj06dOhQ9392oiFJ2deaNWvCGX78+PGffPJJSUnJrl27 Jk2aFFYWLVqUbKisrMzJyRk+fPgXX3xRUVERNiQXhfPnz3d/kfz8/LAyd+7c 0tLSS5cuzZw5M9z82c9+9oLukyRpAGpvb8/Ly0vO+ami0e914WkuHJKkl6Wq qqpwSg+n8bVr14bfF/bt2/fKK6+Em8uWLUs2uC5IGoKFH4+3/LANGzYkJ7c1 a9Yke/o9f3711VfhZtizffv2sCH8M5wtw8qBAwde3D17uaupqVm4cGGqEBEN ScqyOjo6Ro4cGc7w+/fv7148fvx4ctpPXmwwuRZs3ry5e0O4UoeVd955J7l5 +fLlcHPEiBFNTU3JSvgguQqXlZUN3p2RJA1o3b+U9RCNfq8L/W6QJL1ErV+/ PpzD33zzze6/GS4sLAwr4feI8NtEl+uCJH3fRx99FM5ss2bNevDgQbLS7/lz 6dKl4eaKFSu6v8iqVavCypIlSwb/+8+OkqcEhv/CBw8eTB6YIhqSlGVVVVVN nz59woQJ3RgRChff5PGr8vLy9vb25C8Kbty40b2hoqIi+SuC+/fvh5vbt28P N8OFOPUrJ0+c3Lp166DdF0nSAHbmzJnkKfMLFixIFY1+rwtPc+GQJL0shbN6 8ohQTU1N92JnZ2dRUVFpaemjR49cFyQpVFJSkvyp582bN5OVfs+f4eY777wT Nqxfv757w7Zt28JK+Al8kL//rOmrr77asGFD8iroyV/wEg1JGgpduHAh+f0i XH/LysqSj5OrbXfJdfn69etdv7gE7969O3XDnj17wuLbb789qN+6JGkgampq mjhxYjiNX758OfnLsW7R6Pe68DQXDknSy1J1dXU4e0+dOrWjo+Ps2bPbtm3b unVrQUFB998buy5I0oMHD8J5ssdfdfZ7/gwdOHAg7Bk3btzVq1e7vj+jJi8D 7p2GBiTP0ZCkIdK9e/emT5/e/ZyL4uLi8HFubm6Pbcn7gCevfDt//vzw8aFD h1I3hJveSkOSXtISxfj444+7P+4WjX6vC09z4ZAkvSwlLzA7Z86cHi9LvmDB gubm5i7XBUnq6tq/f384oY0ZMyb1eWf9nj9DbW1t69atS9bHjRuXfBD2d29Q TERDkoZC4WKavLrI2LFj6+vrw8rZs2fDzddee63HzkQ9CgsLw8fhAh0+Pn78 eOqGb7/9NizOnDlz0L55SdKAlJB0OIGnvrpvt2j0e114mguHJOll6cSJE92P wq1du/bkyZM7duxIXscjPz+/y3VBkrq6ZsyYEU5oGzduTF3s9/wZamxsTH7Y Tm3Lli3Jz+GKjGhIUtYXrqSJTYQrbHFxcbKYvALV1KlTe2wOv5KE9fDrSfh4 3rx54eNwsU7dkLy9ePiCg/PNS5IGpKqqqtGjR4cf/isqKpKVHqLR73XhaS4c kqSXpXD+Tx5h27ZtW/di8jIpycNErguShnhXr15NTom3bt1KXe/3/BluvvXW W+HjadOmFRcXt7S0hH/m5eWFlcWLFw/23cjGiIYkZXfV1dVTpkxJnudYUlLS vV5ZWZks9tifPB0y2bly5crw8Zdffpm6IXnS5aJFiwbhm5ckDVTJCwnOnj37 +C+aOXNmWHnvvffCx/X19f1eF57mwiFJelk6f/588uBb6kugNDU1JW/2HT7r uiBpiJe8bNTChQt7rPd7/rx3796IESPCzXCe7N5QXV2dPA4fdg7SHcjeiIYk ZXE3b95MXsZ2ypQpPf6ooKWlJbkEt7e3dy8+ePAgWaypqQk3N23a1OOvDkLh ZvLMysG5C5KkASm5HDypwsLCfq8LT3PhkCS9LCUeEWpra0tdT/6K2HVB0hAv nNlGjx4dzmYFBQU9PtXv+bOoqCh8MGnSpB7/4uzZs4f1eh0MpRHRkKRsLVxk x48fP+z7V4hqbGzsvSF5dOvbb7/tXjl58mRYmTBhQnIzecX1119/PfXfSq7R Bw8efK7fvCRpYNu5c+fmH/bqq68mf3UWPk7Uu9/rQr8bJEkvS21tbcmrvqe+ hXdYTB7Ba2ho6HJdkDSEu3HjRjibDR8+vPc7X/R7/iwtLQ0f5OTkpIJvZ2fn 5MmTw3pRUdEg3YfsjWhIUlb26NGj5AVGZsyY0dLS8t0PS/bs3r07bHjjjTfu 3bsXboZ/Jn8wsGvXrmRDuCInzxnvfnPwY8eOhZsTJ04MX/OF3C9J0kDV4300 up7iutDvBknSS1TyjOzXXnsteeGU8BvExo0bh6W8NYbrgqQhW/L233l5eY/9 bN/nz7a2tjFjxoSb69atC59K/pU9e/YkzOFVp+IjGpKUlRUWFvbx6iJXr17t +v6FpxKwmDJlSn5+fvJ2GzNnzuwmj9Cnn36a/FnC29+XvCzkyZMnX9w9kyQN TL1Fo9/rwtNcOCRJL0t1dXUTJ04MZ/Lx48e/++67c+fODR+PHDmy+y0wXBck Ddm2b98eTmgrV6587Gf7PX9+/fXXycMveXl577333rx585Kbe/fuHcQ7kbWl Pp1QkpQ17dq1qw/RuHbtWrKturr6jTfe6F6fPXt26htXdX3/lwa7d+8eNWpU siFcoHfs2PEi7pAkaYBbvnx5OLEfO3YsdbHf60K/GyRJL1H19fWLFi1K/mwp OaufOXMmdYPrgqSh2erVq8M5befOnU/a0O/589SpU8nLTCXl5ub2+MFbkiSl XUtLS3l5+d27d7ufDtmjzs7O27dvh99NerzvlSQpK+v3utDvBknSS1R7e3tF RUUfL4TiuiBJj63f82dDQ0M4PdbX1w/mdyVJkiRJkiRJkiRJkiRJkiRJkiRJ kiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJkiRJ kiRJkiRJkiRJSqP/Hwuuj6Y= "], {{0, 156.}, {1059., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{1059., 156.}, PlotRange->{{0, 1059.}, {0, 156.}}]], "Output", CellChangeTimes->{{3.763318538108815*^9, 3.763318565210346*^9}, 3.76332012412394*^9, 3.763334280368699*^9, 3.763362491569319*^9, 3.763362746709404*^9, {3.763382935901332*^9, 3.763382942393959*^9}, 3.7634041554181757`*^9, 3.763417105921311*^9, 3.763743998233479*^9, 3.763744187782196*^9, 3.7637448979281683`*^9, 3.763809923345448*^9, 3.763815805099676*^9, 3.763914841977969*^9, 3.7640056782735357`*^9, 3.76408110177481*^9, 3.764588861198346*^9, 3.797848679362348*^9, 3.797922479338172*^9, 3.798178770591461*^9, 3.798186192863658*^9}, CellLabel->"Out[18]="] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell["Transformer architecture", "Subsection", CellChangeTimes->{{3.728644832191423*^9, 3.7286448555810347`*^9}, { 3.7286721153422413`*^9, 3.7286721189663763`*^9}, {3.763323292847169*^9, 3.763323298806449*^9}, {3.76332580808814*^9, 3.763325821575308*^9}, 3.764089362881872*^9}], Cell["\<\ Each input text segment is first tokenized into words or subwords using a \ word-piece tokenizer and additional text normalization. Integer codes called \ token indices are generated from these tokens, together with additional \ segment indices:\ \>", "Text", CellChangeTimes->{{3.763325957099697*^9, 3.763326016499756*^9}, { 3.76332606073736*^9, 3.763326102686018*^9}, {3.7634171905498867`*^9, 3.763417205197248*^9}, {3.76341739721682*^9, 3.763417435375107*^9}, { 3.763979095711894*^9, 3.763979133242399*^9}, {3.7639886710850143`*^9, 3.7639887539776173`*^9}, {3.763988797642166*^9, 3.763988902361676*^9}, 3.7640893705785522`*^9, {3.764529696250965*^9, 3.7645297004897957`*^9}, { 3.764589057365163*^9, 3.764589058108364*^9}, {3.7645891375416393`*^9, 3.7645891519573717`*^9}, 3.764614722261842*^9}], Cell[CellGroupData[{ Cell[BoxData[{ RowBox[{ RowBox[{"net", "=", RowBox[{"NetModel", "[", RowBox[{"{", RowBox[{"\"\\"", ",", RowBox[{"\"\\"", "\[Rule]", "\"\\""}]}], "}"}], "]"}]}], ";"}], "\[IndentingNewLine]", RowBox[{"netencoder", "=", RowBox[{"NetExtract", "[", RowBox[{"net", ",", "\"\\""}], "]"}]}]}], "Input", CellChangeTimes->{{3.7634175067506933`*^9, 3.763417517756712*^9}, { 3.76374411982522*^9, 3.763744122047738*^9}, {3.76398896391179*^9, 3.763988976050241*^9}, 3.797848642300631*^9}, CellLabel->"In[19]:="], Cell[BoxData[ TagBox[ TemplateBox[{ RowBox[{ StyleBox[ TagBox["NetEncoder", "SummaryHead"], "NonInterpretableSummary"], StyleBox["[", "NonInterpretableSummary"], DynamicModuleBox[{Typeset`open = True}, PanelBox[ PaneSelectorBox[{False -> GridBox[{{ PaneBox[ ButtonBox[ DynamicBox[ FEPrivate`FrontEndResource[ "FEBitmaps", "SquarePlusIconMedium"], ImageSizeCache -> {12., {0., 12.}}], Appearance -> None, ButtonFunction :> (Typeset`open = True), Evaluator -> Automatic, Method -> "Preemptive"], Alignment -> {Center, Center}, ImageSize -> {Automatic, 24}], GridBox[{{ StyleBox[ TemplateBox[{"\"Type\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox["\"Function\"", "SummaryItem"]}, { StyleBox[ TemplateBox[{"\"Output\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox[ TemplateBox[{ TemplateBox[{"\"matrix\"", "\" \"", StyleBox[ TemplateBox[{ "\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"2\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], "\" of \"", TemplateBox[{"\"indices\"", "\" \"", StyleBox[ TemplateBox[{ "\"(\"", "\"\[VeryThinSpace]\"", "\"range\"", "\":\"", "\" \"", TemplateBox[{"1", "\"..\"", "30522"}, "RowDefault"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"]}, "RowDefault"], "SummaryItem"]}}, BaselinePosition -> Automatic, GridBoxAlignment -> { "Columns" -> {{Left}}, "Rows" -> {{Automatic}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> { "Columns" -> {{2}}, "Rows" -> {{Automatic}}}]}}, BaselinePosition -> {1, 3}, GridBoxAlignment -> {"Rows" -> {{Top}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> {"Columns" -> { Offset[0.], { Offset[0.5599999999999999]}, Offset[0.]}, "Rows" -> { Offset[0.], { Offset[0.4]}, Offset[0.]}}], True -> GridBox[{{ PaneBox[ ButtonBox[ DynamicBox[ FEPrivate`FrontEndResource[ "FEBitmaps", "SquareMinusIconMedium"], ImageSizeCache -> {12., {0., 12.}}], Appearance -> None, ButtonFunction :> (Typeset`open = False), Evaluator -> Automatic, Method -> "Preemptive"], Alignment -> {Center, Center}, ImageSize -> {Automatic, 24}], GridBox[{{ StyleBox[ TemplateBox[{"\"Type\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox["\"Function\"", "SummaryItem"]}, { StyleBox[ TemplateBox[{"\"Function\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox[ PaneBox[ TagBox[ TemplateBox[{"Function", "\"[\"", TemplateBox[{ StyleBox["\"x\"", Italic, StripOnInput -> False], "\",\"", "\"\[Ellipsis]\""}, "RowDefault"], "\"]\""}, "RowDefault"], Short[#, 3]& ], BaseStyle -> { IndentMaxFraction -> 0.1, ShowStringCharacters -> False}, ContentPadding -> False, ImageSize -> {250}], "SummaryItem"]}, { StyleBox[ TemplateBox[{"\"Pattern\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox[ PaneBox[ TagBox[ RowBox[{"{", RowBox[{ RowBox[{"_", "?", "StringQ"}], ".."}], "}"}], Short[#, 3]& ], BaseStyle -> { IndentMaxFraction -> 0.1, ShowStringCharacters -> False}, ContentPadding -> False, ImageSize -> {250}], "SummaryItem"]}, { StyleBox[ TemplateBox[{"\"Batched\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox["True", "SummaryItem"]}, { StyleBox[ TemplateBox[{"\"Output\"", "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox[ TemplateBox[{ TemplateBox[{"\"matrix\"", "\" \"", StyleBox[ TemplateBox[{ "\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"2\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], "\" of \"", TemplateBox[{"\"indices\"", "\" \"", StyleBox[ TemplateBox[{ "\"(\"", "\"\[VeryThinSpace]\"", "\"range\"", "\":\"", "\" \"", TemplateBox[{"1", "\"..\"", "30522"}, "RowDefault"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"]}, "RowDefault"], "SummaryItem"]}}, BaselinePosition -> {2, 1}, GridBoxAlignment -> { "Columns" -> {{Left}}, "Rows" -> {{Automatic}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> { "Columns" -> {{2}}, "Rows" -> {{Automatic}}}]}}, BaselinePosition -> {1, 3}, GridBoxAlignment -> {"Rows" -> {{Top}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> {"Columns" -> { Offset[0.], { Offset[0.5599999999999999]}, Offset[0.]}, "Rows" -> { Offset[0.], { Offset[0.4]}, Offset[0.]}}]}, Dynamic[Typeset`open], ImageSize -> Automatic], BaseStyle -> { ShowStringCharacters -> False, NumberMarks -> False, PrintPrecision -> 3, ShowSyntaxStyles -> False}, BaselinePosition -> Baseline]], StyleBox["]", "NonInterpretableSummary"]}]}, "CopyTag", DisplayFunction->(#& ), InterpretationFunction->("NetEncoder[<>]"& )], False, Editable->False, SelectWithContents->True, Selectable->False]], "Output", CellChangeTimes->{ 3.763417518007599*^9, 3.763743999563806*^9, {3.763744630050034*^9, 3.7637446339333563`*^9}, 3.7637449108093967`*^9, 3.763746328207345*^9, 3.7638100676492043`*^9, 3.7638158155137033`*^9, 3.7639148421928053`*^9, 3.7639890382499113`*^9, 3.764005693313197*^9, 3.76408110982201*^9, 3.764588931371828*^9, 3.7645956520946608`*^9, 3.797848682644313*^9, 3.797922480942421*^9, 3.798178772319181*^9, 3.798186090646594*^9, 3.798186194631308*^9}, CellLabel->"Out[20]="] }, Open ]], Cell["\<\ For each input subword token, the encoder yields a pair of indices that \ corresponds to the token index in the vocabulary and the index of the \ sentence within the list of input sentences:\ \>", "Text", CellChangeTimes->{{3.7639889102828608`*^9, 3.7639889307873163`*^9}, { 3.76398904932439*^9, 3.763989148375695*^9}, {3.76398925365703*^9, 3.7639892820735598`*^9}, {3.76408303145467*^9, 3.764083034092511*^9}, { 3.764083131171382*^9, 3.7640831532675447`*^9}, {3.7640924650326157`*^9, 3.764092484376627*^9}, {3.764588971936974*^9, 3.764589025019669*^9}, 3.76461470291188*^9, {3.764614823895678*^9, 3.764614831343031*^9}, { 3.764655638984601*^9, 3.764655641997635*^9}, {3.773419605809214*^9, 3.773419616103622*^9}, {3.799769777099064*^9, 3.7997697930270157`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"netencoder", "[", RowBox[{"{", RowBox[{"\"\\"", ",", "\"\\""}], "}"}], "]"}]], "Input", CellChangeTimes->{{3.763383118430627*^9, 3.763383139217394*^9}, 3.763383773061701*^9, {3.763988983484301*^9, 3.7639889891076193`*^9}, 3.763989042677826*^9}, CellLabel->"In[21]:="], Cell[BoxData[ RowBox[{"{", RowBox[{ RowBox[{"{", RowBox[{"102", ",", "1"}], "}"}], ",", RowBox[{"{", RowBox[{"7593", ",", "1"}], "}"}], ",", RowBox[{"{", RowBox[{"2089", ",", "1"}], "}"}], ",", RowBox[{"{", RowBox[{"1000", ",", "1"}], "}"}], ",", RowBox[{"{", RowBox[{"103", ",", "1"}], "}"}], ",", RowBox[{"{", RowBox[{"1046", ",", "2"}], "}"}], ",", RowBox[{"{", RowBox[{"2573", ",", "2"}], "}"}], ",", RowBox[{"{", RowBox[{"2183", ",", "2"}], "}"}], ",", RowBox[{"{", RowBox[{"103", ",", "2"}], "}"}]}], "}"}]], "Output", CellChangeTimes->{{3.763383131376433*^9, 3.7633831395963717`*^9}, 3.763417171878956*^9, 3.76374399961908*^9, 3.763744634029052*^9, 3.7637449109159937`*^9, 3.763810067747102*^9, 3.763815815618141*^9, 3.763914842246482*^9, {3.763989038741749*^9, 3.7639890430509167`*^9}, 3.7640057129548693`*^9, 3.7640811098557053`*^9, 3.797848686008052*^9, 3.797922483150771*^9, 3.798178774567219*^9, 3.7981861981213417`*^9}, CellLabel->"Out[21]="] }, Open ]], Cell["\<\ The list of tokens always starts with special token index 102, which \ corresponds to the classification index. Also the special token index 103 is used as a separator between the different \ text segments. Each subword token is also assigned a positional index:\ \>", "Text", CellChangeTimes->{{3.763989206410557*^9, 3.763989210552516*^9}, { 3.763989286950203*^9, 3.7639894098632383`*^9}, {3.7640893984036617`*^9, 3.7640893988510036`*^9}, {3.764589092933003*^9, 3.7645892138381157`*^9}, { 3.7645892833651867`*^9, 3.764589363918144*^9}, {3.772283125886462*^9, 3.772283126830949*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"net", "[", RowBox[{ RowBox[{"{", RowBox[{"\"\\"", ",", "\"\\""}], "}"}], ",", RowBox[{"NetPort", "[", RowBox[{"{", RowBox[{ "\"\\"", ",", "\"\\"", ",", "\"\\""}], "}"}], "]"}]}], "]"}]], "Input", CellChangeTimes->{{3.76338376776565*^9, 3.7633838269794903`*^9}, { 3.763989489233639*^9, 3.763989495881793*^9}}, CellLabel->"In[22]:="], Cell[BoxData[ RowBox[{"{", RowBox[{ "1.`", ",", "2.`", ",", "3.`", ",", "4.`", ",", "5.`", ",", "6.`", ",", "7.`", ",", "8.`", ",", "9.`"}], "}"}]], "Output", CellChangeTimes->{{3.763383798098482*^9, 3.763383827949637*^9}, 3.763417216822151*^9, 3.763743999797264*^9, 3.7637446344198647`*^9, 3.7637449113638487`*^9, 3.7638100682104893`*^9, 3.763815816035344*^9, 3.763914842407913*^9, {3.763989471936702*^9, 3.763989496305593*^9}, 3.764005724746593*^9, 3.764081110555564*^9, 3.797848688251143*^9, 3.79792248444383*^9, 3.7981787759879093`*^9, 3.798186200316805*^9}, CellLabel->"Out[22]="] }, Open ]], Cell["\<\ A lookup is done to map these indices to numeric vectors of size 768:\ \>", "Text", CellChangeTimes->{{3.7633261135381413`*^9, 3.763326152355667*^9}, 3.763383011966428*^9, {3.763384065437243*^9, 3.763384085531546*^9}, { 3.764589426849443*^9, 3.764589428061653*^9}}], Cell[CellGroupData[{ Cell[BoxData[{ RowBox[{ RowBox[{"embeddings", "=", RowBox[{"net", "[", RowBox[{ RowBox[{"{", RowBox[{"\"\\"", ",", "\"\\""}], "}"}], ",", "\[IndentingNewLine]", RowBox[{"{", RowBox[{ RowBox[{"NetPort", "[", RowBox[{"{", RowBox[{ "\"\\"", ",", "\"\\"", ",", "\"\\""}], "}"}], "]"}], ",", "\[IndentingNewLine]", RowBox[{"NetPort", "[", RowBox[{"{", RowBox[{ "\"\\"", ",", "\"\\"", ",", "\"\\""}], "}"}], "]"}], ",", "\[IndentingNewLine]", RowBox[{"NetPort", "[", RowBox[{"{", RowBox[{ "\"\\"", ",", "\"\\"", ",", "\"\\""}], "}"}], "]"}]}], "}"}]}], "]"}]}], ";"}], "\[IndentingNewLine]", RowBox[{"Map", "[", RowBox[{"MatrixPlot", ",", "embeddings"}], "]"}]}], "Input", CellChangeTimes->{{3.763383180924172*^9, 3.763383246288885*^9}, { 3.76338331479277*^9, 3.763383363464138*^9}, {3.7633833961740932`*^9, 3.763383410758196*^9}, {3.763383524413321*^9, 3.76338353764106*^9}, { 3.763417357579316*^9, 3.7634173622654943`*^9}, {3.763810838690617*^9, 3.763810842648672*^9}, 3.763989529506411*^9, {3.764005734003913*^9, 3.7640057530589933`*^9}, {3.7645894373616247`*^9, 3.7645894443500757`*^9}, {3.798186080246503*^9, 3.798186083225348*^9}}, CellLabel->"In[23]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsvQd428id9+8kd5e7tDflkrvkcu+l3SW5vZTnLpdL8r9NTzbZjbPZ7rXX fS333i3JlrstucqWrN67qE51kmLvIMECFrCDAEGCXdLaa+9m/Y6ELP8KZcuS 7PXKznyeefgAgx+AKQDm9wVmhl9es/35jA8uWrRo99+Cn+dX7//Zrl2rD77w SbDy0rbdG9dvW/f6b7ftWbd+3a4frPkQiMz5wKJFa/5q0aKJ5dsQCAQCgUAg EAgEAoFAIBAIBAKBQCAQCAQCgUAgEAgEAoFAIBAIBAKBQCAQCAQCgUAgEAgE AoFAIBAIBAKBQCCQx46hoaErV66Mj4+/3wm5LxiG6e7uLisrKy4u7uzsJAgi zQBksLe3t6SkpLKyUiQSjY2NzdVgwSKRSEpLS3EcT4vHMKy+vr6wsLCjo8Nm s03f8Z4GC4o//vGPKpWqubkZJBj8OhyONIOFX8XJZHLg7gQCgZQlTdMgBiS1 vLwc3KE3btxIO1QikeByuaWTgAWw+nCzAoFAIBAIBPI4o9FoMiaJRqPvd1rm D/B4N27cmDGF9evXA384ZQD84d27d081yMzMDIfDszdYsHg8ng0bNoAEA600 NR54zhl/ztQCmY3BgmJ0dPT8+fNpCQa6OGXwSFSxy+XKuDtSqZQ1UygUmzdv nrpp3759sVgsdRyr1bpt27apBtu3bweRDzMvEAgEAoFAII8l77zzjlAoTDlj j65KoiiKlUjAi1apVDqd7urVq2ym7HY7a1NaWgpWT5486XA4gDOZk5MDVk+d OpU6yD0NFiY3b97MyspiMztVJZEkCcoESMXBwUGCIHp6elgbk8k0S4OFRllZ GUgekDlyuRzDsGvXrrEJ9vl8rMEjUcVA6TRP48CBAyAlmzZt8vv9tye/IoF6 ATHl5eXgAh4aGtq7dy9YBVc1e5C3336bjTlx4oRGo9FqtSBTYBVE3rp162Fm BwKBQCAQCOQxg2GYvLy8qe+iH12V1N3dDdK/a9cu4D2yMX/84x8PHToEIpua msCqzWYDyxs2bEgmk6wBWGCFlcfjmY3BgqW2tjZVg1NVElu5jY2NqRjgjYOY /Pz8WRosKMLhMCscLBZLKpKVOeznpEe3isF9t3XrVpAMhULBxlRVVYHVc+fO gcuYjUEQhJVRrAjyer1sjQOpyxpQFMXGuFyu9yUXEAgEAoFAII8HbG8r4HcJ BALWV3x0VRJw7/ft2wd+p0a2tramXr+3t7eD5StXrkw1KCgoAJHAbDYGCxMU RUEKDxw4kJubO1Ul3bx5k9UUwJ1OGRMEkTHZEfH69ev3NHjYObkXHR0d7LeV qZFAI2AYxl63j2gVv/POOxcuXABpKC4uZmNA1bD3I8MwKTMgl/R6vcPhAPa3 J/tYspooNRYJVBm711QVCYFAIBAIBAKZKyMjI7W1teygDKCVHmmVdEfOnDkD MgV8Y7Ccn58PloEwnGrQ29sLIi9fvjwbgwVIMpnctWsXSKHNZgPO/1SVxHrR QO+wTnUK1pF2u933NHhouZglbGc5hUIRDAaBrgeX7tDQEFhOGTyiVSwUCkEC tm3bFo/H2RiQKRBz6NChW7duGQwGDocDRJxWq019V7o9KZrYqgflMD4+fuPG DfY74NatW2GPOwgEAoFAIJAHxaP+LWk6rPMJVADr8LOKCUROt2GHpdzTYAHC KqOWlpbUckolYRjGDuFJ24Wdu8BkMt3T4CGkf06w/QPr6+unztEB1H1quoNH sYrfeustVux0dXWlItmegSdOnEjrEJubmzs6OpoyAzXI7rtpErCwZcuWtOk7 IBAIBAKBQCD3w2OmkhAEYbuTpfrgAZ8TrMrl8qlmarUaRObk5MzGYKHBuvcg bey3gzSVZDAYwOrBgwfT9jp8+DCIB+VzT4OHkIU5wY4yy5ic6o3P5wNxdPr0 abAKLl12bM6jWMVarZYdKpX6kHR7cna7lDIqLy9XqVQdHR2sDiotLU2ZgbpO mwRv7969cFASBAKBQCAQyAPkcVJJQD6wEik/P/+tt95iI1mPOjU6ngU4zOxL +9kYLChomt6yZQuotdR/QqWpJLPZzPbaStsRyCIQDyTSPQ3e6yzMFSCOMiYn u06NxHn77bfZL19VVVW3H80qvnjxIjh7UVHR1EhQiazq4XA4qUggDNlI9iZ1 OBzsanNzc3QSdhQeuPJTkzpCIBAIBAKBQO6Tx0YlsSP0ASUlJVMHaLCzRvN4 vKnGw8PDGZOTh8/GYEHBdh47duyY/F3Y2d4qKirAciwWI0mS1RRpO4KYjMnZ 0e9p8LCyMltYjVNfXz81EqyCSLDp9iNYxeFwmL1W0/7M12QysfFT+9clk0lW +7OdIVlNlJrwgaWysnJ6EUEgEAgEAoFA5s1joJLeeecd1mdmX8KnTUrQ0NCQ 9nIeAFYz3p027Z4GC4q0v0ZNA0GQsbExdvnmzZupvd588002kmGYexq8H9ma CVbjpEkAlUoFIo8ePXr7Eaxidvr6I0eOpMWzAhZw48aNqfHsv2KxnSGBQAbL IpFoqoFer8+Y/J/c9zjhEAgEAoFAIH8pPAYqqaamhu1xlOY6srCjeLKzs6dG sm6nQCCYjcGCoqurq/HP2b9/P0hqXl4eWKYo6va7SkqtVqf2YjXFzp072dV7 Giwo2JnA2c9GKUA5gEhQ9bcfwSpmO0mCfKXFA3HEjkKaOocGiNyyZQuIZEcw sZOHp+0Lrnz2C+NDSDwEAoFAIBDIXwKPukpiR9kApFLp238OO38ycDLZvmSp wfsymSxj8o9ox8bGZmOwwEkbl3T73f/DOn78+BtvvAFWwS/7AaKnp2eWBgsK oA5Y7TA0NMTGBINBtsrYoUaPXBWz81EAZTp9E/vZ6+DBg2ynu9R30tQ4MvY7 FMiOz+djY0D5HDhwIOPdv1GGQCAQCAQCgdw/j7pKOnr06N26n126dIm16ezs ZD82XZ6EHeUx1Ue9p8FCZrpKAp4/KwqA81xaWsq60Dk5OUA5ztJgodHW1sbW 6cmTJ0tKSrZu3Zox+a/BqT8SeoSqOPWvvuwEfWmAO5Gd5XvHjh1FRUUgvxmT M36nxouB3dmvYJs3bwYZAdXHVuW+ffvGx8cfblYgEAgEAoFAHlumduZ55ABO 8oYNG+6pkt555x0ul5uaPBn4nGkdlu5psJApKCgAaZbJZFMjg8Hg8ePHU0Vx 7NixNJ/8ngYLDaFQuHPnTja1QBeUlZVNHVf1CFUxTdMZk3OAp6ZhTCMWi50/ f55VUmzVoCg61SCZTAJxlDIAFBYWgr0eSvIhEAgEAoFAII8VQFIFAgGgBdKG xs/e4JFjbGzM5/NFIpG0GS1mb7DQCIfDFEWlPiGl8ThVMdCABEEAQXQ3g1u3 bpGTTFWLEAgEAoFAIBAIBAKBQCAQCAQCgUAgEAgEAoFAIBAIBAKBQCAQCAQC gUAgEAgEAoFAIBAIBAKBQCAQCAQCgUAgEAgEAoFAIBAIBAKBQCAQCAQCgUAg EAgEAoFAIJAHS2VlJZfLfb9TAYFAIBAIBAKBQCALhX//939/5ZVX3u9UQCAQ CAQCgUAgEMhCAaokCAQCgUAgEAgEApkKVEkQCAQCgUBmJhaLXb16VSAQvN8J gUAg7zN/OU8DqJIgEAgEAoHMwM2bN0+cOJGRkVFVVfV+pwUCgbyf/EU9DaBK gkAgEAgEcjfC4fC5c+cyJpmNX5QBgUAefR7I0+BRB6okCAQCgUAgd4TP52/Z siXlOM1SJd0z5vEG5vfx5nHK793ycsf4eTwNHnWgSoJAIBAIBHJHWHdoz549 Bw4cAAtnzpx566237rnLPWMeb2B+H28ep/zOSSXN42nwqANVEgQCgUAgkOm8 8cYbU18ds2RlZUWj0Rn2gioJ5vfx5nHK7+xV0vyeBo86UCVBIBAIBAKZTllZ GXCEduzY0dbWVlNTA5Y3b94Mfs+fPz/DXlAlwfw+3jxO+Z29Sprf02DhEwgE Ghsb77YVqiQIBAKBQCBp3Lp1a9OmTcALKi0tBU5RTk4OWD527Bj7Dnl8fPxu Oz5OPiQE8pdJ2l0876fBwufcuXM/+tGP7rYVqiQIBAKBQCBp0DSdlZW1YcOG O85/5fP57rbjVDOw+vw1AwgvFKGvluuXVepfq9atrtNlNCPrW5D1zUhGsxaE 090DKsllRpuNSc5suiZ6+qzhhdMjFy/swQr/L3bt/xYW7Fp1ZfjVIuP+FqPL px+N6AJBTOslhc6w3EVRpCpGiZIBQZIcSvgH4t7OqK0sMLSE5D7j7/wl0fzf vvpv+9t+FrWWg60JR0uEuyZ05uPB038Xav5VWHEgrDkaGV4fu/yVeOYi/MAX N29sfnqd6SfrvT/e6frfbPzXJ8xXm7kWVXFEfzo8soXJ/TST84Fwwb+FRTsi loKAtUqNYyIH1W0KnOM7MxpNa2s0h1t5dtW5sP64eKRoR5nwqXPYsovCoYqX rTXf99b+u7fm37xVX/bW/Ku/8+eB4aVB8UZGcyTJqMbi1okQM4MwGlJE8eqQ cm9QsZNBjkSM5yKWaxFna2w0Grt+K54MxWlZguAm/H2jtGQ0JE9SI3FnfcRw 2qc8K+oue2Wj7efLyE3bB4YK1xIt/0Nw/jcw+BLNX+6X7LToG5QuUumL4eFx 5o034+NjsaiTMV1kdDmMNiuk2htS7AwqtgclG2nhWrdgQ01n0dY60dIK/YtF 6M9PW3583LqpSC4U1kawfLuu5FCTaHk58lqFcU+75ZLIWSyziwxS0tkd93Ci 5vwAfzXZ/RQxuBx3KKwB2uZzujTFRPczRNuTvvpveSu+5K38ElgAhQDOFVLs jVqL4u4Wp6XjUItieblueamqrDnfxF3jHliq7t91uk9yqNdU0N+nGT4Y4L8G dmF0x0HJOPXX2kSdezu1e9vkTZzjppanibYfU33Ph3WnYnh11FwArgR/96+J th+BwvcU/4O76O8nw2c9Zf9MNP1P2HghileC7IRU+4PSTSHlHpCGBDnkdosa JYrtzdp19YZsrs0QSBLx62TQEbI3hI15tPma0Y1pfcEha/jsgO8PV7DnLyNF zVVmwUFSvseuzO1DLe1GaljFswxs8nQ/Qw4sCRjKmXjMF7teoyK3NGNr64yH e3BrcDQ09maQRALSA1Tv4sDgi4zqQFh3zKu7LFR3lcispXKcozWRvpEENaxA RfsakMWXTauLlBL+Ja8qx4+cRg1dA5h30EqanGjQxw94BjUm/q5WzcpaZHMr emzAekGIl0hMAp0giNdFnQ0UqfGEGXt4VE1Ei+WuSyIchDveuWkqad5Pg4XP d77znc9//vN32wpVEmT2bN26FVxO4PdhnpRhmO9MMjQ09ADtT5w48b3vfW// /v3vvPPOA0rpYwiscQjkL5na2lrgAu3cuXNgYIDtY5Pi5s2bd9srzb+CKgmq JKiSHgmVNMNdPO+nwQLHYDAsWrRox44ddzOYpUpqaWmpq6tTKBTTN8lkMrBJ pVLdV0LnApuYFI2NjXw+32q13rp166Gl4S+TX/7yl+By+t3vfvcwTxoIBBZN Aur9QdmjKLroXQYHBx9cYh83YI1DZg8o5MLCwlWrVu3du7ezszMSibzfKYLc F6BJ3bp1K3CBgFMEVvv6+lJOUWZm5gw7pvlXL5WgILxcqgfONlBJK6q1GxsU eznC/RzhnjbFjnZkR7v+Qh9PKy9gdMesinOZVcKVV9ENV4RlxYdt1f8FQklZ 9rpiwfIyY2ab0e1DRyN6OmDQO81Si0VtMwf88lhAnKCFcUoY9QvD3sGAtcE6 sN7cvcza8Zy7+Ue+hu/6238WwYrjvp4YXh/pXxu8+LnA+b/3c571yI961Lme 4T2eqz/wHv2C9vB/btvd/twO0293uX6XafnDKf3S8+rK9na75lpYfyYwsgs/ /03r8X9yFj9Jio+EbNWkrVnnMCmcnkHMVyhy7G7HtrfoznaLHUhhxHReISnN qelfli/bUsAVNazCOU8BB97X+qS3+f/ztPzExX3Zwd/oEu/zac7Eg4rRmCkW MdMhE0mbfX6Vy9xok+bYJFk+bV4QKw7bayOengl99MZ4POGPByRxX3fC1xP3 D8bJ4TjBjeFVEcMZQpWr6C/ddNC0ZIs7M3tQVL2D7H6G4j4bHFkbEm+k5Aft uhqt3Yg4Ha4gHR4bjY3FQgyOI9csilyH4phfujMo3RiQbvKLt7qFOyyCfQ3c iqzWkS0NmrWVyFN52M9PWbYWS0XCSpA7kMczHdJtjYatTdgxrrFUpKuVaiUG kcfZH/JwSXM5ztuOdS0x9W/SmJVAmmlwG6YqJrjPkp2/IFq+76v7dxD8Ld+n BSuD4vUh5b4Ylh93VHvMjWe75Dsa9TvqNY1dJTbeTt/IJnQk5ypffpZvrhIM oKLjtHh9ULoVqJWo5arHVNEr6z7ZpznJVXRxL1q5K/19L/mG17k1+Q5jI46U m3rXGzpeQVsW62t+oiv+jq7oCUvxV1wl/+St+Iq/9cmIpTjmbo3aykNITlCx K6DOImxtXhfPYBMDlZTdod3fpj83aDIThJ+hKMpI460hrIi01GgdVqmL5ppD eYO+ZSXW5cVoRWeDRXKcUh+2afLbteZ6LcFVCM38Q57B1eTIloClKRSPeaPX q1Xk5mZsTZ0xm4vbQmPM+JshSkcrsgMDLwSGlhGqoz7kvB0pFqo762TaOoW+ G9H6fBKGEsmN8iNthmUlls2V2oHhIlRy1qS8JEN6uSZvH+Y3OPRAJQU9g6iF f6xHu4ujO9RlOMuzXRY5yqSmEYQfstdGHXUBr9BLWXHSjbgctTJ9qUgLwmzu 4nk/DRY4e/bsAV4BgiB3M5ilSvrABz4AjvPZz352eufD//3f/wWbNmzYcL9p vX0bSK1Tp04JBIKZzRbdhX/8x388fPjw6Ojo/adkHpw7dy43N/fGjRtz2isv L+/Uu3C53Nnskkwm6+vrjx49arFY5pXS+fPY+MwYhqUum97e3geX2McNWOOQ 2QCee0uWLEl7IH/sYx/r6up6UKeYZevwcJhTYurq6jo6OmKx2GyM33777dOn T6caBR6Pd38pvS9AtQ4NDVVXV1+/fh2sTn17XFJSMsOOaf7V6w16ENbWA32k e61Kt6FGcqm1ZqT3kHQgq0fUWqcwVqm9PTo9jrVH7eVBa43eMKzQi9X6EZuh nTGWkGhpq4B7qF21u81wbshI0OaxuDlKDPmUpxzDW9yig1HvQJJRJ6NGmsHd tNdCuGVW7Fxnz9FmTnHzBXXLi8An93f+gtEejWJXI8YLIfHWQPuvfG2/Efcd axL0VIyISgeHrrQ0XKwpO1dTt79WtbPWfLBBf619YGiwVMQrxOSXAvpcSper l1w4U1Zy6GpZfmOrSK+xEzhOWH0+FeGTOd1Kuc3cafS2G3zDGB4kJElKQFo5 Gv4ZQcc+KTfTh1xlzGVha43f2oFb+Rg2MqTi1QgGgNvfKR+gPEMJcsjuELZp kDKp4eqI/vyA5GRnX253H1ejtPlsVMhFM854zJaM2RKMNuHjxl1NcUdNWH+a 0RxmNNlh/ckwyJrxCqEvU0mGpSKxTsEjsY6YoznmaIri9VG8LmypIJFzLtE+ t/hgwNoYCyGhsM3oNJ1qFe+rEeQ3VCk5a8jOX7q6X1AMHW0T97VK+UPqIYVu QKodbBMJVlUgzxUa9tcOSoZzQ8pdpPooYtXLHLTM7tXo+43iU0bxSZ2+TW3X q5yOQRQtHuw/28Y53d6V06M9zDXl9iu7hso9I5sDI2sDglUBwYqAYCUtWhfS HGKQw4w2O6Q+GFIdoDTHTRahzm7QOzC3RxX2C2IkP+SX2v02zO90+LQ03gZE TcxenfD3JShelBrxekVGXGzAJR6PIuxX0V61AVMVDcgym6WHGkV7K7l7yjr3 lLbtLWnaV1yTXVLSWrnBVPck0fIjauC1eABJJshEyBBzc6LWYq+pspLPz+uX XxhUNssVEqNch4mM5n6//kpAd44yV/o8ci9ltxL2Nr2rWu0ulXuuST35Qs81 sUuIKklHf9DVYzAPnhzQ7es25gtQNab1uDUBv54Meon4mJ0ZK5MRG+rNq6qM 2V24K/JG8sZb8bA9bLoSUuz0KTN5yu4ulbhbJRKpOEZ1Aaa9ZkYbjG6L0eeV 4d4WhCiXE8US16kuaQ6Hf7x95NIwUqFyNSAulU0X8ImifgFNiE0uo86JiazW JsRZqnTXqzCJfpA2F0XM+TRymlRkE7JMj+ywWXbKKDsDwmzu4nk/DRYyf/zj H7/whS8AHTSDzZxUEgA8t9M2PUCVlJ2dDQ712muvzWzGpuSJJ574wyS/+MUv QC4++MEPsvHAowMZv//EzIm33nqLLSKn0zmnHf/6r//6u9/97m8nmc08IXa7 /dOf/jQ40Te/+U2lUjnf9M6Tx8ZnBhQUFCxevBg4JLD/1QzAGofck2Qy+f3v f5+tgs9+9rMvvvjik08++Vd/9VdgFTyWr1y58kDOMsvW4eEwp8QcOXLkW9/6 FigKICSBCJrZ+NatW08//TTbInzmM5/ZvHnzg0jvA4BhmKlDEjQazQzGaUMb NrboQVjfpF9ZM6GSNtcKyzjXLAMZDt4GubyGqze1G0kBZnbbuTFnXdTVFPQM BTwjtFcYJnhx/yBY7VRJj/TpDvQYLo0Y/TQ2Fsfi3h5KtM3X/Tv/4HKwPBrR JuMWKuzDg4zeHx62+ve2qzc0yE411khblhGtP5hQSarMqCEvgp4JyXZQAy96 +pYMDl2+xhfl8Q2nhnTZnYr9HMmBNtmhLkNWF3amV9suaLcr87zqUxRyOoSe JpBzSknhjireimJJJgfpM/pMVNhCUZRfHib4AZ/Q7DZKnKTIGVR7/OGQYTSM RN1dPukhZ/8KD39T2FYX8/VGSSFJoTjlNxFkB4rnjxgu8JEaqYJw9iV8PSbL cJlIfXoQze5F93RoNzcrdnFU9RqHMcAQsRgVC8Yj5mQETQRlCV9n3FUfxysm +olJtgSlW4FWipguRcwFDFYesHMpfDDo5sVISTIoT9DShH8w4esDiimoOUEM LvUPvgb0VJwW0yGDxmbYUqN79RqSWdYkbFhBtP7Q3v47Qf/pSrGmUmYSohKX vddl6xFrBzfUaV8uQzMb+iRDJ4PSzQFVlo8weZiohya9lg6feIdXtN2ANEpt FqEjwEHdpwZ0u9rkOzjKTc3Ihkb9gVZJY2+lV7wHpHYiyLYG5duDyt1h3bEw ejKsPRxU7ApKNgeVB0NefihoZBh7LGIbjVlHY5bRmC0ac0dinghjihMDcTcn 4e1M0JJESJkIqSK0LOQXhkhxJGxNJBk6GtW66ePd2IpyZEUZsrRI/Wqh6tVr yqVFiteKZetKeBXV2WjDr4m2n1LDr8ejruT1NxIJX5wYjDnr3ZaWvD7xLo76 YKemSa6wOSW0jx90ttOqrKBsF6W77CN07nAUC4TrtJ4CqeOKxFGicNdqfM2I V203BEgZ7RchNunBLmR9i+70sEXhpDyhMB2N+WNJb3zcyoyWSIn1teaVlcas Locnen3s5tvJmCNiLWW0h7yaU10qYa3C2KjUilUcr+akHznlNpQgLqfay0hd oX5rsMtMN+rI/e0GUB2bGrRHuFix3F2jdsktOooQxamRWEDCMKZQ2GLx27qM rgq1p0ljkekHaFNBxHg+qNhH8deQw8v9/DV+2U6/fA8Id7xzp/e4m9/TYCED dN8ddc1U5qqSPvWpTyUSiambHqBKAgeZvUrKzc2dGunz+YBiYjdVVlbef2Lm RDAYZE89D5XU19c3e/s1a9YAhyQUCs0xgQ+Gx8lnhswGWOOQe3LixAm2/IE+ Sv3DoNFoBIoJRH74wx8GjuT9n2WWrcPDYR6Jqa6uBrvMMN/sdECLthBUEoqi HA5n/fr1Kafo4MGDMplshl3S/KuNLToQ1jfrVtbqllfrttSJyzvKLLztzpGd Wk2zALMP2UMyO447BJSjg3Z2Rf2DcYoXD/AS5ADQDkFXb6dSnNOnPdiDXh4x +Ce/JSV8XbR4K8l9mhp+LeTuZYJoKIz7QkAl0aifHrQS2zjGFXW6Iw3NopaV BOdH/q5fMdrDUdPFiOkio8miRRmEcBNfWFI+IrkkQHOH0X3d6NY2/c4O/cl+ bd6gppgnGpbU+NWHQ9pM4MNHTOf96BWFvHJTrXJJmX43x9qsJ6WeiMobIAlN xC9mSKmTMOq8XsTrN/ucEVqepMWMo9UpzsL61lmGdzqxdrdL6PJq7X6nKRBD yHgz6s/l208NGUtFKq+tFbjoZmN7lUh4blBxrFe1p129tlELyq1GhRv9FBGh CYYgKCNB6iiviHE0RK3FYezqxEcB0W63aJ9TccqOlOC6crehMuJoiLma4t6O BCVIAklFyyKUOOwXhVzdtPoE2fsHqve5MHohEZTSDKaym1+vND571birpLO3 YYuN87Sh8zX+cH6dwlCvtirN0oCrN+Ds1qAD25s0y6vRnJY+2fCfVJKHMDhD jCtAENaOoHxvQLYHRVoEmA048016IpdvyexFD3F1u9vUO1uVR9p4nP5ir3hn QLqVkO1zy7Pdihyv6lhIfyZsOAu0UlAxIaBC6swkJRiNGscS9vEEPhkcYwln MulLJolExAwUX9zdmvB2JIOSJKOOhzRBUuHxiD1eKU2bo1EiwJAq3Hu4E3ul RL+kBF1aZlxWblpabn6lxPRCkeG1YmVp7VkD53l/zzMB4bZY1B+/fiueoIG+ iHs4blvniR7ZxibNDo62UqTSWWRup8BrbQ8oDwVl20ndZQ9hdITHTYFEM+Io l5lLpOYKBV6j9TbpPGq7MUAqaL9EZ5NndiObWnW5PEzpJLwhmoowzlDEQMVV vliplNjTatveZDk74PCFo2NvJJMRS8ReyaAnvPqLnRpVjRpvVJml2s6AHsjz M6SpBHVatG5C6fSKrTgo3m69ZU+bcXUt+nq9MbvPWqpy1yFumc1AEIowJY3Q qmjEGo3iNr+j0+AqVbjrVWYJ0kcbr0QMeSAXgeFXqIHnA7xXafnOoGovCLO5 i+f9NFjILF++HKibmaeemJNK+sQnPgF+jxw5MnXTPVXS6OioUqkE8uGer3BB IztvlQS4fv365z73ObDpbhJ4fHxcrVbjOD7Xj023bt1yOBwSicRsNt/xj4ZB /MNRSR/5yEdAaztLY1DgINkqlWo2vRCBVxMOh1OrYFmj0bz55ptTbVifefHi xewqKBatVjvLLM/+MgCQJAmM33jjjduTM6vc0weeq/08AK44OMU9u83cvHkT lCSCIGxiHjKwxmeTsFny3tU4eIYAe1BTsymZOT21gLHFYgFt1kN7kTI2NvaZ z3wGFP5///d/pz0bwXXI1ssMA2NnzyxbhzTAtQEqESTy/hNwn4kB1fe1r31t 3759s99lgaikrKys6VNa5eTkzLBLWhOc0YSAsK4RWV2PrKpDtjQpLvdwecJK qbRGbZLrfBRKJcCvxqYHvqHOIgl4h6PEYNzbFbWWh3UnCfWpFl5TZqdsT6fu PE/nI3WjEV3cy2U0R+iRdaRkj9k6grosqNdlI2wev9nuMQqNxi1N+mWVuuza FlHLWn/7T0ju4ojxUtxRF3c1RW0VYeOVoOGqSd8oRPp71YJ6sehQp2ZdA7q9 RVPN6+0UdQxLanWiM6RwAy3aENadiuJVFN6q1nXvaNGsqNVvazOdE+BlSneD 1mXx2UjaQgctBGVy+gwOD+JxihhzScSYb1Pl9/ReqeRcqeooqR8ZbpDIW5VI rxEXu5kRV7hC7Tk2iOX0IgVDI071BQY54VRfHJE3cCXtjSPcMz0j6xq0G5qQ WrHSYpf4PWK7UzaAIt2ITqgdcKjOh5T7/YpstaxALCkfFtW081qq+rpr+zq6 hhq88hOMKjusOz3RM83VGHG3eX1Kpw91OkUeVZ6/ZzHF/X3YUJAMG4NRSu3w ra3Efn/FtKlMVN1eyu8/K+DlCxXtYoseON52h5DxdNHODr2xN6sT2dJizOse VgvzQso9hOq4xKwZtPj4VpcB44dMhUFjoQId5CDWWq2vQevm6G0deqwdMTSJ eA2CvnZ+s5x/lhCs9Y+s0QuPCsRlPGmtXFkZMOZHzBfDxryI/nh4YqqKvGRQ OhrDxhL4WBwbi6KjUTQZxZh4IJRgmJAl7G6LWIujeGWS5CeD8iCp0mDKZomy SSxXIgN+C8ePtSjUXUfa5Esr9CuqDXvbbFmd+MF2fEOd5cUiw7IiVVlzEda/ lQaVqzpBh0l69AYTo6MBRYLod+NDJ7mKTc3ajU3I0W5t8bC6fkTeIx4gNGfD SE7AXObwWSzBURMZkWDaEVQygEgalcgVsb1QahMa5KRrIOjuNVl4eTxDdh9W JNKD65lwj3jdYhmmr9e6q9S+CjlRKvVXyYhOxBOgTEnGkKDEMUd91FJIWaqG DWg76u424FpMyNhro/bqkL3R4RgGFWG1DRkN7aiuSaTiZLVL19drNjWjZ3l4 u5HqswRULg8Orn/S6ibtjoDHEQBV7qlUOc6N2IvEep6yK4BeiKAng5It1ODL ZD9QSSvC+hMR0zkQZnMXz/tpsGABzeVHP/rRn/70pzObzUklZWZmslpp6j/t zqCS6urqvv71r6e+Q4H0gHOlNd+9vb3f/OY3v/SlL/3DP/wD20kDCIf/8+f8 13/919Rd7qaSbr/r1P3gBz9Ii+/q6praKw+kBDRzd3vD2dnZCU7KeoY3btzY v3//pz71qUXv8vd///epr2nPPPMMaPj+6Z/+CdizWz/+8Y+nJb61tXWGgp2r SpqlPXDwlixZwqpaAKiCr371q01NTVNtQCV++tOf/vKXvwzU37/8y78AM1D+ u3fvBl7fCy+8wNbF5z//eeABpnZhi/e5556z2WxPPfXU3/3d36W+MIId76gf b8/uMkhx5coVUJ6s5Yc+9KEf//jHUql0Bh94lvbDw8P/50588pOfvGMyXnrp JbC1vr4eZOrIkSP/8R//kboA/ud//ge4rNN3AZfT8uXL2XJjyzztXLMcdzY/ YI0v5BpftmwZWK2pqTGbzb/5zW8+8pGPsLuAxOzatetuxTinp1ZbWxvQKYum 8LnPfQ7YzzAo9YFQVFTEng7kbvpWtnUAl01qwObrr78OiuLJJ5+cbrx27Vqw CVxyqZi5tg4mkwkUKSg0dnQPuAXYtIEy/M53vgME/tTTAf3OHmH6ODUgYNlN U4cFzaOpSmOuqmchqCSfz5fyhc6fPw8EL7t84MCBGfZK86/WNGgnQqN2LQhN 2s0c5PigsUphadLaFe6gLzbOjN3EmdEBC9mEuLpQm9vBC7t7Yngdo86m+p5z dj9f1563p5m/lYOcHtB4fMpkSBEn+qLmq2HkuF93kY+qu1Fnt9Gjw5GAV0C5 hpSoYGMd8nKp/lB1m7Alg+z8ZaD/ReCIJgL8ZEg+GZSJoDzq6ws5WjxYk1Td eaRNtqwC3VKnEIlLHZoLpOYoKdpEcp+l+l5gdGfjRH+QkuvsyoPdEz2pNrbq NnH0W9r0mb1muSfkCMfJWDgYxCKEMOruCVuraWFGYHipjLvvWANndaVydZUy ow5ZX6/byTGc4+NtBhJ4tvlixyGu8WCn8mJPj423JcBfHhBvprSnKP0ls6a0 erBrfZ0GhPrhLlxXRhmL9JraM/3KQz36Am6Hirud7H3W0b+qQ8ytkSEFIiyz 07i8TL+mVJ5V2WbqXEr2PE0NLWGUe8P647TxEurQq91+jd1kUVzxdf8WbI2Y ykbjRGjsTbUnsabStjjftLrCdKbfVCYxNSiNUgtK07pIUBMleXFfe9DVZsJ6 84b12X3mIp5Ar7zK6A67tbm1EnG+0FQsxQQGdcjHBwEsgFXgmddq7GavIxxx MCET5WgjLTWk4Sop20cOvODpe2Vo6FKJQFQg0jXKJIS5ImotnAiWqyDE8PLR sBZIpPGkazSiTwYliYAkGtQS0ag3NkYErEFrJQP0lCkv7uMmAiKPR94sU2+q 12+sUdV1FlmHt7r5GbL+zKMtPStrdRuaDFdFrmq1t0zmzezEX7pmevUaUt7Z bJPnMvrTQazcGwq4o+P+MB0OoQla4vZIT/aqNzUDOa9bUaV/rRR9vQLJaRY7 9OVRS0EA51i9dpRMmPw04eYB8YhbOrqUI0f7URC4ymGfpYnB6+1YV5UcK5A4 WxQKi64iYLjgQPJbBJydHM2ONmOVyqf2xm10zEX5Il5u3N0ad3PY35CbK7fb eTZSaCfMHnOEEscpfozoC9vLGWtxyHQxoDxESXdgwswTHb2bmxU7OtBCqUvq imh8McQfRckI6o/oiLDKG1Z4wlyMuiCwZ3JNZwdUPeJWUncmrDsSFG+gBl4g +54PCNbG7GVxdxMIs7mL5/00WLCANgs8xsvLy2c2m5NKAo4HaGvAAhAOqU13 VElvvvnm008/zbZQH/7wh4HL8cUvfpFdBSoDtGUpS+BQLboX//qv/zr14Gzk HVUS8GfAJuCHTI1cs2ZN6lDAS/ybv/kbdhk0auxkHWkATwls/c///M/r16// 6le/SkvMV77ylZQlkEj3TDw42gwF+16opO7ubuAqsGf/whe+8K1vfSvl3E6t plQvQfYDHAtwO9OyDHzj1C6sz/y3f/u37AHBVQFUYcry5z//edrr7tlfBrcn 37Wm+kwumtSbrH+YcrbTfOY52QPle7cKumMZsjkF9z7IFGuW8mwBwEtP++YS iUT++Z//edGkPwb8drA7yG/aiTo6OmauuHkDa3yB1zh78G9/+9uszT2L8fYc n1qNjY1sIYDaBGf5wQ9+wH7fYTl37s6vCh8I27dvZ8vqjhPXlJaWsmlITTXz 6quvgtXvfve7043Z+R+mvuOaa+uQ0supcVJAWoIyYZfBZdzf358yttvtbDy4 VNJSYrPZ2E1Ti3oeTVUaC0ElAW04+15/77zzztGjR1lH6MyZM8XFxXv37k25 SWl976eSNrRhdYN29aRKYvXFjg7DyWFrucpdr/MpvFF//EZk/JY9NM7FAsCb bUFwBz7CuPujeENIle3n/s7RubihPfcAh7+zQ5c7jHh8qnhIFSMGIpZrYf0p Qp/fh6ibEUez3q20IoSLTziG5DrBhlrkpRL0YFWbsHUj2fXrwMArcWdTkhaN MqrRsDoZ1iQY1USPPncLbW/UIh0nOyVrqpHdjVKV9JofOctoD9Pi9WTvYiDT GH1ezM+nAwiCI5MqCdnQOtGfajNHd4hrkrppnIn6o8EgbYwRvLi7PWotCwjX UsOvSHr35TR2r6g2rKw2rK5FV9ehm5uNp4bwBsTfqPNfFOFZvaasbtVlbo9t aH1geElQtI7RHo2gebi2sGGoY0O9ZkO9tmGoE0euUYarOk3V8V7lni70UneH smcryf0d3reiSThYLDaf4zt3cawvFhlfLVTtKW03cF7yd/yC6n+RVuwJ6U+S xny9w6R002q7FVMU+Lp+Q3b/ljGVxWJ+avSmwp1cW2n9fb5xTSV2dshernA2 aXGF3RJm0EQYSVC8hLct5OSYMe4FPnp0ECsVCFFVAaPLcWlzK4TiPJ4xX4gN ogjtl4LAM+quSS15I/YGLW4nXeNJ91jMGvd2xpx1EawwqNxPDrzo7nulbzD/ Ck9+YcRULVH4/qSSrk2GoiheFaG14Yg9EnXEQkicEsbIkTClIEKkh2EIP0pj pYzuaNhwNubtjJM8l1vcKFGtr0MzqlU1bQXYwHr30EpZ394TrV1rG3SbWw0F YlcNq5K68JdAERVpy7ubbYo8Bj0TtJR7goQrkiAYMhzSJWix2yM53afZ0qJb XTcxJeMLRejSEuRAgwTXV0WtxQFHh92HG6kERtJBHy/m7fDhHf1q/qkBBIR+ 1ZDX2hpyNNmwnkoZdlXsbJHLLUg5jZ5zaC+18Fq3t2q2cYw1akJHxt3huC/o jXq64q5GcBHGPR0Jb3fIM6hyOEbwgAgnzV4rE1BGgUL0D0asRREsP4yeCSl3 BSUbrCN7TrZzgUra2WEolrlUnojeH9NNhDjij2v8cXBbyb3RHnMAVMSBHuOp fmW3CKiks2FdTlC8nhp4nux7bkIl4RVxDzh1yx3v3DSVNO+nwYIFeD6gUUgm kzObzUkldXV1tbe3s01P6s3wHVUSO7ceABRjqreDQqFgZcXUF4lAiZDvAtp3 sPX5558n/xzg3U09OHvk6SoJNG0gy2DToUOHUpGVlZWs/Q9/+ENgcHuyhwxQ Lh/72McWTU6LN71HDauSPv/5z7O+zRNPPAHaR5AGsKPZbAZSMWUJItkUsn3O ATKZLC3x7Hwgd+OBqyS/3886zF//+tdFIhEbCRyYVI2k3o6mfGbggPX29joc DtbrA2RlZYH6feaZZxZNvi9N9QtiC4T1Ny5dusQwDIi0Wq1AUbLxaR8vZn8Z 3J4cXM8aAyWu1+vBScEuQOanvtOl+cxzsgclkFYv4H5nze5YjKmcApYvX+7x eG5P9t559tln2Ujg/k21X7VqFVssYrGYjQFuIStGwIlmqK/7B9b4wq/xuRbj XJ9a//Zv/wbiwcMcXAypSBzHQUaAvkt7fj5Yfv/734NTf+Mb37jjVlA4bEZS n9XmpJLm2jqkVBJ7salUqrfffnt8fPzy5cusVgJFlBKkc1VJ82iq0lgIKgm0 m6A1Bx7ObIxBAU7vXZOiu7v7bjum+Ver6rUgrGvSHR205Apsl8R4mcLdrCM6 DKTcE7GHx7yxN9RErErlzeXbC8Q2LT7RdY3wiCnDtYkBRCMZI6LyailSpnC1 6lxOyhVgXLRfFjIVAi3jVv/pi8YlkY2jNUnNOpkZaVMia6rQF4oMByo7RtqA oHgmMLws7moGKmlinA41HPd2xL3tcaI3Tg6HCb7LwetXDNfxh9qEvbi+NGi6 EjbmhpT7AoKVExOymcoDpN4dcMvsjh1t6Ko63VaO8TTPdlXqrFCCtOoIQh30 y8Leoai7PeZsitqrGN1JRptjVFyrFsrODjtPDzmO9dtz+qw5/baj/fbjA46T gw4gbfLFzmIx1iTmu4Q76JHVQcnGiQnfdDm4Jq+Z17yjdWJ0TIuQ6zJW0li5 EW26wANliJUM9SO8zAB/uW1466Wuwcx2dEeLdU019nxhSiW96O/4OTG00q0r cdn7HbgA9XkRIqx32mxAJXX+muh8yqkutHpx4FT3mkOv12AvFhl2NKJNClRq 0avsRhthC0U94agrSopizuagrQ4zdVwTG/ME9lqxxKguBIl0anKL+CMnBvRn eOYmDaZzYagbGzDjtWpXmdLdZfK5QsHR8XBylA4HDQylCXqGafQSJVjpHl7V 3l9wokeSzdVdHRa7DSVRLD9qKYjZS4HfHrDVD+sV7VpDu9aoxUS0qzvuaYs4 Wwhjjc9Q7TeUhUz5YdP5sPlixFIctZX5zPXDKt5Fnu3iMDYgqHaK9nlF27SC 41f7Bw72mA/1YId77Ee4+KEu+5YmbGk5uqpCVdVeivH2BCVbafXRACGjAiaa VEd9/XFPuxfvKeCrD3PR3R2m9Q3GpRXoykokq1XqMDbE8Kqwp9cf8oLLlYjF oyF9IigLkWIjLh1AZP2IDLWrKFJDkYge150awA50Wwr4WqO5L+jqIJ3dQlR6 TWwvkLjaDJTIFVH7wgafJ+ztBxIy4eMCNZoMCBlKJne4efbAsJ1SujyY12rz GpwuSRBk2XCWQXKC8u20OMMi2H2kmZtRq9zWYqiQ2ayU1x0kTCSt9kUU3piG iGF00hYaFTqZyyL8INd4ekDdI+MGzMVR86WQbCc18PLEtyT+qojpYtQGyvDO T4O0u3jeT4OFCUVRH/zgB19++eV7Ws5JJTU0NADPhG3ZU13Np6skjUbDvuMF nlLacUA5s4cSCoXTz/K9731v0bzGJYGmsKOjg3UVgI+XmnADiES2s9w3v/nN tD+9Aq4je5yNGzemHZ9VSSzAR5rNcAPgA7D27/W4pHvaL168GCTjM5/5zPTx aKwPnOqEmfKZUwPNgE8FVn/yk5+wTnJFRQVrwDp1t9919r74xS8aDIapR47H 4+y7cVCDqcg5XQaJRIJ19b/0pS+lFTj7n19pPvBc7acDnNV7+swgkWkzdEWj 0Y9+9KPTLxtWbuzevXtqJOvVf+ELX5ghGfcPrPGFX+NzKsa5PrWADVvsd/So 7zmp2n3Cdk2c2k1uKm63m01wqlTnpJKmMpvWIaWSdu7cmZbxrKwsdlNKr81V Jc01MdNZCCoJAAoHXOezmby3qKhoBr9oZGTkbjumvY5eVTcxHGljC1okdzfo iFYD2WEK9FjoPmtQ5okaAwlLaFTgYC6K8P09xhOD2AjuN5JBnMR9tvagJjOg zjQbO6R2B/D65J4wHoq5wzGfHw0YC0Lq/bji+JWB4ewe7SGu8arY2qJzcnSu Egm+vNL03DXjvsouYcd2qv/3tGDFpEoSAqEE9FHUUhi1XIv7upOMKhHRRWmF D+c6zS1urClkrYzYyiMYEALHadmWgHyn38rxBDyWQJRvozc1mZZW6He2WapU vn5bUGDzuB0DIUdrxNEYddTF8OrJUBMDy446t50rMGHtBoqjJ8uVnmK5+7LI eZhrzag3rm8wnR12NCD+HqOLr5d7ZftpyYagbEtQtSekmcgUh1+3v0u9vxvp kA14zI1Be53F3FEsMZ0X4nUigUF6OijbbBEeyG4YXFOhXVJieqnI+FyBcek1 1d4yoJJe8Hf+zMPbaLUMmgififBP9MWi4ka33a4s8HX8ytP+S1SSL8HMA9ZQ vZZc12B6tQLN6tCKDQqKkFCkKhDC6XicjkUYUh5x1AetFVZjS6XcdFXiaJbL MM2ESnJocq8O8Q9ztTkDplI5PmwlBHZ/L0a2GchWlOTbg+7YWOLGjdj166FE jI5HA0EHbW2kxZs8gg1NvUX7OaIdbeqzXIFTVxgxXYhgV2L28pijxoM15g9J D3UjIHRJBzym+pi1LGy4RI5s9A+vDoi3MfpcUHdR7GpYf5LRHSV057T67m7U A4IG4XjVRwnlQYP8fK1o5JzAfmrItrkRW1NtWlllXF5peK0SXVetqGq9aOau pAZfpkWbIk5OlBiKenuijtqovcyL1VZLVOcFpmMDlp1tppU16Noa5Ei7zGnm xJyNMYIXiVORN25Fx98YTXpGE7Zk1BQKqHxuvtcjoGgsFAsSkajSRe9qM6+u NZzsRXV2JEQjTFBv9uJDVqoPXPaWYJ+VHrRRUtwR9Ani/v7JjqCK0QjChAwy l2/IFhiw0Xx7QOwkZQ43YtNQ6HmQU0ZzaGJuQFGGhbc7u6F3TZVqc6OhVmai gtZIxG6lfDJ3UOwOq3wTE1LQoze0/liBxHmIazoziHCVw7S9AZRwSHkwMPQq NTEuaTmDHA8b8kC4452bppLm/TRYmAAFAZ7hPT0997Sck0qqra29Pdk9e9Hk S1GSJG/fSSUdP3580WQfjDt+RgFN/9RGcypzUknAkXhiEuB7pPpUpB0ZuGRs ZHNz8/Tj/OIXvwCbgLZKi0+ppG9/+9tp3WzuxsNRSTdu3AA5ndppJI233nqL /Zp2x+8XoBBYd5pdTfnMqW8NwHleNDn6jF0Flz1rwFb07RlnPAO3yaLJTlap zxBzugz4fD57rumjG+44g9lc7aczG595at+zFGxnHiBAUjHATWVvkLQOrqkC fO/+XhPW+N3sp/M+1vicinEeT60f/ehHiyZHmR08eHCuj6D7hB3g9uyzz95x aygUYvNy8uRJNubhqKS0IUi3p1wkqUlfH75Keu6551atWjV7+/duXNL27dvB BZz2eXQ6p0+fnsEvmmHqwjT/am2DDoQtHEOp0tOC+jvNFNcSGLTTPJxWeMPG QBSjY2A5j2/d1Yke7jfzcEZPJWwk4bH3hrRHgtrDNqxb43Kz78mtoVGcGXMT ZspwNaTaa5flnO8d3Nep3tVlOD+C12t9TYi/UOxZVmH+Q6FxdyWX17HPM7DE J8gIOdsjlCRKiSOu1rDx4sRkDt6u0YhuNGYeDWvjRG/M3RJzN8ec9TFHbcxW FkbPBhW7aeUBwt7loX0YneDZmA0N5iVl6O42a6ueknsjGo+fcPYyjoYIXh2x V0RtZRPBXhX3cOKedp+LL7bauViw2xxo0fkaEU+FwnWi37auzphRbzrPd7Ub AjyrT2ZSE8rDQdm2CUWm2O9THjTJTrbyG4D0y+bquxR8l7WNdrRiGLdMhl0W OxolQpP0dEi2xTJycF/14KtFmj8UTEik5woMS68p95a1Gdpe8nf9wiPYZrGJ jFTESEWBRPr/VVLnrz0dv9KKLvGNxi5ToEpFrGswLq1Ec7o0apM4QfGjASkd tHkiCaBGA4QygjeGrFU2I6dSbr4qcTbLpZi6gEGyneozQCUd4WqP9GPXpK5u M91rmchppzHQYaSEDsYTG0/cuBm9/iYRe8MTHXfTBGFtC0i2e0a21HOLd7eO bG5RnurmOZCCiOl8BMuP2ctijmoX1nC2V7qdg4DQKhpwTaikkog+d+ILSM8z FH91SJfL2KpCllJadzKgOeLV5OoN3UMWHwiooYPQnCTVWSblpWap+IrYmcd3 rK0xLyk1vFSMvlqOrqhC19coqlvyTJ1LSO7iAH81ULWgpuKupoitLGIp9Joq aySKiyPm44OWXe0moHRer0VyOhQuayeo00RgJJEMJN+8mbzxxtg4NT7qHU/g 4OJJULw4JYiE7aFEwhO9LnXHtrZgy6rQnB6DEsfIkCUYttsDXoU3JPUwfJzu twQGLITEhk+qpMF4YCQe0sYjxmDILHESA9ZAnzU4YAsO4yGBnVBYdaQ+DxR4 SL0/KNlEC1/Hhndm1veuqlRtbkDrZGgwaIhFTBbSJXYGBE5G4Y0S8evB0TcR f7xE5jrSZz4/rOtT8Wm8KYZXhVRZQB+BwgzwVzHaw2H9CRBmcxfP+2mwMPnW t7712c9+9m6DgqcyD5UE+OEPf7joXWXEOgBTVRJoCxZN9spbfifY16RbtmyZ fpY5qaQ0gNz47W9/K5fLp1rm5+ezW2mann4ctgfOBz/4wTSnLqWSpje1d+Ph qKSysjJgP4PLnXqpDrym6SX/05/+dKoLd0+fOeWtzcZnvnz5MmuculnmdBmk dk+dK8UdfeC52k9nNj7zHXPKfrtJmxeFlQBpLs2FCxcWTQ4qmSEZ9wms8bvZ T+d9rPE5FeM8nlputzs1U8GiyS9TZ86cMZvNd8zpg+XJJ59cNDkO6I5bURRl k1RXV8fGvF8qCcBegevWrWNXH75KApXyiU98YvYd+N/T2Ru2bt0KmnXQpsxg kxqdfUcGBwfvtmOaf7W9zQDC3i5TvthVqvDWanzdZr/EHVB4A6ifsgUoB02J ce8loXV/j/H4kIXPqiS/z23poOV7KdkeVDcxv/SQPTTiZIyBBBZM4l4j+a5K yuMO7u5Q7+g0XBY7241kl4muUBCvlWNAO2yvGOrsytUP7TKMZOP2fpdX5fXJ KZzDGC9EjBeBezwxTCmKjjLaifEvxFDMNxDzdsa87UArhY0XgEoKKvdT9nZ/ 0GUPRkZsgY31pldK0X3tli4DiZKMye8nPCO0mxt0ckK2qrD5ath0GTjbcVdj 3N3qtA/2IFiV0l+jIjpRJ9fgaNfhxWL8KBc/xnVUKoh+a1Ds8KutCKk+GVLu 88kzDdJTGnGuUHS1jt95YhA9NWTqVsvt9iGvc8BgFVYorFelziYJ3yQ+RovW Am95X3Xf0mLNc4XG568ZX7g2Mc31gcoOU+cyivs0IdzhsPNtlN9CkTp/GAhM xGGzKAu8XU95Op/SS/JFJlMfFqxV+1+vNy0pRw+3q9QGXoLoZQie02dQexmV J+Rwq0PujqCz2WLpKZUYLwrtDWKhUXkJeOxO5fFSPv/0kP40z1qicHWaqG4s AH7bDCTHQPLxoDeaSF4fD4+N6ck4ULgqh8tiaA1Id3pF2+t7S/e0Cv9MJZkv TXzgsxZ7TFVFfPnxfsPxfmOvasRraweiNWy8RPa/6O9+ihhe7dFddVvb3ZZW t77YrbtiQ8qkuuE2vbtd71RpW7zKw37lPpMir1UmLpS6Low4X6/9k0oC8pb9 llTZnGfsfMXP/R3QC1HLtQlRjFdFrMWg4tzGitIR+alBYyYX29pqXFGNrq3V HelQuewDQEcnaFkySY5eHx+9nhwbJ8fHvONJx2hUnwiK4rQkGMaJaNzGjAkd kR0c86pa9AjXKDBbLD6bg8Itfi9K0HpfUOmmQKVLHB6t08r4eBPzkJCCIK0P hKzeAC52+gdtdL8t2M8KJatPatb6tadC6gMT/ycl3kCPrLEMb8+q566pVG1p 1NdLtEFKHQtpjR7bkM3PtdA8PGQOJu3hMbk3WqXynheAS87I1/JoazUoXkZz OChaTwvWBEUbGE1mGMkBYTZ38byfBgsQtmGa5dN1fiqJ7fgBPHav18v225+q kkALuOherF+/fvpZ5qSSwBnFkwBlRBDEHafJTY2SuONxysvL2a1p7kRKJU2d J3lm3muVBLwg4I0A4zvOWZFihhHrU2G9rwfrM6dGUmAYxsbM6TLYvXv33Wrq jj7wXO2nM2+fmZ1AIM1n3rt376LJl/mp9/8SiYQdL7N06dIZknGfwBq/m/10 3scan00xpp5C83tq3bhx4/z589/4xjemljZ4onI4nPf0X7ZXrFixaHJ2iztu TT0YU3+o8T6qJPaz14svvsiuPnyVFIlEfvKTn3zyk58EVTwb+3mopNHR0ezZ kZWV9aEPfQi07DNM8VRYWDiDX1RVVXW3HdP8K+CsgnBZ6NzdYd7cbDjQZWrR 4GSYiMR9kbAlGtJEaaXZpa2Wm8/wbJfFDokrYgAqyed0IaV+7rPursW9/efz h2W5AluB1DWMhyTuCGLXefQTPe5s8mMnO/u2tCg2c/SVKq/GHzcERvux8OoK 67NXTGsqDKe5mpIRVZlY06Sxc3SuXtSkRbtpNDdiOBtzNiQowcQ/kzJoLElF x+Kx0XAyZk3GDMmgJIZXMqq9jHpf1NkIDGjGpcHxzQ2GJaX67C5M7fDGk4Fo gvYytDMUdPltBNY88a860q0h1f7o5FTVKm3b8Xb5yirzziaTzKzz+LSBAOqj PQ4miTPj1tCYJThqoSMOP85Yy6Pmy6i6vGCg70jXSHa3/DzfUKFy1yLefguh cJEqt1+IE7Uab7nS0yEeNPZtJjt/ZWx75VBl28pSxcvF6CslKJBvq8pVmXVc oJ7okbUhVVbY0RqlpAG/Sop7Bq0BnsGESAq93Yt9PYstymItbhc7o606elWl +bkCw4F6iVzRFLWWeEzVXHn/sX798QFjP6r3+zUMrcFcSB7flNVnLhzsR/g5 tHCtR7S7WyWs19iqNd46ra9RR4BQj/hqtF4Q+jGvJwhEhJcMua8IsUNc49Ee WRuv2qc44JXvr++v3MMRTVVJYeNZBjka0maTujy1DRU7SIkrZCedEQadqAtn E9n7LNH2pK1/nUTTO4i5hiwuodUhsuGDJnuVwnpswHyiT9s9VIzzt/lH1ltE R7oUI1VqT6HUvanR9FoFKBz9y6UoCCvLFWUNZw1tL/q7fk31Px9UHwghRxj9 ycmBTgU4Wn68XbyhDllRZVhabni5BF1ZgWa1o26/aTRqGotZx0Zd42O+yUCM jxOsUBqLY4kY5qK9CMGIXZFucyCz17SzHT3cZyqS4jVqFziAyu1hot5Y3BuJ OsNhW4QxRYOauK8z4WmlPEMGt03upmUuethOD1rpPivNnZScPUaXQC/3SvdP 6CPR6zR/RYC3FBvccKSpO6NGuaNZ2yAcCTi7GHe32CAH5QBukBKFu1nvB0IV 7A50FhBNYpvThPWBa55BTzDoqTB6JoyenviDKtXuoGI7CLO6i+f7NFiAsF1T 3lOVBGBfUwOfB7gEi/5cJbGjIb72ta9p787UUcYp2KZn2bJlMyeGbb9m1gss wG1gje+od06ePLlociRC2p/L3I9Kcjgcs9yFZZYqSafTffjDH/74xz8OPLEZ zDQaTapwZih81mt6sD5zTk7OoslX3KmpruZ0GaT+mHL6fCN39IHnaj+dB+sz 37x584knnmAPCDzGL3/5y+wt89WvfvWef7hzP8Aav5v9dN7HGp/h4MeOHVv0 59+G5vfUSoHj+IULF372s5+lpv5LfT15L2AvIQAQHdO3prQtqCM2hlVJ3/72 t6cbz0Ylzdw6zKCSwC0AHraLJqcxZGNSKqm9vT3NeJYq6Z5N1fQ07N+/f9G0 aUzuxjxUErhmfjZr2DzOINnA4wX4P6ASp05mlaKtre1uO041A6sVKg8IQCht bTWurtXvaDU0qe1MfMLVBJ7nxLzctNjqUtYrTXkC+1Wp808qyetwaa4RXb9x tv+6s/vM2V7R0QHsvNDebw0KnWG1FaikqymVtHlSJdVofAY6aWXGebYoUEm/ v2JaUYkd6bVdFOH5YgdIQ7XWy9FiCl03rQcq6UzMUZsghxNBeSJsiI1Gotdv xd7403iTUUYNtjKqfUAlxZ31Y2EdE3ZoHXagkoAYOdKNIS7vG+PU6BjtjSYc 4VGc8nmx1oBsLy3KCMl3RSyFUXu5XNlysEn6/DVDRg2qNquDpCrJ6OJxX3T8 enj8pj9x3RkZdzBRT8AZttdELYVaTd3Z3pEdbert7egZnrVW621GiQErDVSh zBPhO0JAjIBcdIj7jdwMf9uPDa3PHaxsXVkqf6VED1K1pAxdXanKqu+18PcE xesZTXbc2Zik+CG/GPjJfZh/ADVqxAXe7t/7en5vUxXrXbjUHeOgwZUV2LNX jfvrJFJZbQS76kRLOcKuPR2avZ36Dq2eoNAwY8A8hjPDpoNc09W+Pg0vMyBY 4RZu71eLWhE7EEcNk/oIhFqtD4gmEPowj4d2jSedBI2fHkQ3tyK7W0UNA1U+ xUEQwEK6SjKcYbTZIfUBCjlldhn1/4+9+wBuY7vvxX/HL2NPZjxxJsnYcSZj O/NPnHgydsYvsT1+cZL3T17sOPk/27f4VpUr6YpqVL/qjaKo3imKYu+9gQ0k CkH0XhaLRe99sYtOUldXukX/A64vBoZEkdKVJVL5fWZHAxweQNuAPV/s7jnB mDGc8lOBVNI6Q6vT3sHw+C+CA/9onSiZ1vBRDBmzkFxHLgKMWaI1Cs/BMdOR EfXgZLWdvzUoeM8iPDammm7T+mvkKCWZVzXlthqKPL+uM65pUDZ0np+/KPGn kYlfxZS70V5E60/Eicu5lIQ1HO8Xr2/VvdmYS52/rjGuacKPDBG+iJ0Z5XZ2 xj2HgtKM9zcpCU0zHvSnTMrhIgOqACVw0ywicnzC/P4IfmzCjFJ/jcLbqfcq 3b5U2jeX9THjQM0kcbRcGf9Q2tsb8nK1bqfQTaOJ54hxcyeSSLRcuROjuJtv kPmlaL/aSArXRwWrovy3LJxNZT0jm9qVu3o13SJB1DVIeYaEmLROar4mct2Q ulu0frQhBozhKReF9hyNx2uzsmOG07ShDK3nBMqD5qtx/Dxa9tytcPLSh35y i1LSE38bLEPoqxgdtX+nV9zdn//h9KX5zrKYy1EKUxJznyw6JC1laMtCTPL6 +c9//uhq+Wbhom/I5/OZyg+9RQs1Wl767Z69GU+QkvLNS7VavcSXMJZ+xV0m k9m+ffsf/uEfPuJWKdReZQbyWEr39U+3zfzTn/6UaSLmSx5rN+jp6WH+rwdj 4EPbwI9b/0FPt808MDDwP/7H//jjP/7jn/zkJ1/+8pdRnkXtKNQAXrTrj08/ /VSpVIrF4nv37j265kPBFl+o/oOe4xZ/xJujr7uib6En+9Z6kNVqzfeh99AI w/ice6BcLmf+iwe77EBBEh2G0J+++93v5gs3bNjw0nw3fQ++FTNU60IpaSlH h0ekpPwQt/n72lBgZ0pqamqKKudHCV8oJS3xUFWksbERHUkLe0l9tN/pFXcd HR0om2/duvURddD6zLeCjh07VlFRgY5B+RKBQLDQC4vaV7fkHjRdE7u29hLv thl39uE9KjMVt8+mbbl7OmKSTHTa6pa3K/HzAvsNiVPuiZrDlNNv9mqqgqM/ 97B+Njl+5iZXcEWA18qsQldU4aM1DqNJ02iVnVKJL5UP87b3q3cMYh1atylC OmL0tC2yviV3xd2aZvPRMesFgf3StB3FtFq5t0NlEes5IVM1SdwM2vt9bqHX p/IF8VAyHpm5Q2ZnMxn3TMY6E1el3O209iClPUi7uukYFiDdCrujtBt/u9FY NmrReYMoIiWzMSKc0AVSOrfPbhoJqSuiigMx9bE4cQ3FDY2q7ezg9MYW3YEe jZ4Qkj5+MiyOxWyBVMabmrXGsoZw2hCMmX3WWG4U2it6dfNVjvDAqPbAuOmK yNlrCA6ZQnwnpfQnVf4UyoadukCz2j+MUhJ7c2joX0wDrxxvGyhpVq5pxte3 mUo6zDu79ecGeTbRoZislNIey910Exil/Fy508l3RAQEYZDd8o/+MjD2S6uq DoW+wnNJe9ulHHGvE6vXa9s6p9n7h3UHRo0svckXNscos9lnvi62VfBtDVNc gwClpNV+8Q6BQTqGu1mmEMvoZhksQ3pLv8HVg/l7sCDP5g2EzbMJYyiC3RAa j44ZT43JWIK2oOpwUHloiFdXPsQ+PDhVyR7P9XFnRUHpCm04SWkPRw3n7X6C iFBmMh2Mh9Ip+0xck0tJY/832P+/LOyNU2oeiiGj5si0Iyp2R/m2ULvGeY6L XZhUsHlVTv7mgGA9MX1kSD5Vr/RUStybO4l3UOSp++xc0m9S0qvMuSRKc5g2 lKPsELfcSNhrnXhL2ZBkY7vundzpJ+NrtfjqRpSSzL6Iezbrmcl40plc2Eln /DMzvtkZXy4uZV2zGVsmZfOQbr3fL3EFRk3eo2zTHpbxKJu4KnbdQnud3qtw e1Ip91xu73Jm0vZ00pKmtfMpqTvomZQ7bBw7icIR24oCYG4aN4cmzH4uYZdg ooBkLynaQAo3MP/ap3afHWTv6tEcHNL3SSVh9wTpnRDiyka5tVLiuilzN6py v0t06oPj1ijaf8R2t8nMJrGzNFYeN11CaztprUIPYvLtpHQzmpbyKX7ib4P7 n/so87uAmkwvFXTm8whPnJKQn/3sZy99pjAlTU9PM4Vr1qx5rNlmLt748z// 80fnO+bNl5KS8v1TodZC0WUnKpWKeZ+1a9cWveoJUlK+K6dbt24t8SWMx7ov iaZptCHQ6n1EnX/+539+aX4Yx0XvSnjiNnO+S7S8fEtp/fr1+cLH2g1sNhtT GTUMiv6Uv8SosA38uPUf9HTbzOhztFCz9hHQPonahMxsPG6LKw+2+PLf4gut xvy30Lp16/KFT/CtxePxHvqdme/yYqFjwVPZA5nI8JWvfKVwPGKEGYUcKRyg hzkFib7H0HdmYWUUVZie+hZKSUs5OuRT0oNfkkwa/b3f+738RdEff/wx8wtD UY8KFEXl0+VCKWmJh6oi6FD70B1sIb+7lISOcSjjo6N20Q5ZZHBw8BHX2KC9 caEXFv0cvWfYiKadg9jaNuM7zcZtndoOwVTU2Z/29qa9ffPjafYTlol6seYE x3JZQBgcqqBfFvOMR7VnQhMvB9m/csnKLHif2cqzu6SRuI9MRXRuZzVXdqSb e7h3at+Qbu+w8fCYkaWSupzcgGtCgfG3durebMRWt2KbenPDv5YOGkv7TNt6 iUPD5ja5xeSxEh5LrwI/PWI4NohdmDD3G4JTTkrqjoQjxmRMk4kIU67OBHYm ZjhnMo0ILeZxwtepde9lmbb0Gc9N2RVeKpLJ2mOpaom3fMJ5lmMdVBMmF+7z qf0OdlT+Pine5JcdwdStKh1bpxsKqM+SqqM+1WmpeqBd7UDt2Cqp+5LQcU2A twp47uld5PQ6s6hsUCZslttQFBoyRYRuWupNoCRlo2bQpA+n+/FQhz4wpuCZ J7eFhv+Pdfj1m2Oj5WP68kl7tdQ3hEfZpoDIZAgbLuXuN9GdoHXHaO1RErvo Dlrtsbgj5PUYW0KcNwOTb2DKeiFhHjeTbeogc1/S+lbd/l5Z2aCojCU5OaY9 yTGfmbL2YV6VL2wMRfUBUuympl20ipC7JUejvDciom0+n85L0Z4Y5fIoHFiT HWs02aV6X0QbTJt9jrC9L2mpomyNNo8G99kIj8Ft7Y3pj8U0B+ySQ1reYTX/ BC69FPePZyNTuY4HLVVx45m4+SYVd1EzWXrug9RMDAWTmTiW9rJCo/8R6P17 fHTtqJTdpgv0GX2WgDMYtYcjZrdPbbIJUBBwqy6Ep9/z8d/VcvdXsdmHxoj3 h4nVTfjrdfP3JTViq1qwja2K5p6zpqGXw2M/I6dWp1zdmSAbZcmMdzDt6fPZ WFe5ygPDhtI+/N1W02s1pnfqzYeHnP54eu7DWXoGbXEaBVsiFKHjjkzKPJsy 5e5rSxiycR0dlkS8PJeDI8MEB4Y1m/r0e0fw3OlRqbtd45LbLUkKVcOouCdA x3xULBh1xN29SWeL1cLqU+tvyNxVMneNwlun8LZr3AKz2eXX+n3SgH0oIiyJ 8N5EexStP50wVwVMjf0qbZ3M0aT0TBAec9BrCfqEDv8Qnjud16zxXRQ6zgps 5wT28wL7BYGjWmTkq9hR082E5UYSBefckvYnrXUocKHtiKaHfnKLUtITfxs8 laPMU8e0KN58881Fa36elKTRaF76TNF4ScwPhsw8FF5cZ7fbUXsg3xFxkdra WuZV77//fv7o8+GHH6JjX2E1ps5SUtL9gpujUaZj7s5AxwUUTP7kT/7kpfmO 8h68RfoJUhKaW6Zn4G984xuFnf16PB5m4JWFPPWewNEaZkbD/PrXv97R0ZE/ CKIIjxqTP/rRj/IDXT1xmxn5yU9+gmHY/fmVOTk5yQxh+eUvf7mon5PH2g2Y X5KRnTt3MhdxoVll7v54aBv4cesXebptZubG8NWrV+f70F6K/GgyjPymeSyw xZf/Fs+vxn/6p39ivh/QauRyucy3EMpERV1SPNa3FhMS/+Zv/mZoaKjwRHM2 m80PwrvQOD5PZQ/M90/+ta99bXR09Pbt2+hLj7m07KX57vgKO+VmsVhM+U9/ +lOmUxGXy7V///78PCyUkpZydMinJLRKL1++zFzEmMlkVq1axZTnB7BgMIOk f/GLXxwYGLg/PyISWsmFY4UvlJKWeKgqskx6Au/q6kIRaePGjY+OSPfn73HO t4LQKkKfr8J2EfrrQi8sal+h5iKaNvbo32nB3mzCtnaq23jsiLkpZW9IOVvS 7o60u5MwjdSKVMcmzBenjCarkPRyk64+SnMyPPlqeOLVqOo4aW+nfGN0UJBJ e7MzYZ3HUzGGr23SbGjV7h4y7h/BT7CxUSXfZ+0P27tVetb2bg3671a3Yeu6 9Bt69Ou7DSijrW0xbu8x10k9+iClD1A1Is/WDmJdM76711In9+fu47AGAkFd klRkwlMpd2cCv0AaL2mNY6OYpUvvr1d69o+ZdgwZUdNX5osH0nN4NHNqwrWt x7xn0NKi9BoCMS8Z9HmVYfG2CP+tiHRH1FQbdQ5FbZ2kYj8p3uwW7uUIm68J TeenHWUcK2rGHxvVVI6P2Sbfi3BeswgOTKim+3T2fmOQYyWV3pTGnzaTWXdy 1pWYNUbTA6ZQpyHAVqKUVBoe/nf7yJuNnLGLPOyKMDdSrcyT0PhIo5sgTVW5 G0/0Zbn7/WXbSPXRUIQIJJOBWCBkao9w3w5y3tIr6nkmAmWxRlWA6eNudYth fZt+Y7t2S5duHwuv4FsvCO09hqDETWkDSV0wiUXSWCRjcar9sqNR7q9J0VY6 Yoyn03SKjvmFUePlCHbZ65qyh8PW2IwrYI1am+JYeW5so6gynnDEaRPt6af1 xynNgbBsZ0C4JSAqDatOpML8LK1KhzlJW20cPx+33kqlfOkP5jIf3p2Zi89m fbNxPO0ZCY38NND9fePw6mHxeLPG32vwukJWmibStCEelcX8nJhnmNSfIac3 eLlrVZP7L4+M7BwylvYb327EX6vBX8v13oCtQam5TdHym5T0HyiZpv3s+SG0 hLms5GP5neOVfNWRUcOuAXw9Skm3TG/Xmg8NugPJ27fv3aNmPzBGUzIfrQmQ JGVNJ/CZJDaT0M/EtfOLkBswK+wYUGNj7w+p0V63i4WjDX1D6m7TuGQ2UyKm ztDaKOX2UAlXLO6LuOOu3qSjyWwabJWpc6Fm2n5N5LoucjXK7UICI8OyRGiK dvVEp9dHuK+hlJQgrqVc7RFH/wRm7Na7e7EAzxHFQnEsHBe6YqNEpB8LtWn9 Z/i2kxzLiUnzMbb5yDhxnqMdl49HidpcVxXu7nRgPO0fS9qbo4I1kcmX0bSU T/ETfxs8laPM7wJqHaG206KXvnyelITkh1wsSkmzs7P/+I//yPwJvfYv/uIv fvzjH//pn/4pU4IaVw/tKxgdZfLDXKKWwHe+853vfve7TAvQ7/fnqzEVlpiS UIzNz+RL81eqMKOuvDTfmXl/f/+DL3mClHS/YMhLdAD6q7/6q7/9279Fieml 3/6V+EFPPSXdn59/ZqUhzIVAaGa+9KUvMSXl5eVMtSduMzNDWyKozca0FZml fnAAjsfaDVBjibk+56X5n3z/6I/+iHn8rW99i/lpvagN/Lj1izzdNvPWrVvz +xhqK/4/89BugFb+L37xi1u3bj10fGe0g+W3C8I01Z4AbPFlvsULh6xF0FI8 ejU+1rcW+ubJr4ovfOELqJGPIu2//Mu/5LfasWPHFlonT2sPRKmBOS9TBGU3 j8dTtGh//dd/nd9DmLj90vxVeczeuFBKWsrRoXBUWWYn+eY3v8mcomL+i6L+ 5To6OvKV0epiOtVH9a9cucL8XwulpCUeqoosh5TEXCm6fv36RSMSgr4H8q0g tMNv2rSpsF2k1+sXemFR+2pznwFNJT2GVc3YW43Y1nZVG4cVwm4kiOtJe0Pa 3ZX29BLEeK1YfWzSfElgJGzTMR8n4e4Pai5Y2BvM4xucinNeS3fAzY4EhDNZ 79xcGPd5LnHwLe2a0i7tkTHiBMdylm+aVHP9lvawtVmt7dnRrVrVYljdani3 U7euKzes7aoW7J1mbEu3qVZsx3x+zBeoEjjWtRKv15k2dVpuiH29WGjY5LN5 9cGAigwIKUdPHL9CGq/qjeOTuLnf4G1SufaO4NsGjOenrFJ3JJCI46HYqUnb tl7TrkFzjcwvdCUM/qjJoQmJSqP8t2PS0oTpasrZmrTWUPJdpOg9z/SOKWHt LaHumtBcwSUOj5mYlGTnvBflvmYWHBiRT3eo7O3a4BAemXLQYlfcGE56Emlv MmUh6QnCyzK6BKoJK2crSkmOkTfbpsYrhcabMuew0W3w+cwBp8Orjpkq48Yz Ue1Jl+SgVbDXLjsVDBGRRDwa80aJ1gjvNymJbzKjYNiu8e8fNGzv1m7v0e/s xXb24e8P4uWTRKXYWiOzDeKBaTcl9yeVgSQWTuORjNWpCcqPk7w3Y+JtyYg2 nSHT6WDCP0ljpylDRdA55gm7XDHK59dHiJuU9hBtPJOh1NmMO5uypvysuPEU rTsSU+yIiTaQ4pKY+kgsKKFiGBUQ0NZqGqugiMog6fQn04HMHJ2hZjK+2YQp 7UMp6WeBnv9pGn5nQjrUrbWxDDaTB/cEjaGwLhaW5frT9o/EsfOUqMTHW6/i HLzAGtk2YNjci3Y542s1RiYlrW7FStrkLT1nTIO/Co3+LCpYnwoK0jSWJuUp 3whKED7bYCVfwaSkDe34G3X46gbiCMvlS3wwe/cjcuZDFBWl3oTKTwVIB0UR cRpP0MYUhaViulSIn/INRJ19WuPI3iHV+h7dnmFjpdTZpPb0Y26Vw0JF9fGY wRPxGENxQyBu9rtp9wBKPWZiuFmqOc23neHnLg29InTUya3ThJGMqJIRUdw9 gPJRbneS7cwN8uUbiHhGRw1Eu8bdoQtM2EhtKKUPp0We+IQ1xjJFUV4+w7ee 4JiPTxCH2WaUxM9O6sbk7KilIWWvy529DXFyHes5WsnHSUlP/G3wtI4yTx06 TKP5aW1tfXS1JaYk5gj4YErCcZwJUNu3by/6E/oSbmpqYroVyvuDP/iD9957 r+i4WSgUCv3rv/5r4UvQ+6OGR+F15szoSEtMSYyWlpY/+7M/y78neod/+7d/ czqdD62cvwPisVISglo7+ZYJA7Vedu58eP8hjN9FSkICgcCrr77KHPTzzafv fe97bW1t+d910dIxf8q3mZkbw/Mj7ygUCqZCPvsz11iiNnlNTc1Xv/rV/Jt/ +9vfXqhbicfaDSKRyL//+7/nWzVoZa5atSqdTv/gBz946WFdUT1u/UKPbjMz S/rKK688+KeHtpnRB+GhrcQ8tA8/9Iay+vr6v//7v2fqVFdXP2KGHw22+HLe 4kxKevnll9EmzueCR6/G+4/zrYUSGcrCaHMXHoxemh+XdtExcZ7WHsjlcr// /e/nh65DsQ4dWdCmebAmSrs/+tGP8jOJMuO6devQIoyOjr4036f9Qv/FokeH fEoaGRlBGzTffwX6XKC4ke9mJA/tq5cvX86vNLRf/fCHP2SxWOhPzEV3PB7v iWfmQcshJaHNVFpausRuD9Eae8Q1Ng/duIyi9tX+UROa9rJMa9uNb7cYt3bI W9ldAVV5XF+WRM22wEgmzCXsogap4cSk5YqAIJzyWEgc8XFVqu7OwcrW/us9 nK5BGXdcIxETqkQKpaSQL+KY1Ejb+NzWKUG91FwndzUpHVI9J0jUh/CbSmXr 9i4VykSoVby2Yz4ldenWtOnXtOq392iaBGKbbcJmnazlSDe06N+oM5V0WC4J 3C0af4fOKzCbZTZca1M5rMO0uQpNLtso5tTJ7OZhPVHag69pxctGMRmhi4Y1 Dq+2WmQ+NWEtYzvQO9TKfW1KB0sh9UoOkJLNMfkOSnOU1pdTuuPMSDdByU6j qlqASScNmlYFflVkvSrAWwQ8t3AXKVxnEJyo5fDPTJjKOY6LAtdNqbde4RPa fd6Yj0x4A6QTc2rRjOH6Lg9nQ2jo/3WOvDYonmyRW9rVjikDaoaPey1DfnNn zHCG1p+wK86McqpaRup7OR0OPx5L+OIxPIbXhCd/HZj4NaaoE5otXFuUbfL2 SKWdQkGXRNqtxHs0tj6tbQwjpkxGAYFPWd08W4Rvj005KIk3gQKCzqr2KivQ 3MbkO5N+bprC0jFNwtVFaY/k7iqytob94lBIF3GzScNZSrmL0p9IkvpUOpJK +ZPBqYT5esJ4jlLti0k2RsSbAsrjJqcK89qtjumA/nJMtdevPsk3KFmYZ9gU MgYC8ZRvNmlO+8dDo/8Z6P2Bc+wNnaJObpySYNPDGm2/2jCq1ettykRYlAnz kkQlJd/pF25VTx2/OMouHcS29BnfajC+eis3vV6XGzJpfZOiqaMC7/2vIOvf Qvz10ZA2GveREQ1ta47jF1y6GxfZ03uHtOiFG7uM77bimzqJcrbLGb+duH0v lLlDkFlNMKULJswhvy3kcYTdnog7EHMHSUcsJE35R0jXoB4f3zekeq9Hd5Rt HMRcQhRcPT5TwOONuNwRt9zlGzUFWbh/ymwn/bx0cNxin2qSGSp41gqe7Szf dnbKfkNiZ5sc3ogjHCUiPiGpPRZT7EI7Usrdkwnxgt7pZonxAt96UeDs0IUk noTcl0RbR+Shp5wUSr7nBY5yrhVNJzm56Qofm1RNkkxK8vXneiwJcVOuTlK4 IcJ7HU1L+RQ/8bfB/ad3lHm60DyjL3/Uonh0tSWmpM8jlUqhlgBqmD040MlC UHNOIBCIxWKDwfAUuwiLRqNyuVyj0Tz0TNZT8dFHHxEEwcy82+1e9GD0O0pJ DNQ8djgcbDYbw7DfxSLbbDYOh/PQUV0etPTdYHZ2Fm0mNM9LvNHvces/dajB /JWvfOWLX/xibW0tagFqtVrNPNTEQl8I+f6xF2oP0zTNtC0//4gDsMWfjcfd 4kxK+sUvfnF/vmVusVjQtl7oKrgij/Wthb5/0DYSiURqtXrpFzY8xT1wZmZG IpGg9bPoVx9KGXw+H23EpZzUKPSIo0M+JaE/3Z8/CKLdT6/XP3o/QSsWzQba sZ/gWPNYh6rlkJIe1/vvv8+0gvbu3Yv+PXLkCPP0wd9FCxU2n9DT03wbmsom bRu78NWtxm0dktbhOu/0NhQcEraaNCnKJvQWH9YgJ8o4livTZvSYpgy+kK5P oSntUL7XqtzRq90/rD85YayREpGEb24umEkQlG88Yuv02QbFVhvHGuJYA0Zi MmysDOouKKU127oUbzVj77Ri73bo1nflpnWduWlPn7ST2x3QnvZrzjSN9ZY0 Kd+ox99rN1dwnTdluV4mmlSeFrWnT2NRGnmUtSFpr0/6hhMRSSiolhOaDU34 r6pMBzqVCs1w0tUZsg8MaYxNClel2FM2YdveZ9rRoz3L4jsUFbT2EKXaQ4pL otNrSeG7pHhjTFwSU+yhTJWUnxvxi+RWQ6/e1aFxjKplftUxFCjkgnMnernv tWlXtRrXdRhLeoy7BrEBDeYLGTK0Ph1TJvzsuHeUMt0Mc98KDv7ENfoyV8kb 1tvHMatGzwpry2OaIzHNYSrX6cQBpfjS6YHxDW3qff0agwunaCJDSuPGKxH2 y8HxV0zKWpndJvGQarc76JqIuoZQi52iTHTKSyfdcdocj2rIiFbnsrGJwBAe HjKFR+d7FRDjKqfqQkxWGlPtS7k60v6RtHcwbq7MXd2n2E3jl5LOtqSnL26t pVT7c0utPEBGiVg6FUsn4hF1Mndm7RaNmv3STWHxZqesbBLDR0zBab3YISuL CtfaBNsvjoyhlLF/FB8x2kO0ZzZlSQcmwuO/DPb/r9zll7rTMVuT3dRxfkyy u197eNgwqNZRUXU2Jk3aG2jtkYBsn0Z09tLY5PZBbOtnKemVW/irNcZXa42r 6hR1rSexrp/5B/7Fz9ngCdnd8Yw3YIjoL5LSLTbhvvIB9pZu1fourKQX2zqA 7R01XZh2WMnZSOaOP/mBjcqaohljJC330hI3JXXHNP64KZI0R6hA0JDwT5Lu Eb1pcj9LtbFXd5aP67yuaNwbTwX8NImq6QKJfix4RYhSsKVZQYTCmiyttnp1 6PF8SrKe5FrKuJaLQvsgHjRHU84Y5Q1ZafOtBH4xaatH6SZLqX1o32UbtvVg 2/uJayIP154bR0zuT2hDaXUgxXdQl0XOCvShm5q/L2naUS3Bp7Rc0tqUcjSm /UOZ6HQmMpXy9MWkW3Kd5glWPfSTW5SSnvjbgPEUjzJP0euvv/6FL3wh3wXr Qz2DlAQe4XFT0u///u8/9CpB8Bxt2rQJffa3bNny0L/m+zRgfqN+EHOPA2py L7HZDJ67x93ij7icbzl4YfbAR4+X9Nz9+Mc/XkpzIu+5p6S5ubmFfjres2fP I15Y1L46J7CjqZxrK+k2oZRUmktJ9T7hdkq2PWFvyFCybMpkCxJtKuu5KdtN idXux+KUwR/S9yu1pZ3q99o0O/oN+0dNZRxztdTmp3ypbCAVN6Z8Q6jVR9q7 pDYLxxrkWAJG03gYuxzUnlZKqrZ1KN5qwt5pwda269d1MhOTkiRd3M6Qtjyo OdUx0bOzQ7WuldjabT016bgmcl8XuW/JvbUKb7vKIjXwYubapLUm5WOlouJw WK2YT0m/vIFSkkKhHky5WiOOntwdIjp3vcJTPmnb3INv7tSUD/EdqjO0/iil 2U9Kt0ZF75GijehBTLaNUu1LWG6kQ5PxkFDnNIyaXIOYg2uQBVRlKFLJBOeP dHPWNmvebET5zrC2Q7+lV9+j1Pj8yixaS6QoExjN+FkJ4kaU9yaTkngq3ojB zjZatQZWRFc+f0IHTYfQJBdfOdHHQ2tgR69R78QoCstGhXHscoT9SnD8VUJZ o7Sb5d6wweukfJMp/0gqNJVMmJNpXyrtSSfMWVqTorS4x8Ixe1l4gIUHh4kw iwhPG5VOzVVKuYfWHk4529K+obS3L2GuojSHYuqDcdPllLM5d0uX5SZa2Fz/ 1cqDKCWR6TRKSXREk0tJthpadyIm2xqWbHPKKyZx8wh6W4PEIT0RnV5tm9p2 YXjo/QHJvmHtMGYN0b7ZlC0d5IZRuBv4SWTyVVpfgaKrC289NSLa3KXZ3a/v kGn9QQ0VlqFUS2keSEmN2Ks12Cs1uaD0Sg3+Tq2ytuUU1v1z/+D/9nHXuwOE i6K9PnVYd5aUlFin957sH9/cqVrfiW3uw7YPYfvHTBeFDjyS9SZuO+lZgsxi kbQhnJJ6YyJ3VOwmUVxS+1PaQNwdwGk/P+pmGwjecbZ2x5DhqhA3+e10wolW qZ+OGMMpVLMXC10ROS5NW1qURDiim4lr7X5Dp5q4OG29MG09zc9lpStC5wAe xsJpS5R2Bu2UpSFBXE06GjNBTjam8PqUp8Z1m7sM23pNV4VuriMm8tBKf0If SmsCqSknfU3iPDt/lxOKS1fFznqZaVrHI22NKUdD2jeYjvJzt1B5emKyHVHh ejQt5VP8xN8GjOV5lEHNbzRXly9ffkQdSEnP1+OmpO9973uwvZYbpjP8hW5A yw8/+tBT0h6Ph7k47bn/aAyW7nG3+HJOSS/SHricU5LJZEIzduPGjaW/5Lmn JJfLhZpARTcg5D3inHJR++q62IWmC1POki7TqtwVd6o2znAIq0IN6WSQl0o6 U9lwIB5R+WJ8ByVxhUNhfZJUkAGhSjfcOdbUMtrSJRL1611DeHjMHJJ5/Cqv 3+hQBo3VtPaYX3N+SClvVtiaFVaxrN0r2e8T7pBxT25rk6Lm8dsoKDUbVrUY 5u9R0qNpZ4+yUzAetjZT9lY1Lh7UeluV0RpJqIztfH/IemDEdjZ3b4ijVqTj y7pDyiMx9ZGQtdsb0Jn9Di5uW9ds+tVN/EC3UqkdTXl6aM+QyW3Uup0Cq7te ajs4gr8/pL3IFri0V+PG0zR+JWZtIW29MWs7ZTibG5QHq0g66tOhiWR4yhPU Gn0OvcdGOCQRDUpJe+WCC8f7OBvaNG+3YKtRSmozbO7SdQs5bnN32t2e9nSh SJLxDyeIKpK/KjT0v12jrw5LOR0qa6/OKTdroh4O7R2PuwcTlqqE+ZpK2XyS JVzdZtjVn7smLRaRZ4KTCexClPtGmPOGT33RYeM6PCqXVxONGMio0RMi1G6H 0OFWuFyOkC2dMGUSpkAYJzy4zmlCUbRb723T+tkGnQPvTJivJ603076BdGAs FWDHfWOkeyTqGqbd/Ul3N0pJKF3S+mPzV9ydjJHWWDoTS9F0UJIw30jgF+OG chqlOV1Z2HiD8FmMwaDFKQmqT5HiDX7x9klRy4BoeFDK0btMVIbMzoTTEXmU vzo89vOo4N0Efjnl6nCbe8tYso0dupIuw6lxfadCO6SSq2S3/OJdvuktKv7x CyNj2wYMW/rQps/dDZfr1rvG+Ksq/K1qdW3bBXzg16HRn4e4awJEa8A2FDK3 kLoKSr3fJj15rIe7vkWzuiXXKeKRcfMprvWy0MkyRSasMbaVHLfExizkpDWM 5s3sMRgc2LjeWi3y3hL7eLjd4SN8QdzqNY3jtkGDTUBgPu90PMChgwK7D5d6 ckN9TTkojj3Gs0fkbh8VUWVjslBIrXYSfIudY3Z0G3xtOn+HPoj+R1RT5Ayr HUTU2oTCdW4QLntD0tHsMnWeGhGjhLhjwFSv8GORjJWaQf+qgymZLzHlpJo0 fpT0G9X+bn2w3xgeMzp1xBRlqUlaqlKORrR1Uu4OlLloQwWd6wjxxFI+xU/8 bXB/GR9lPv74469//evf//73H1EHUtLzhVLS0aNHR+cVdo63EOZ24+985ztb t2593BFswe/I+fPnmbbZxo0btVrt7OzsJ598EolEhoeH/+M//oP5E6rz4Atv 3rzJ3KXyd3/3d0x/X2BFeNwtvmxT0gu2By7PlIT2ij179nz1q1/99re/fffu 3UdX/vTTT8fHx5kjwg9+8IPn26jAMIxpApWWlqJ/Gxoa0N5++PBhpvARB6yi 9lWNwoumq+LfjFyztV3XPi2OusfTwckUbU5kE4nZOXr2NjVzh5z5kMpkkjFt OipK+MdCeI1ddMAqOmIzj9nCIUtsRhtKD+HBbr1/Ui2ySctJ4XqXcFcDn3ue h53nGca5Nxycde6JNyUju7a2ilDbGE1vNvxmYp6Wduk6RNKol5MO8WOULZhM 26jbQmfq2IhrVaNpXavp8KilbNJ6lasc59X4+evD0xschlbMbZW7yUFD8N1W 88u38IN9KpWek/az0gE2TWFUwuaMOIYx6xkecZytv8kXe7Dq3EipznYqYogl wjHKEXe0JYjKpLky5WpHKSkd5idIDU1bKcpMhcQoSlDq9xXCC2X93JJ2zerW XEpa02rY1K7p4PQ4NJfi+JmE+WraM5D2jybNt6JTa8PD/8c5+kavcKJOSrSo vSJnMJok46lwIm5NewdQqtIZhk6NSt7t0O8Z0OosIjIgyHWegJ0nBauiU++Q mrKYrYVys0jfVIAK++MJfZDsM/huyZ3tWqfc6UolXbNpZ4o2UlFlKCgn3Jo2 teOW3D2oNTrsnNx/4RtMB9np4EQqLKBJQzQeitJBOixPegdzKclWRxsrKO2h mPF8jLKTGZSSYjE/nzaeoQ0naOPpOMpKxA20WuIJSzztp8OSmOEMKd0cle/0 Yjc9phavpRfFt9RsMjObSsWwmHBLlPNaTLw1aa1G6cxjGz7BUm5o1+c6e+/W 7xnSHRmWD7Ir7ZxNXu67yon951jDW/pznSuuazesbc31HPJ6Lf6LG6Y3qjS1 HddNo+9GJl+J8N+Oyt+Pqg6T6qOU4SSNnbKpLx/umlrToF3ViO8aMFfw7JeE jisixw2pq0qWG4qoTumtV/o6NU6LQxT2cux2XpdIsa2D2NppbZb7tT7SFaPc MTJAB4K0LxLRU66ehK0uYm012YQ8W4hnp3ShlDM+F0hlwwkyRcoz0ekkKY/F TBHK4SddCm9k2pUbWHbEEu01hgaMXi5uRC9P2m4lLNdRyqb1x+2qc+VD3M1d qt1DRKc+FMrcoWfvoqCk8CeF7jjfQfUbQz25wbZyOUviSSjdfptdQBNVCSI3 nmzCcg29Va5XcOstFGbRtJRP8RN/Gyzzowzzq6bFYlmoAqSk5+sv//Iv//wz u3fvXspLCILYu3fvj370o0cP4wWemXv37qEPUdF93PnHX/va19C3xIOvQs2h b37zm1/60pdOnTq1fIZaA0vxuFt8eaakF28PXJ4padeuXf/5n/957NixpdzQ 9+GHH37jG9/IHxSOHz/+DOZwIW63m2kCMfcjMGOmc7lcpvARI/cV/siMntbI PWi6JnJv6SJW58ZLMrSL1ZGAJEXKkklPYmYmMXc3efte9sOPZu5+nP1gLkNp MmFByjcUN10lpVtJ6Q7S3h+lvYFkiojE23XBOqVvUC40i06Q02udgh1VE+yy cU3ZuJo1ftXGXu0ae03MKt3WKny7CXu7yTgfkYz5aVunvkOsjPpF6agkmXTH Z2YCqQ/lnsyREdfrdfiqJnzfiPnYpOUSRz7GqfRNvhnkvmXSNMpsdq6d7tBG 1rYQr9QYD/epVUZ+JjieCXFmE9hs2hKO2Tgmy8Up4tQkViOQ+UyNuZFS3f10 zEqlElTcH3d2JcxVcXMV5eiM+jhRv4CKahJxSzJOJMJCSneCUu1RCc+dGxzd 2SUt6VCtb9euzaUkdcdEm11eHtcdSRjPpL39mcBY0lJDCtaHR37mGHu7Y4pd JcRr5R6OPeZOzgVT2Ug8mPKNpD3dOmzkzLhsQ5dh35BOZ54m/byUbzBuPEdO r4kKVse0J2hbI+3ui/k5ATrqS6TVfrpR7TkrsN+U2QVWZzLhnks7Z2hdhhTT IaHdLWuQW6+JnT1q3O4UZYKTmSAnE+JmwrxUVByPO6KZ2Uh6BsWlpG8od6rC Xk+ZLsUMp2LEdYp2UJk0lSIpH4c2lNO6I6i1nzBdSlqqkigzJi2ZrD8ZlZHY lbBsT0R5MGG9lRsP19ObpvSZOZSSkqmYMSYpjfLfImU7KUsD7R112NgVI/Kt ndoN7bpNXfptffq9/YqekevWyfc8nFUK9t5zLBaTktZ36t/NhSnD63XYL27g v0lJY+sinFeivDdI8aaYdGtMuSc3whR+zq65fqhrenWD9p0GfFe/5QzfcVnk zE1C9ADFJed1iatS4m6UW+02Du1huS0jrXzhqnrDOw1EtTio9KEElPUlUtmZ 0NxsIBvXogVJmK+HTdV6gjNG+McspC6UDqRuU9k5Ok2lSUUmMp0hZTNxw0zS nEjY8FBU7Y9LPPExSzR3UknrGtMbwta2XDfp5utR3cmw+pBZdurkIAelpD0s os8Qomc/SH/woZ3KynzxKRfFdZBDeHjAGBozR8XuuMqf0nlDLqcobq5OEtcS xJW46SKaUFxK2RtTjmY0PfSTW5SSnuzbYPkfZZgxxA8dOrRQBUhJADwV6Dvk +vXr//Vf//X3815++WUUe7u6uh7x6zGbzbZarc9yJsFTtPQtLpFI+vv7HzGi xPPygu2BiUQCreeBgYFFT9mApUBhs7DJFA6HZ2ZmTp8+zTxtbm5e6IVF7SvU tEbTBYFjex/xXge+p8/YIdP7A9pYVB9PBlOzc6nb91BKis/dpdE0k0mRqnSI n/IOxk2XSekWUrqNtDSTYW2QdBI+V5PKUyX19MokZvllSrHHKTt+nc07MqI7 MqwZGK+0Ta53T74lHd29s0O0phVb02pc3YKtbjGuas71b/YGSkkd+nahLOrl JUNTnqDZGKBl7vSwkTrAcqCItK4NPzxmKeNYr3Ll45wbvok3A5y31NK6MZ2p WxeulvjfbTX9us54dFCrISTZqCBLCmdo9UxCF4oYJnH8koCo4OC1IpXf1p+7 N8czSPmFZEAZ84lo8604fok0XnHiHVp8SmMSm916d9jqCxMBrzCqPhyTlxqm j9ePdpwdHDnax93ZJVnXpt/UoW5nN9ukR2nNvrihPO0dyATG58cDLQmP/392 9ppGLjs3XtI0M15SUuOjcK+TdvamnM16Q//FSenWfuzIqN5gKUhJwjWR6TVB zWm/pcfvZPs904GoIxjzqt3uGpn9BMd6ZdrGNdvjCcds2jZDKVEUonwTVoeg Wkqcn7a3q3CrS5qJTGUiaPFFaEpTSjrhjmTmwukZKqpLevpSzpa4vSlC1AXx mpClnYq74tl4PBWkfWway6UkWnec1p+ksXOUpS5KWSPJYCCkcxmbbIpzTs3l mKMv6R1NBzhpypjJhDKpQCqipGQ7SMGqsGyvl+hwOfmYeerm5HQFS3hiSFw+ qjw9qb/EUbC5lQ7eRg93tWLi/bMs1uZ+fclnKWlNq+GNBuxX1fhbt9R1nddM I2vDE7+KcF8npZvRmqdUe9HqTeDnHZrKY73T7zZrVzfjO/rNJyZtZ6ZyHSDc RElf4bkpd1dKXPN3+hBWy0jM2e0y9bVy+UxKuiUOqv1pd2LGn0xnZoKzs/5s PHcfFkpJQbxaZZzsM3h6sfCUk9aH0xYy5SIjCVKVISVoykZFaEpG5fawzxQi 1X5y0hJu1fjb1K4RnSFkbU3aa0lztUt33aK6opRWHxsUbOpS7xky9em9dDaZ nkubo3GBkxq3kCPmaI8h0KUPDOKh3CWsnrjSE7C7ZHFHe9LRmLDeiBMoJV1I EJeTtlsofKFpKZ/iJ/42WP5HmX/4h3/41re+tdBfISUBAAAAoAhzJ0JeWVkZ c7ENo7e3d6EXFv0c/f4IjqZ9I/gRtuXYhPXclKNb79f6Y0SYDqVns3c+un3v k/jte3Zq1hDOEGGSDIiS/vGUqzNhPBuTvEeK18fUB2LG8yG8CtO3XpnSneRa GqQ6wspN+Yc99rEz48ptvfpt3arO8Trb9Ps+0XbV1MlDQ+Kt/YZtAxiaSgeN W/qxVU3GN+qMW1o17Txe1N5DOnr7RdMH+o2bOqxbuiw7Bk27WWg+Taf5tly3 YNPKSe5179ir7tFXu4euHOoTb+nBS+b76HunGTs5Tug9jtkZz0zGlgnzMoGR oGuMq5NcFeDnp6zNCluUss9lHGn/REyyJzT6yzD7NVK6lVLtdshO9E607ulU 7eg0nONYW1Tufp11UiXwTm3Odcgw9a5TtNcmPqSaPt3Abt/Sq9zaI+8Yu2nj b4+JN9LKfSlvfyY0kXS20Mq95PRqG7+0om9sW6dqY6dxS49p5wBxYMhwZVzk 0l6lsdNGTW3jtOQU13p9mjA7RFSAP5+SLpDCdaHpDQZVgwDX88xOmQULO/rj zg6NfujMuHxjN3ZgGOtXY1QMy8Z1uduOnE1Rc71B33ViXLWLZbwmxAi3MXcG LYnPpu25800ZbyITC2fmQqlMLIha400JS2XY1mVx6/RevyngpxP+TNo/kyBS 3qE4SknaQzEUfqdWhabedUuOyGwWoYvk24NjmJ2lIyZxq4+OxbOpzGwik7Cm SWk6xEm6eyjlHpRo7IoKjlrQq3OwdLjB0GczNLnwVq9zwh8yBIKaMH4zKt7s 5b+rnNx/ljW8uU9f0pvruGMtSklt+lUthrebsQ0tyqbu86bBV0Mj/x6Z+GVM sZNS76O1R+LGMwnTJaeuqnxoelOn5t12w8ZuA9qF9rCMJyctbAspcsVHiOgN ibuCZ7suMJh0DRHjFbumsml88O063dv1RK0kjIVngpm5SCadTXtnM65sTJkL JsQVr6FyUs66MJUbq/e8wH5F6KyWuHq0jmiMmE2ashFBwlJF60/QpktJUpNK uoN0UOjw3pS5bkqs3QpNwNKetNU7id4JrbRdiddLjXtZ+k29+n3D+j4NTtHm RMKidDm7dd4Gla9O6UNprmr+4sAuLDhgCrMtAcxnT9L6LK1MeXsTxLXcoMPG s7mrH+enh35ySx52X9ITfBssf1VVVT/84Q8X+iukJAAAAAAUuXPnzvbt21ET aMeOHUxbqLy8vLa2lnmsUqkWemFR+2rHEIam3cPG45NmFHAuCR3dWEjlT2Hh TCD9QfbDjz74CKWku2ZyRh1I6f3RqF+Q9I+kXO1xw2lStAG16inUlNUdDenP YKrqi1zNsQlzncRAOCSZyJTHM10xrt7SrdvcqewYb7CKDvglu9XTFUeHJaWD hu2D2E4Whtr2KCgxKWlzq6aVx4vYutDUwuVtbNG+Vou/04SX9uP7RvGD46az U/ZLQmfNtIrDq/SOv+Iafbml79KO9unVrbkxalfNT+UThMGXG7ZpNutCsSXt 6w84Bjla4TWB8YLA1qJyksng7Tm0eBxyan2w94ehgX8iResoealVcrR5uHlt o/rtRuOhEdsNiadVaR2STbl5GyOc16L8t0nplph8h01ypGuycVuvYluPvHP0 hpW7JbcS5Hs+S0mttGo/KrFO7TjePbax9Tfj567JnXtSnRzk2lXnUZMbV99s EYnPTtluis0WZ1FKek+rbJkwmscIv9BiitrakvZatbbr1IhobYd+z6C+R6WP kahRrUn7h9GfwqabWm3rkRHltgHD5SkM95hmk8Rsyjybdc9lfTMzwUSW/iwl SeP2hoT5WsjWbfLa0AY1hBJU0p/NeGYTxrR3II6dpLUHSfHGKPe1IOdNp3Cf gDBz7LExS7TfGO7SB4dNEU/ydnzuXuaDO5mkPRMWpP2slKuVUu2NybZa5OdG 1ZJmjb9Xa7WbuklzTdJenwmMzSSxmbg2abkVk2zx8tcpJg+glLRpPiW9y6Sk 3EV3uaC0sUXe3H3GNPhyaPjfIuz/m0tJmv207iiTkly6qlMsweZuNXrVhi59 SY8O7UVov+U7KKU/OWElb0jd5VzrVb4O19REDOftqitNo/1v1+rebiDqpGE8 PIPWA4lSElrejDMbUzApyaOvHJcOVXCw4xPmsknLSY4VpekmhTUSs8ymLNnI VBy/SKn20PqyLKVA8YpMBEQO7w2ps1Js6VCoA+Z2tJh2YmBIq69VeCsl7t0s Y0kuJen61AaUZxO0UWa3tmtct+Seapn7usSFJvS4TRfowUIjRMDgdaUSFrSW 0r7BXErCzsSx07nzeobctJRP8RN/Gyx/iURifHx8ob9CSgIAAADAgyQSCdMK Onz4MGoRnT17Nv/r8SN69IWUBCkJUtILlpKe+NtgpYOUBAAAAIAin3zySXV1 9dGjR/OjSeZ/Sd67d+8jXvhg+woAsLIUfYqf+NtgpYOUBAAAAIAHHT9+HLWC uru70+l0IBDwer2HDh1CJRKJ5BGvgpQEwEr34Kf4yb4NVjpISQAAAAB4kFgs Zn40PnnyZHV19c6dO9Fj9ODRr3qwffXfLTfB8r7YXqTlXWhZHix/sm+DlQ5S EgAAAAAeisPhMHdtM2pra+fm5h79EkhJsLwvthdpeZeeku4/0bfBSgcpCQAA AAALuXfvXjgcDoVCS2wRQUqC5X2xvUjL+1gp6f7jfxusdJCSAAAAAPC0lAAA Vr7n/UWyLEBKAgAAAAAAAIBCkJIAAAAAAAAAoBCkJAAAAAAAAAAotGnTpnPn zj3vuQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAALAmPx7tx48bc3NzznpHPhabp0dHRhoaG 2tra4eHhYDBYVAEtIJvNrqura25uFovFs7Ozj1th2ZJKpfX19U6ns6jcYrF0 dnZWV1ezWCy73f7gCxetsKx88sknarW6t7cXzTD61+VyFVVY/ps4m81yFhaN RvM1SZJEJWhWGxsb0Sf0zp07RW+VyWTGx8fr56EH6OmzXRQAAAAAgBeZVqst mZdMJp/3vDw51OLdsmVLSYFNmzah9nC+AmoP7927t7DCkSNH4vH40issWz6f b/PmzWiGUVYqLEct55LfVrhCllJhWZmZmbl8+XLRDKNcnK+wIjaxx+MpWZhM JmOqKZXKbdu2Ff5p//79qVQq/z42m23Hjh2FFXbu3IkKn+WyAAAAAAC8kD79 9FORSJRvjK3clBSJRJiIhFrRarXaYDBUVVUxC+VwOJg69fX16Onp06ddLhdq TJaVlaGnZ86cyb/JohWWp7t37x49epRZ2MKUFA6H0TpBUZHL5QaDwbGxMaYO QRBLrLDcNDQ0oNlDMUehUFgsllu3bjEzHAgEmAorYhOjpNP7gIMHD6I52bp1 aygUuj9/FgltF1TS2NiIdmAej7dv3z70FO3VzJt8/PHHTElFRYVWq9XpdGih 0FNUeO/evWe5OAAAAAAALxiapi9evFj4W/TKTUmjo6No/vfs2YNaj0zJJ598 cvjwYVTY09ODntrtdvR48+bN2WyWqYAeMMHK5/MtpcKy1d7ent+ChSmJ2bjd 3d35EtQaRyWVlZVLrLCsxONxJjhYrdZ8IRNzmNNJK3cTo8/d9u3b0WwolUqm pKWlBT29dOkS2o2ZEr1ez8QoJgT5/X5mi6Ooy1SIRCJMicfjeS5LAQAAAADw YmCutkLtrunpaaatuHJTEmre79+/H/1bWNjf35//+X1oaAg9vnHjRmGFmzdv okJUbSkVliej0Yjm8ODBgxcuXChMSXfv3mUyBWpO5ysHg8GS+QsRP/jgg0Ur POslWQyLxWLOrRQWooxgsViY/XaFbuJPP/30ypUraB5qa2uZErRpmM8jTdP5 aiguYRjmcrlQ/fvz11gymSh/LxLaZMyrClMkAAAAAAB4XEKhsL29nbkpA2Wl FZ2SHurcuXNooVDbGD2urKxEj1EwLKzAZrNR4fXr15dSYRnKZrN79uxBc2i3 21HjvzAlMa1olHeYRnUe05D2er2LVnhmS7FEzMVySqUyFouhXI92XR6Phx7n K6zQTSwSidAM7NixI51OMyVooVDJ4cOH7927h+P4wMAACnE6nS5/Xun+fGhi Nj1aD3Nzc3fu3GHOA27fvh2uuAMAAAAAeFpW+rmkBzGNT5QCmAY/k5hQ4YN1 mNtSFq2wDDHJqK+vL/84n5IsFgtzC0/RS5i+CwiCWLTCM5j/x8JcH9jZ2VnY RwdK9/nuDlbiJv7oo4+YsDMyMpIvZK4MrKioKLog9sKFCzMzM/lqaAsyr906 Dz0oLS0t6r4DAAAAAAB8Hi9YStLr9czlZPlr8FCbEz1VKBSF1TQaDSosKytb SoXlhmneo3ljzh0UpSQcx9HTQ4cOFb3q+PHjqBytn0UrPINFeCzMXWYl8129 CQQCFI7Onj2LnqJdl7k3ZyVuYp1Ox9wqlT+RdH++d7t8MmpsbFSr1SwWi8lB 9fX1+WpoWxd1grdv3z64KQkAAAAA4Cl6kVISig9MRKqsrPzoo4+YQqZFnb87 noEazMyP9kupsKyQJFlaWoq2Wn5MqKKUZDabmau2il6IYhEqRxFp0Qq/60V4 XCgclcx3dp2/E+fjjz9mzny1tLTcX5mb+OrVq+h/r6mpKSxEG5FJPQMDA/lC FAyZQuZD6nK5mKe9vb3JecxdeGjPz3fqCAAAAAAAPqcXJiUxd+gjdXV1hTdo ML1GT01NFVbm8/kl852HL6XCssJcPFZeXq74DNPbW1NTE3qcSqXC4TCTKYpe iEpK5ntHX7TCs1qUpWIyTmdnZ2EheooK0Z/ur8BNHI/HmX21aDBfgiCY8sLr 67LZLJP9mYshmUyU7/CB0dzc/OAqAgAAAAAAT+wFSEmffvop02ZmfoQv6pSg q6ur6Md5BD0t+azbtEUrLCtFQ6MW0ev1s7OzzOO7d+/mX/Xhhx8yhTRNL1rh eSzWozAZpygCqNVqVHjy5Mn7K3ATM93XnzhxoqicCbDInTt3CsuZUbGYiyFR QEaPxWJxYQUMw0rmx8n9Hc84AAAAAMB/Fy9ASmpra2OuOCpqOjKYu3iOHTtW WMg0O6enp5dSYVkZGRnp/m0HDhxAs3rx4kX0OBKJ3P8sSWk0mvyrmEyxe/du 5umiFZYVpidw5rRRHloPqBBt+vsrcBMzF0mi5SoqR+GIuQupsA8NVFhaWooK mTuYmM7Di16L9nzmDOMzmHkAAAAAgP8OVnpKYu6yQWQy2ce/jek/GTUymWvJ 8jfvy+XykvmBaGdnZ5dSYZkrui/p/mfjYZ06der27dvoKfqXOQExNja2xArL CkoHTHbg8XhMSSwWYzYZc6vRitvETH8UKJk++CfmtNehQ4eYi+7y50nz95Ex 56HQ4gQCAaYErZ+DBw+WfDaMMgAAAAAA+PxWeko6efLkQpefXbt2jakzPDzM nGy6Po+5y6OwjbpoheXswZSEWv5MKECN5/r6eqYJXVZWhpLjEissN4ODg8w2 PX36dF1d3fbt20vmRw3ODyS0gjZxflRfpoO+IuiTyPTyvWvXrpqaGrS8JfM9 fufvF0MvZ86Cbdu2DS0I2nzMpty/f//c3NyzXRQAAAAAgBdW4cU8Kw5qJG/e vHnRlPTpp5+Oj4/nO09Gbc6iC5YWrbCc3bx5E82zXC4vLIzFYqdOncqvivLy 8qI2+aIVlhuRSLR7925mblEuaGhoKLyvagVtYpIkS+b7AM93w1gklUpdvnyZ SVLMpjEajYUVstksCkf5Ckh1dTV61TOZfQAAAAAA8EJBkSoajaIsUHRr/NIr rDizs7OBQCCRSBT1aLH0CstNPB6PRCL5U0hFXqRNjDJgMBhEgWihCvfu3QvP K0yLAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAIDnKJVKVVVVTU9PP+8Z+Vw++eQTlUrV1dV1 8+bNjo4OsVj80UcfFdWxWCydnZ3V1dUsFstutz/4JnNzc2w2u66urrm5Gb3D 7OzsM5n3zwvNZ2NjY09PT1H5UhZn0XWyrJAkyeFw0BKh5eXxeHfu3CmqsPw3 scFg4CxAKpUW1lx0YfV6PbPDd3d3o7d9hgsBAAAAAPCCu3v3bkVFRUlJSUtL y/Oelyd3+/bt8vLykt929OjRZDKZrzM+Pl5UAbWWC98ENZj37t1bWOHIkSPx ePyZL81jQ6EAze3+/fsLC5eyOIuuk2VFqVRu27atcG7RIqOMn6+wIjbxtWvX ShZw8ODBfLVHL+wnn3yCwlHRy1EJKn+WywIAAAAA8EJC7cNLly4xTawVnZIa GhrQIuzatWtwcNDhcIyNjb3//vuo5PLly0yFcDi8ZcuWTZs2cbncYDCIKjBL TRBE/k3q6+tRyenTp10ul81mKysrQ0/PnDnznJZpqaRSab4VXVi+6OIsZZ0s HyRJollFs9fY2Ig2MY/H27dvH3paVVXFVFgpmxhtr97f1t7ezswq2o2XuLBC oRA9RXWGhoZQBfQvWnZUIhAInuWyAAAAAAC8eFCDqrS0NP9D9MpNSffu3du6 dStaBD6fny9UKBTMcs3NzaGnFy9eRI+7u7vzFVDrFJVUVlYyT+12O3q6efPm bDbLlKAHTMvT5/M9u4V5TBRF5TdiYUpayuIsuk6WFbR/onlDoT5/ukSv16MS tOnRDnB/JW/inp4eNA8nT5788MMPmZJFF/bGjRvoaXV1df5NamtrUcm1a9ee /fwDAAAAALxImB+rz507x/yUvXJTEkmSx48f3717d771i6AGJ5MdAoHA3bt3 mYX1+/35CsFgkPk1/oMPPkBPh4aG0FPU+Cx8Z+aipv7+/me2LI/l448/Pnv2 LHMOoiglLbo4S1knyweaWybO0DSdL0QJAsMwl8v16aefrtxN7HA4mOwWCoWY kkUXFj1F0a/oMzswMIBKLly48IznHwAAAADgBYPykVqtRq2vycnJFZ2SHsps NjMtZNTm9Pl8zGOmhZnHtEW9Xu/9z5qd4+PjhRXYbDYqvH79+jOd9SUbGRlh ripkFrAwJS26OEtZJ8tHLBZDc3X48OF79+7hOI4SAco1Op0uf6plhW5ilOXR QhXFtEUX9v78iWBUZ+fOnW63+/784jOXmC7n28oAAAAAAFaWiYmJFywl3b59 +/jx4/kTBxaLBT3eu3dvUTXmRn7mvhWUGdFjkUhUWAE9Xba3JqHmMQoFqJ2c TCb9fn9RSlp0cZayTpYP5mK5iooK5rK6vAsXLszMzNxfsZuYz+ej/3379u2F J+8WXVjkzp07zc3NTDnaB5gHqH6+AgAAAAAA+JxesJSEGpCoPYmWaMeOHUyf YDiOo6eHDh0qqskkKb1ejx4zHf0pFIrCChqNBhWWlZU9s5lfIrSMR44cQfOm UqnQ0wdT0qKLs5R1snwolcp8WGhsbFSr1SwWi7kZrb6+/v6K3cQnTpxA/3tn Z2dh4aILi2QyGebWpEK9vb3MXUsAAAAAAODze5FSEmo9Mo1h1Kq0WCxMIXP1 3eHDh4sqo0Y1KkcNbPSYucEHNVALKzBdQKA3fDYzv3TM3f21tbXM0wdT0qKL s5R1snzk+/EbGBjIFzJXnSHJZHIlbmK3283MfyQSKSxfdGHR06tXr5bMd2OO dvLZ2Vn079GjR1HJlStXnvViAAAAAAC8oF6YlBSLxQ4ePMhcg+RwOPLl4XCY KSyqz1yqxNS8desWejw1NVVYgbkgKt+d+DLBXF1WMt+nn2Iec4PSjh070GOT yXR/CYuzlHWyfBAEwSxy4RVl2WyW6bEB/XUlbmLmkrmLFy8WlS+6sLdv3968 eTN6ipb6/2fvPsDbuM584TPJvfFu1htn403bzU3ZZJN1fGPnbuJ4fa8dZ1Mc x46duCR23CPbkqhKdVmVqrbVe6MoUhKLCsVOsfcKkugA0YneKwnSEmVL/F7x PJ4PBlhAiKRI+P97+PABBmeAc+YMZt4XOHPAFaCdn12EFTqHCQAAAADELD6y JLPZzC5CoUQp7MP5YDDIws6BgQFuITcJHptJLCsrK+zT+8GP5w1LTU2dmiZE iV1KM5JNmzYNRtGcaLbJ9MGSIHLlypXQ5ewLFD6fP+O6mOrJpnDv7OwMe2jM xgqFQrqxfPnysBXZDyuHfVkGAAAAALGJgyyJAsukpCQ2dCoQCEQWYAlUe3s7 t4TH49GSJUuWsLss9Vi/fn3oWiwurampmdTKj5dOp8v+JPYtybx58+h2eXn5 YHTNGXObTB+UL7ALc0KnlaCFLNHw+/2DM62L2SDJOXPmRF5JNGZjNRoN3UhM TAxNCa9fv75y5UpaTjnUFLUBAAAAIK7N9Czpxo0bbPqyjRs3BoPBjz6JlSku LqYCW7Zs6e/vHxyaBI998F5UVMQKUBTKRmdxV/c3NzfT3aVLl9Jz3pZ2RS/y uqRomjPmNplW2DdB77zzDhuHRp2emZk5O+RapJnVxWyKBsrRhn109MZSQxYu XEh309LSuJnP2ZTmlDphxB0AAADAhJjpWRKfzx9lBBr7QRkKg1mEvHr16pSU FHb5UnJyMpdGkfz8fPbx/v4h7DIQHo93+1oWrcgsaTCK5kSzTaYPr9dL6QzV MCkp6dixY9u2bWNfn3GXUM2sLmY/cXv06NFhHx2zsQ0NDWz3pjzr1KlTbGIK UlZWNoWNAAAAAIhnFFlRfHXmzJnbXZEYFRUVjZIl6XQ6VszhcGzZsoVbvnnz 5tCL3weHPrEvLi6eP38+K0BBaV5e3u1o0LiZTKbZEfNgR9OcMbfJtOLz+Xbv 3s1SG1ZbkUgUWmAGdfGJEyeoAgUFBSMVGLOxHR0dbIgds2zZsubm5smvOAAA AADEoWAwaDQaPR4PN1QpzPXr1202G0XXYdfOz1DRNGfMbTKtDAwMUFY4yriy eOriMRvr9/upsex3wQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACm rYqKioMHD/b19d3uigDAbYajAQAAAADp6OiYPcTr9d7uugDA7YSjAQAAAMCN Gzfq6urmz58fZVw0GwBmvgk5GgAAAADEJZfLtXPnztDYKZosacwl8Q3t/TSI j1aP1Iphl8dwNAAAAACIS8XFxRQLzZkzZ9u2bfSfbkul0tFXQZaE9n4axEer x5UlxXA0AAAAAIhLKSkpLBzi0N2SkpJRVkGWhPZ+GsRHq8eVJcVwNAAAAACI P1arNTExkWKh3bt383i8uXPnctGRSqUaaS1kSWjvp0F8tDr6LCm2owEAAABA /CksLKQQKCkp6dKlS0ePHmUfI69atYr+nzt3bqS14iN6BPg0i3wXx3Y0AAAA AIg/58+fp6AobIwNLaH/hw4dGmmtyGmy/MbCm3+6887ODY7WpfTn6tzgke72 yPZ5ZPs98oOersMeZYq/+5LfUGDQ1dQrVCVyc6nCwTP6dN4+raevTus+JzSl dehP8roPNGj21qszaytE1WtslS/Za9/yyo/4tZkBQ27AWtnj4jntHXmd/PUl whWFoj01UrND09eju/nX201/Rrt2Y4F01hnR66fFK/LkO6tVe2o1m0tVfz0p feaI5E+HJU8flD51QPrGvvLa9x7QbP+i/cC/uEvf7rFUO2y8cy2dSy4K5mR2 bs5v1Eozfep0t77E4jIa/R8Y7XpTVaIp7zHTpUcshb+zlj1rrfizreZVe/1b 3bULG8s2rc0seSFFlHiqpu7CHHXuH6zlL7nlacEPeoJXPwg42ryaM17FMa86 PWCt7rE30n+/PtenO+dVpTlbl9nr3zY2LOngnS0S64oker64xCLa5xZucQuS nbyVTt5yZ9syZ9tSZ+sSReO6jJK0uZltidmC8zyRwSoJ+qRBnzzoVwT9Srmx 61ijfE2J9P0qqaxb5HLSn0Rm0lYobeUKs0TT7ugu9ndf8KnTnJ3rne0rHS2L 7bWv2ypfdDQu8DulPf29PX1ev7XB03XELdrm5ic7mhc4GhPtdW/aKl60lj5r r37dqzztN5ZQP3pkB50tSfSoreoV6+U/aYueb8xfsCS7+m+ZfPp7I+Pm39Kc Dn5Xg8NUGzBf9qpSXPx1zvYVN5+t7HnD5Rck5UsvNVacaO1O7zDUat20M4jN Lr1V6aPtYy53dWyipzXnPXrzL5f+ftGa++buvItL8kQbSmW0w8jsPa3qnj1Z rifmmf97lum/3zT+8i3jr+bqH1+ufTVVRHXYdrmDr2xzWFv8rk63V28L9Fn9 PU5rm0eTYZWfbO/IWnapZVZW59b88ramwy7BZhdtcPH7HvEum/SYRF5RqTRV dukFrScNlW+ay//aXb+qQ6NuM3olZqvDKet1d3qcIqHR0qJ38YxujdPmCxhd Xn2HTjvnrPS5Y+I151taOwt9umyrJq9cIr0o7M6VWOp1bpUrqHT1VqmdmZ2m U22GNJ4hg2/M5JtK2hsUjVto4zjq3rI3z3e0LNK1rKttzkxpkp9s0TR1SWyW dp+j3WoVdepNtRpXtdqZJ7FeFJmz+KYTrbqdtarddaocsblF7xVaAiJrQO7o 6XL20t+w79zILCm2owHAZFu4cOH9999P/6fyRV0u1/1DKioqJrD81q1bf/az n61aterGjRsTVNM4hB6fEZ588klq/t69e293RQAmC73N582bR1HQwYMHpVJp aICUm5s70lqR8RWyJGRJyJKmf5Y0+rs4tqPBpLpw4UJGRkZra2vkQ83NzfQQ j8eb4spwsrOzq6urFQrFtWvXpqwOn06/+c1vEhIS/vCHP0zli9pstoQh1O8T VV4kEiV8rLy8fOIqG2/Q4zPCv/zLv1C7Vq5cebsrMl1Qdx85cuSNN95YsWJF fn6+x+O53TWCW1VSUkIh0JIlSwYGBuhu6JUIdPYfaS1kSciSkCXFX5YU29Fg Un3mM5+hs/BXvvKVvr6+sIcefvhheogqeeuvQqnW9u3ba2pqRi+WMIKvf/3r GzZs6O0dfjtPtl27du3YsePKlSvjWmvnzp3bP1ZcXBzNKj09PZmZmZs2berq 6oqpprGLm5hZLpdzuw0mRRkFenxiRXmIGy9kSRw6Ar/44othp4Y777yzoKBg ol5ikjoxNuOqTEZGRl5ens/ni6bwRx999O6773Knp6qqqlur6a1SKpU5OTlN TU3sbuhIm1F+JwVZErIkZEnxlyXFdjSYVCxLInS0DHtoArOk9evX01O98sor oxdjNbn33nv/NOTXv/71j370o89+9rNsOUV0169fv/XKjMuHH37INpFWqx3X iv/zf/7Pn/zkJ78fsnv37jHLq1SqL3/5y/RC99xzT1tbW6z1jVHcxMzk8OHD Tz31FIUB8Tf+agKhxydWlIe48UKWxPT09Pz85z9nO8NXvvKV559//pFHHvkf /+N/0F06QRw8eHBCXmWSOjE246rMxo0bf/zjH9OmoESSkqDRC1+7du2JJ55g 56a77757/vz5E1HfiXHy5EkuKKItMEpJXJeELAlZ0kzMkka/Lim2o8Gk4rKk f/qnfwoEAqEPTWCWRE8SfZa0Y8eO0IVGo5EyJvZQWlrarVdmXBwOB3vpGLKk y5cvR19+1qxZFAY4nc5xVnBixFPMDNFAj0+sKA9x44Usidm6dSvbEyg/+vDD D9lCiURCGRMtvOOOO6xW662/yiR1YmxiqMzp06dplezs7OhXoXPrdMiSgsFg eXn55s2bQ8OnzMxMt9s90ir4LglZErKkmZgljf4uju1oMKlYlvTFL36R/m/c uDH0oTGzpN7e3ra2NkofxvwIl05tMWdJ5IMPPvjqV79KD42UePb19bW3t6vV 6vF+2XTt2jWNRtPY2CiTybiTbyhaPjVZ0he+8AU6x0VZmDY4VZvH40UzCpFi idC9i253dHRcvXo1tAyLmZ966il2lzZLZ2dnlE2OfjcgFouFCvf399Ntu90+ Zgw83vIxoFCcXmLMwSoDAwO0Jfl8PqvMFEOPR1OxKE1Gj0d5iAsVzVEr5ixp XDsMQwdA2k+oPqN/GUHPzDagVCqdmitG6aR5991303Z44IEHwo7S1EC2hyQl Jd36C8XQiYNDeyltDarkrVfgFitDO9L3v//9ce0t0yRLonfZ7OEIhcKRVkGW hCwJWVJcZkkxHA0mFcuS1q5dy3Kl0IF/o2RJGRkZP/zhD7nvof7hH/7hhRde CPsepKSk5J577vnOd77zta99jQ2NoMThrk/66U9/GrrKSFnS4MdB3X/913+F LS8oKAgdlUc1oZPLSJ8r5ufn04uyyPDKlSurVq36p3/6p4SP/fM//zP3bdqT Tz5Jp5t//dd/pfLs0X/8x38Mq/zFixdH2bDjzZKiLE/xyYsvvsiyWkJd8L3v fS9sGnnqxC9/+cvf/e53Kfv79re/TcVo+y9btoyivueee471xTe+8Q2KALlV 2OZ95plnlErl7373u7//+7/nvmGkFYfNHwej2w04Bw8epO3JSn7uc5/7xS9+ 0dTUNEoMHGX5ysrKu4bzpS99adhq/PnPf6ZHMzMzqVEbN2783//7f3M7wIMP PkghYuQqtDu9+uqrbLuxbR72WlFedxYb9Pi07fEYDnFM9EetkbKk3Nxc6nF6 /sgoOpodZnDoTERb7Mc//jHdpmTt6aef5lahnYEOwpG5G5t9iCvG2kvPsHfv 3g8++GDYjT8hjh07xl7uzJkzkY+y8xTVmbt09K233qIt88gjj0QWfvPNN+kh 2vm5JePtRMoNabtR97Gre2jbsrpRb95///2dnZ2hL6fVatkzRF4xRzkseyj0 sqCY9yjOeLOeaZIlbdmyhQVCixYt2rVrFxcXNTQ0jLQKsiRkSciS4jJLiuFo MKlYwEOBBx3h6QYlDtxDw2ZJV69efeKJJ9h54Y477qCQ45vf/Ca7S1kGnUG4 khRQJYzl3//930OfnC0cNkuieIYeevzxx0MXzpo1i3sqihk+//nPs9t0Kikr K4t8EoqU6NH//M//pHP6b3/727DK/Nu//RtXklKkMStPzzbKhp2MLKmwsJBO 0OzVKXyi+IQLbkO7iRslyL6AYyjsDGsyxcbcKixm/ru/+zv2hLRXUFbIlfzV r34VFjJFvxsMDn3CyY2ZTBjKN1l8yAXbYTHzuMpT5jtSBw27DVlLV69eTY1i xb7whS9wq1CUHvadi8fj+V//638lDEVBFLfT6tTesBfKy8sbveNihh6fzj0e wyFucJxHrWGzpPPnz7MQ+rHHHgubVSbKHYawDJQ2RXZ2NitDz0lHIa5uYRf7 9PX1UXzOHqIY/qGHHvqP//gPLpGk15q82XUWL17MqjrsFDopKSmsDtykN3/9 61/p7k9+8pPIwmz+h9BP28bbiVzmzl0nRUkuvdfYbXpDlZaWcoVVKhVbTjtt WE2USiV7KLTTY9ujQk2HLIlyw3GN+pPJZCwKmjNnTlpa2s6dO9lt+j9sXszg uiRkSciSZmKWNPp1SbEdDSYVi0MKCgpyc3PZAZ/7ZHjYLGn58uXsWL1ixQpu jEFraytLK0I/vqNMxPKx++67jx599tlnLZ9E0V3ok7NnjsyS6IRCZx96aM2a NdxC2oCsPJ2vqcDg0AgZylzuvPPOhKFp8SJH1LAs6Rvf+AYLnO699146K1Ed aEXqGkoVuZK0kNWQjfQmzc3NYZUf/ePTCc+SzGYzi39++MMf1tfXs4UUNnA9 wn0mycXMFICVlJRoNBoW9ZF169ZR/z755JMJQ59SciOm2AZhZ/l9+/a5XC5a qFAoKKNky8M+i45+NxgcurieFaZMXCgU0ovSKqmpqXd9/D1dWMw8rvK0BcL6 ZdOmTazYsJuRayl59dVX9Xr94NCYmT/+8Y9sIQVdoeXfeOMNtlm4zzEoGGPJ CL3QKP1169Dj07zHYzjEjfeoFZkl0SGLxeSU84ZlDdHvMIMh0T57XeruD4fQ iYDly5Q/hj55VlYWK0x7C5dBU+XT09O/853vTOp74emnn6bXpaRs2Eepm1jF uK90x5UljbcTQ7cb7fY8Hu+jjz6iFHL//v2sX370ox9x22e8WVIMe1SY6ZAl 0Rmc4orjx49HWX7t2rXDDrAhCxYsGGktfJeELAlZ0kzMkkZ/F8d2NJhULEui MyBFJixA4gZ4R2ZJHR0d7DNeipTCnodOFuyp6urqIl/lZz/7WUJM1yXRCSgv L+8HP/gBi/GoAmx5T08PGyx3zz33sGnVORQJsOdJTEwMe36WJTEUI0VzuQGd eVn5yb4uaczyTz31FFXj7rvvNhqNYQ+xGPiXv/wlu8vFzNyFZhQc0t1HH32U BcmnTp1iBVhsPPhxJPnNb35TLBaHPrPf72chE/Ugt3Bcu0EgEGCRG4VSYRuc XisyBh5v+UgUxY0ZM1Mlwz4q93q9//AP/xC527B0Y9myZaELWVRPEewo1bh1 6PEZ1OPRHOJiOGqFZUmUZLGOePzxx8O+Ahsczw4zGBLt//jHP6acMbTw6tWr 2UOhYynZqGwSOR0rJQWTOrsgGyQZOkwuVHd3N6sY17/jypJCRdOJ3HZbsmRJ 2AVc69atYw9x+dp4s6TxVibSdMiSCG0cesedOHFizJK084wUFLEhNyOtiO+S kCUhS5qJWdIo3yXFfDSYVCzCOXv27ODQoOiEoQ9RLRbL4HBZ0pYtWxKGRj4M +zUKnfoTIsZpMOPKkiiQuHcIhXDcSIawZ6aQjC08f/585PP8+te/pocotwpb zmVJ9913X2SMMaypyZKuXLlCLQ0dqhHmww8/ZN+mDfuZLW0EFh2xu1zMzH10 TMFzwtDVZ+xubW0tK8A6enDUGc9o50wYGmTFBULj2g2qq6vZa0V+WzrsDGbj LR8pmpg5dOwZhw2hoXiSW0KhLHuDpKamhpbkNuDk/aglenyk8pGmQ49Hc4iL 4agVmiVRzMnq9vvf/z7y8DWuHWYwJNqP/ElxOnCxh7hPpUhFRQVbSNkHxfxR Hj8nBLvU7o9//OOwj1Iqxyq2bds2tmRqsqSwS5AGQ3ZX7mc1pj5LeuaZZ954 443oy0/edUmLFy+m3TXsi9pIcrl8lLholO8o8V0SsiRkSTMxSxrlXRzz0WBS hWZJ5KGHHkr4ODNi0UVolkRH4IShUXmvDod9TDrsl2LjypLCULpBUUFLS0to yQMHDrBHwz4FZdgInM9+9rNhQR2XJUWe4EYyNVnSyZMnqfwoITf3ofr//b// N3LL//KXvwwN4caMmbloLZqYef/+/awwd3X5uHYDbnXutTjDxsDjLR8pmph5 2Jayj+JDP28f/DgFCAsk9uzZkzB0Ucko1bhF6PGRykeaDj0ezSEuhqMWlyUd PnyYu0qrubk5cvVx7TCDo0b7fD6fPRQ2MIBqyNXhrrvuev3113Nzcyd8brdI jzzySMLQdUDDPioSiViVMjIy2JLblSUR9l54++232d2pz5Lee++9L37xi2E/ 6jGKSZ29YeHChbTD0NltlDK0j8X8XVLYEmRJyJKQJc3oLCnmo8GkCsuS2MAP itgNBgMbtx+aJdF5J2Esc+bMiXyVcWVJ9IoNQygzMplMw06Ty420H/Z5UlNT 2aMymSx0OZclRT/v+mRnSd3d3RQDUOFh56zgjHLFeigWfU1szMxdSUFpPlsy rt1g2bJlI/XUsDHweMtHijlmZhMIhMXMK1asSBi6sJ37/L+xsZFdL/PSSy+N Uo1bhB4fqXyk6dDj0RziYjhqsSzpW9/6Vuim/t73vtfT0xO2+rh2mMFRo30u 74gcPi0QCJ5//vnQ6S8od05KSjKbzaM0/Ba99tprCRHXSXG4QzSXPN7GLIl9 7UWbiN2d+iyJsuBHH330S1/6Eu1s0ZSPIUvq7e1dH51169Z97nOfoxgj7JvZ UDqdjiowUlw0yg+RIEtCloQsKc6ypJiPBpMqLEsi7FNHinkoJEj4ZJbEBrd/ //vf7xzZsKdLdsB/+eWXR68MO2uMni8wu3fvZoWHzXe2bduWMHQlQti0S7eS JWk0mihXYaLMkijquOOOO/7xH/+RIrFRinV0dHAbZ5SNzzLKiY2Zk5OTE4Y+ 4uYuFR/XbsD9HGRkXDdsDDze8pEmNmYeGBi499572RNSnPbd736XvWUoUh3z B3duBXp8pPKRpkOPR3OIi+GoxbIk5i9/+UtbWxubE4+Lwznj2mEGY82SmA8+ +KCoqIhODV//+tdZyS9/+cth8xxOILYzE0o6Ih/lsmzaW9gSliXdd999kYWj yZJG78RRthttWzZJ4OrVq9kSLkvKzc0NKxxlljTmSTOyDqtWrUqImFBlJDFk SbT3/nfUWBtHT9komWLDaSLjosiLizmhxdgSZEnIkpAlTf8sKew9PiFHg0kV mSVRuJ4wNFkWG44SmiWxq1PpRDDeSV9Z5vX73/9+9GLcWX7MJ6ysrGSF6WQd +ShFRAmfnNmbiSFL4sJLHo8X5SpM9CPuAoHAwoULv/SlL40y1J/iVTbpLnf+ HcXExsyPPfYYCxG5JePaDc6dO8deKzINHDYGHm/5SBMbM+fk5Hzuc5+7++67 H3744TvvvJPyWYpeNm/ePObUHzdu3Ghtba2vr4/tlzfR4yOVjzQdejyaQ1wM Ry0uS3rttdfYXAHcAMWwK0DHtcMM3lqWxKGMklI/dvUo5SYjFbvF90JzczOr T+TkIVSBr3zlKwlDc1BwC998882EoYn7Ip+K/VTrSFlSNJ04ynbjfuKWu8LO bDazJceOHQsrzP1e+UhZUpQnzTCpqal0Tg+dr3V0kzriLiMj47Of/ey8efNG L0Z7LDeiZuvWrcnJyVxcRGfGkdaKjK+s3aX0Z9Pm2flbHbzV9EcBp0e21yPf 71Ec8SpO+FSpPvUZv7EoYLrc3d1QJVPmiY25Elu12sU3+ztNvstd9rOdxpNt 3cdadLvrlDtrlGdqKkS1m+y1bzqbFngVJ/26iwFjYY+lvMdR77Q2FXS2byoV vlMs2V8nNzv1wR4j/QWG/rRW/bqLXa+fkL5yXJp0vmtbufL9KtX6YsVfT0qe PSp+5oj4j4eklCi9fqCqcMfjHe//QHHwJ/qyZQ5jvdHckdPUsOF8xcqMy3ty 83Xik5TXuHS5NqfW7O812zU3s6T835nyfknRu71mlq12tqF2gapuibh6VdXl rdvOZCYev7z65LnmC29ocp+0lr3klqb09tp6+9wBW9PNLEl53Ks967c1Bpzt XmujQ3+Z4mebMsvessJRP9vQsJTXdjZXqMkVatsFBSbBThd/g6uT0opVlM5Q csESJUXj+rMl6XMyeHOzBOdaBXqzoNfz//8p9fwzLZ3vlXccrOnkawTdZqHO ImnTqvOl5nyJsUPZaequdHUXOjXnbcKdVv42K2+tvSHRXvO6vXmpxSK2eL1W r9NhqHHJDrmFlCVtcvJWUBLnaJpvq3zZWvacvfoNL/WjqdRjLDZLTigb1ipq l+sq3jSVPqct/ktz4aJVF2rnnhfOPieclcV/I5O/9BJlSfUOU03AXHIzSxJs cHasdtArVr5iqvhbV82akrY6ShAo1G/ptgmMDqnZrDJ1GY3NJl2FqS3ZXPK0 Oe8XlvxfWooetxQ/0Vq0YE9B7pJ80fpSWbXGJbIFWtSBg5fcL662/DHJ9Pgi 46/nG369qPv372heOSWkV996uZOv6nDY2v0ugduttnvsVrfVpG82dJ1Xic42 tedsKO5YkifeW1rd2XLIJUh2CTd7pHu9sgM2eaq4q7pcYSqXdwvaThmq5lgq X+tuWMvTqJr1HqHJarIpKEVy2CUSk6HDYBGYrDqn2RswObzGdk33W2myPx2U rMxqb+4od3cXmLSlJSL5OX73RZGpXmvTOCmlcjdo7bliS7bAdF5gyhV254l0 1R01qiba4Wc56t92tC5z8lZ387bVtl440diV0qxpVMgpP/I5hTarSKiRNXV1 1cmVFwS6s536tHb98Zbugw3aI03aQnF3u84gMZpkZqvK6VO7AvQX5bs4tqPB 4C2fZUYRmSWR3/3udwkfC82SampquFP2uF6FDZn45je/OdKvVTLsyaPJkrj5 qf7zP/8zbEheW1sbe57XX389bK0YsiRuAqWjR49GuQozruuSXC4XdQRt3lHK /OIXv0gY+vHEsGGEkWKOmbkp0ThcfDJr1ixu4bh2A4VCwQrT6TjsIW6IUWgM PN7ykSY2Zv7Rj340Ulg7CtonKRJj1RhvnMNBj8+gHo/mEBfDUYtlSU899VRo N7GZsT//+c+HBerR7zCD48+StFrtsF/lDH583VDopIihJuS9wFKGu+66K/SX kQdDZt4L/YEe9mUoHVHp6B1amFIVNkPgSFlSNJ3IbbfIwzW1LmForCY3PJsS W5a6hs2o4HQ6uSn3R8qSojxphnnhhReG3dVHMnlZEp1tKX2m+GHM+Q8pjRpp jE3YVJOhIj+OLhLK6K9C1Nkty3LJT3i7jt783kRx3Kc8GTDk9lhreh1Nvc7W Xq+w1ytWGCTpbV27ahRbK5TJl5XrClX0t6VU9V6V6v1q1bYKxZpi6aoiyeHK BkHHGY/0gLfreI+1usfV1uNopkSJntMuT63qKD9SL9rXoD7b0W30+L3BXmdP j8Hj1bo87WrXwp2mZ5br/7Bc+8puxfws8aJLonkXRC+lCV9MFf3lpOjPx0XP HRO9eJS35Nj5NUfTt6ZnnCqtrpSpa2Xy2oqDtZkvNaQ/2Z79rLlhESUILvFu n7nO71X7bK222kRz0R+sl5918tb79QVOfVkV7/KR0pJduTlHzh0qOPrnmj0P NB58UH3hV4b831rL/+zsSPYbCimn8HVf8KnSvOp0T3eey2twBVw6h6VRqbos klYJmjUtmyyNC3UNK6oaTh+plx2pk1XVpenqltjr3nI0znN1bnALNrv5ya6O da72tYqmbemFma+ntr+RJjhb06BVl/tNxX5jvk9z1qs+bVWcVghPCztP8zqy L/J4Z1qlKc1du2vk6y9Lk0ulme3KZrVGpNeK9CqxVizRipTKaqv0hFv4rlF4 II/HO92mzmqTNnbkGvg7XIJNbvH7lNv6tOc8skPOpsX2qpedDYl+/aUee4PZ 1FLeXvtu7uXtF/OKCt9Vlb1pqHhDWLn6cA1vR41qU3nXwkvCt851rsxr58ur 7fpyv7HAp0r3iHfQa7mF292ibS7xLnvX6W6LUu3u07h8FqvYZm7T6VtqxLyU BkFKTXNlcbLi4mPGC/9lLnjc2b7OJdje2Xp8f2nNsnzx+hJZntRaq3XVql35 fGdatf1omW35Gf1f9iif2iH/0x7Zy6eEr54WbC6RCHRKp0fT65X5LNUu1WmL PL2+o+BsQ11KfevpFsFlubVC5WqR83X8Q07eKlfnRp82228oshuqRRpBscxU JNF18i8Y2zbYWlfpOvbUylXlSmeN2tHWbeUbLCKDodvYYTI2Wc1tNofa7nN3 O/w1UvcLu3S/Wq+ee1RT1K5TWbpFhu6MDkphNKktylqpwGZscJmbzHa11uWl FEbjsOsMPF13o77rgqVlla3mNXvDXG/XMb82y6wpqBU1HG5QHG7S1KjNRo/b 47M5rAJ9+3519VJ+1YY9JeVrCzs2lMqOt+oqVM56jVWkaOiWZZjlaTbNJbdX 5+mx0d+w79zILCm2o8GEnGVGMmyW1N7envCxsKGA7GM68uKLL4YOrlMqlX/7 29+4iYjDHD9+nK21fPly7ph/9epVOuOEFmNlosmSBkM+U6Wcjg22p6MxJSb/ /M//nDA0UV7kJdIxZElUWzYz8Le+9a3QOZN1Oh374ZWRTPhM4LSF2eUA3/jG NzIyMrhTDyXOFEw++OCD3LS9McfM5OGHHxYKhYNDG7O0tPTuu++mhXfeeSd3 IT8zrt2AfX5LFi9ezAZxUVXZ1R/DxsDjLR9mYmNmdjn2q6++ys2hHQ3uN1yY 0BmVo4cen0E9HuUhbrxHrWF/VZayLfaTvv/2b/8Wep1+9DvM4DizpL6+vnvu uYcCfjqFGQyG0MIU5LMfpR1pOMSEvBe4mdK/9rWvFRYW9vf30+GXDS1LGJoY MHRS7ry8PLb8scceY7NVaDQa2oZcHUbKkqLpRG67UcK7e/duNtUG9cIrr7zC lnM/pcGwn2unlDYnJ2dwaLAidXfor5aPlCVFuUeFmSYzgWdlZVGK9Pbbb0cz RTy901kUtGHDhpaWFoqFuLholC9GI+OrjA4d/V3iy7VdeS7VWZ86zadK9SpT KDUImC73uHm9PlGvTxL0K4MBldyoONYo31wmXVUoXZQjeTND8laGZFmeLLlU saVCsbGsa1mBaEm+aF9Fk1CU41OnU7jY42zt8Yp63O1+Q75XesAuPlDDKz7Z KD7SpMsWmAy+PnffFVtPv8bT0+X0Nyl8czdb/zDX9LtE/fNblbPSRG+fE7yZ JXj5tPClNNFfT4n+nCJ6/oToLycEr6e0/u1k04KzrTvLpLlCY4lQ0Vy2TZrx qDL9Xk3Wf1mrX6d41cXf7DeV93qlAVu9vW6utfhpW9kLbsHOHmebyyEq5PM3 X+atzqnalXm85tAjyp1fVO37qvH8Q+b831jLnnO1rfZrTvt1WT7NGa/qlFdz 1qMvcvpdzt5+pdNf1mXO5uvyO+lVt1kaFmvrV5bXpe+tFe+tEZVVHdNU3hzJ RomSi59882sd4RaK4SljUrS8l1aU9erJ9tdO8c9U12oVxX59rl93ztd1xCs7 6JYesksPWSWH1cLUMw2t+2ul71Z2rS2RLsoVLc0XH2/RVqksrQYHz2Dv0Ds6 DXaZTmJVZHtlBwziE6cbWnbXSA/VdJY1n9d1vOsSJrslO2mD+81llH85W1fY a15zNi3wGwp6nC0GMz+nrX1RdtvCjPqs/P1d5XON1bMkNWvTGjsPN2nfr1ZS D759vnNVPo8vq7J3l9Ja1JUeyW63eKdHuscrP0BJtEd3ye3Ru/oGXL09Hke7 11Jj0FVf5rdsrxBuv9yan7dJfu43xvM/txQ+4RHtpMRBwM86UN5IWdK6ElmO 2FKpooTFVd7lKpI4cwSOTQXdrxzv+tMByXOHJS+dEr6SLthcLBXotS6fPuiX Uy7plu62CHeVNufsr2x5v1JytFHZavSIbQGFXmwWHnS2LXd1bPAZcv3WKru1 WaiVFEpMBSJdp+CSqWObvWODjn+wSq4q6bKXKx11WkdTt5OnN5kNTW5Dhcdc 63B0WX1elSNYIfY/v83w6ArdW/uNuTyH1OppN7hO8fQHGjTHm+TVwlaH7rLP WOZ2djl7/M7ePqff6bY0uU1VLlWGvXUlZUmUGnvVZ/3GYrOhplbSTlnSoUZN tdph9AW9AY/LIrC0rDNefl5c8ub23PzFOa3LCyXHW7qbuz18g0OtLLdIDjgl O13Kkz1eRW+vmf6ifBfHdjSYkLPMSIbNkgj3k4thWVIwGPx//+//sYdo3e9+ 97sPPfQQNzqdgqth5wqmYzv3M5cUCdAJ98c//jE7oYeec1mBKLMkSh65SrKA gfvR+b/7u7+7ePFi5CoxZEmDIT95SYf9f//3f7/33nvZxdQUHI6y1oRnSYND 9eeum2YDgagyd9xxB1uyefNmVizmmJn9tCWhmI3FiqzVkT97Ma7dgEIUNiom YeiD1i9/+cvs9ne+8x320XpYDDze8mEmNmamsJDbxyhC+94Q2g1o4z/99NNH jx6NvJpmcGjoPtcvhAVIMUCPz5Qej/IQN96j1rBZ0uDQSYGNc3vuuedCl0e5 wwyOM0symUyhP8571113UaLxxBNP/Md//Adb8q//+q+RP9LETNR7gbIG9r1M GKqDTqcLLUkb+Yc//CG3r7LEP2FoVB57X4yUJUXTiaG/Kst2129/+9vsKyr2 EmHzy1G6yhWmtwmbrZ3K79mzh73WSFlSlHtUmOmQJbExq7NmzYryV7TojM8F QgsWLAj99Dj0d+TDRMZXaTw9/Z3vUKoVhU71zaTg5lcnypM3syRzWa+HH/TL +vzyvoC6r0cjM6qONMg2XKbMSDz3nPiVdNGr6aJFF6XrS+TJZfL1l2VL80WL c4V7K5oFojyfNtPffbHX3d7rk9LzUNDukeyxifbUtRelNd8MDs+LzHpfP0Xa 1p4rKndQ5uhp6PLP2Wh/8i3rY28Zn9ukfP2UcFZW598yO19OpyxJSFnSX1KE f76ZJQlfPCn6a6qI0rQtperzAnOBQNlSukl19gFD2neNmffZql62179FuQkl ekGvuMdWa69PtJb80VbxV494X69H5HIr8wSytSWipRca3jub0njwZ8YdnzHu /YKZAvuhLIkyC5/qpE9z2qtO+zhLKnYEPI7g1S5Xb4n85iDDnA5JV8t2S2OS pmFFaW36rmrRrirh5cqj6orZtqq/2utm3cySRNvdwq0u/kYXfyhLKsx+NaXj 1ZOC05WUJRX49Rf82gyP7ABlHzdHOUpvXgumFx8/Vd+yo0qaXNa1slCSeFGw 8JKQkspylaPZ4GkxeHhGX7vJJ9UrLaocb9cRg/RUSm3z9grx7or2koZzuvbt bsFGj2SX31QSsFZ7NZmutpUfZ0mFlLTqzYLsVsHsTMHbZ1pP5x2Slycaq9+U 1qw73cw/1nLzGpmlw2RJp6mGbskuqqpXccSnOuXT5/v95sCVDwP9vQFHW8BS adRVFnc0JZeKNha15lxKlp37tfH8A5aiJzySfbQBBcKLByqblhWI112WXRCZ KVupVjsr1c4KlbNI7tha2v1qqvxPh8XPHhX/9ZTwZcqSSqR8fbfLbwz6FX5j gVu83czfVtRwYWd5+5byLko9Os1+pSuoNUmtIsqSlrk619MOFrDVOaw8gU5W IDHli3QdglxT53Z7Z7JOcKhSpi6W20sVjhqNq17nbu22WIz1XmOpz1Llcsqt Pp/S0VchCjy7xfDIMt2sfaZLPI/E1kObOrVNv69ec7RRWi1scuoKKQPyu+W+ voC//wN/r8tvqw9Yyr3qM87W5bbqVx0N83za7IC53GJsrJXyDzUoD97MkpwG X7+nx+ey8m1NKy3FT8iKXtp68eL8801L8sTHW7t5Rq/Y7NCpSu2i3W7RNm/X 0aBP3hc00V+U7+LYjgYTdZYZFjvvRGZJYrGYJVCRQwHp0Hfq1Ck2mQ/ni1/8 4ltvvRV2tgplNpt/9atfha5Cz08n8dBzNDvvR5klMenp6aEXONMz/PrXv1ar 1cMW5q6AGFeWNDj0SyVcPMlQILd48eJRVpmMLIlQKEJxETvVMnTCve+++86c OcN9mkqtYw9xMTO7MJybbZ4ydFaAy7jZGEsKsY4dO/bVr36Ve/If/OAHI00r Ma7dwGq1/va3v+ViCdqYr7zyit/vf+CBBxKGmwBqvOVDjR4zs5Y+++yzkQ8N GzPTG2HY2IxD+/CwF5SlpKT89Kc/ZWWOHDkySoVHhx6fKT0ezSGOif6oxeJk LtsNxX7Eipw+fTp0eTQ7DGltbWWPho1hGxz5uiRKzZ566ikuNWbo3PT666+P PgXoRL0XysvL/8//+T/cj+hRgvnCCy/QThJZkvLuBx98kKsk5eB/+9vfKL0t LCxMGJosfaSXGLMTuSypoKCAdi1uanTa4JRucBOecOhds3v3bu4MTn3x85// PC8vjx5ig+4qKipirkyk6ZAlUTdReDPs5LTDSktLG2mMzYEDB0ZaC1kSsiRk SfGXJcV2NBicuLPMxPL5fBRQUWAW+UMnI6Fwrrq6ur6+XiAQTOAUYTabrbm5 ub29fdhvsibEhx9+KJVKWeUpJBjzFDBJWRJD0Y5KpSopKREKhZPRZIVCUVZW NuyvukSKfjcIBoPUTVTnKC+vG2/5CUcB81133fX5z3/++PHjFHd1dHS0D6HA ht6G3PzYI6UVLpeLRXQUNtxiTdDjU+MWe3xwPIe4ST1qTeoO4/V6aQtQvkAv EZkXDGsC3wu9vb2Ur1FPjXkQpiyjsrKStkCUX2pwRulELkuih+gu5Wj0RuDz +aPvsdQFVA16i8Vw1hvXSXM6ZEnjpVarWRTETQK8dOlSdoM6eqS1QsMntgRZ ErIkZEnTP0sKy30m5GjATOBZBqbAeLOkv//7vx92lCDcRnPmzEkY+YILbk4D 9slwJHZlAYXcDodjMqsJE+YWexxGEjfvhdF/L+m2e+ihh0afCSrMdMiSzp07 N9Knx6N8HIHvkpAlIUuaiVnS6O/i2I4GTNycZT4lxpsl3XfffS+88MLk1Qdi wCbDH+kCNO7nR4cd86PT6djgtNsehED0bqXHYSTx9F6YzlmSRCJJiJgifnTT IUvas2cPhUBz5swJjYjmzp1L/yMvkOQgS0KWhCwp/rKk2I4Gg/F1lvmUoCxp 3bp1hUNCJ8cbCbvI95577pk3b954f8EWJsn777/PIqK33367o6MjGAxev37d arXm5+c//vjj7CEqE7ni4cOH2TXj999/P5tlC2aEmHscRhJn74XpmSXR/rl0 6dKvfvWrP/jBDwYGBkYvfOPGjeLiYnZueuCBB257UMF+R5JDud6FCxfY7fXr 14+0FrIkZEnIkuIvS4rtaBBnZ5lPie9///vf/NiSJUuiWUUqlS5btuzBBx+s rq6e7OpBNK5du/bCCy+EXT3N3f7a175G783ItSgI+fa3v33HHXds2bLldl1f A7GJrcdhJPH3XpieWVJSUtITTzxBUUQ0lxZevXr1W9/6Fnd62rBhwxTUcBQs LmJXH9BttnDRokWzh35ZcqS1Ii9tQJaELAlZ0vTPksLG0d360SD+zjIAM4tW q92/f/+TTz750yHPPPMMpb1ZWVmjfGZbUlLS1dU1lZWECRRDj8NI4uy94PF4 Ll68mJOTg51hoqxcuZILmdivt1Mqyu7OmTNnpLXwXRKyJGRJMzFLGv1dHNvR IM7OMgAAAABk9+7dXFy0ePHid955h7sSYd68eSOthe+SkCUhS5qJWdLo3yXF djQAAAAAiD+FhYUUAiUmJlIURDeSkpIOHz68ZMkSur1s2bKR1sJ3SciSkCXN xCxp9HdxbEcDAAAAgPgzMDDALkaguGjnzp0nTpxYvHgx+zD52LFjI62FLAlZ ErKk+MuSYjsaAAAAAMSf9vb2999/f+PGjaHT/y5cuJD+19XVjbRWZHwFADNL 5Ls4tqMBAAAAQPyhuIh9dOxyuSxD8vPz2QxXV69eHWktZEkAM92wWVIMRwMA AACA+EPBDxtjs3DhwiNHjmzdupVdmKDX60dZKzK++rTlTWjvp0F8tHqkVkQu j+1oAAAAABCX7Hb7jh07uAE2K1eubG9vH30VZElo76dBfLQ6+ixpMKajAQAA AEAcCwQCRqPR6XRev359zMLIktDeT4P4aPW4siRmXEcDAAAAAGBmA8DMd7sP JAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAADDxfD7foUOHampqbndFbsn169fb 2tqysrIOHz6ckZFRX1//4YcfhpWRy+WZmZlHjhzJy8tTKpWRT9LX11dSUnLi xIm0tDR6hmAwOCV1v1VUz9TU1HPnzoUtj6Y5Y26TacVut5eVlVGLqL0VFRVX rlwJKzD9u1ggEJSNoLGxMbTkmI3l8/lsh8/OzqanncJGAAAAAMS5gYGBrVu3 zp49Oz09/XbXJXb9/f2bN2+e/Unr1q3zer1cmeLi4rACFC2HPgkFzMuWLQst sHbtWrfbPeWtGTdKCqi2K1euDF0YTXPG3CbTSmtr6/z580NrS02mHJ8rMCO6 eN++fbNHsHr1aq7Y6I29fv06JUdhq9MSWj6VbQEAAACISxQf7tq1i4VYMzpL OnnyJDUhKSnp0qVLKpWqqKho+fLltGT37t2sgMViSUxMnDNnTnl5uclkogKs 1VKplHuSlJQUWrJt2zaNRqNQKJKTk+nu9u3bb1ObotXY2MhF0aHLx2xONNtk +rDb7VRVql5qaip1cUVFxYoVK+juoUOHWIGZ0sXUX+c/6ezZs6yqtBtH2dja 2lq6S2Vyc3OpAP2nttOS6urqqWwLAAAAQPyhgGrBggXcB9EzN0u6du3avHnz qAmVlZXcwpaWFtauvr4+urtz5066nZ2dzRWg6JSWHDhwgN1VKpV0d+7cuT09 PWwJ3WCRp16vn7rGjJPT6eQ6MTRLiqY5Y26TaYX2T6obJfXc1yV8Pp+WUNfT DjA4k7v43LlzVIdNmzZdvXqVLRmzsQcPHqS7R44c4Z7k+PHjtGTfvn1TX38A AACAeMI+rH7vvffYR9kzN0uy2+0bNmxYsmQJF/0SCjhZ7mA0GgcGBlhjDQYD V8BkMrFP4z/44AO6m5ubS3cp+Ax9Zjao6eLFi1PWlnH56KOP3n33XfYdRFiW NGZzotkm0wfVlqUzLpeLW0gZhFAo1Gg0N27cmLldrFKpWO5mNpvZkjEbS3cp 9Qt7z+bk5NCSHTt2THH9AQAAAOIM5Uc8Ho+ir9LS0hmdJQ1LJpOxCJliTr1e z26zCJPDYtHu7u7Bj8PO4uLi0AIlJSW0cP/+/VNa9agVFBSwUYWsgaFZ0pjN iWabTB8Oh4NqtWbNmmvXronFYsoIKK/p7OzkvmqZoV1MuTw1KixNG7Oxg0Nf BFOZxYsXa7XawaHmsyGm0/myMgAAAICZ5fLly3GWJfX392/YsIH74kAul9Pt ZcuWhRVjF/Kz61YoZ6TbdXV1oQXo7rS9NInCY0oKKE72er0GgyEsSxqzOdFs k+mDDZbbunUrG1bH2bFjR29v7+CM7eLKykp69YULF4Z+eTdmY8mVK1fS0tLY ctoH2A0qzxUAAAAAgFsUZ1kSBZAUT1KLFi1axOYEE4vFdPedd94JK8kyKT6f T7fZRH8tLS2hBdrb22lhcnLylFU+StTGtWvXUt3a2trobmSWNGZzotkm00dr ayuXLKSmpvJ4vLy8PHYxWkpKyuCM7eKNGzfSq2dmZoYuHLOxJBAIsEuTQp0/ f55dtQQAAAAAty6esiSKHlkwTFGlXC5nC9nouzVr1oQVpqCallOATbfZBT4U oIYWYFNA0BNOTeWjx67uP378OLsbmSWN2Zxotsn0wc3jl5OTwy1ko86I1+ud iV2s1WpZ/a1Wa+jyMRtLd/fu3Tt7aBpz2smDwSD9X7duHS3Zs2fPVDcDAAAA IE7FTZbkcDhWr17NxiCpVCpuucViYQvDyrOhSqzk0aNH6XZVVVVoATYgiptO fJpgo8tmD83p1zKEXaC0aNEiui2RSAajaE4022T6kEqlrMmhI8p6enrYjA30 6EzsYjZkbufOnWHLx2xsf3//3Llz6S61mitAOz+7CCt0DhMAAAAAiFl8ZElm s5ldhEKJUtiH88FgkIWdAwMD3EJuEjw2k1hWVlbYp/eDH88blpqaOjVNiBK7 lGYkmzZtGoyiOdFsk+mDJUHkypUrocvZFyh8Pn/GdTHVk03h3tnZGfbQmI0V CoV0Y/ny5WErsh9WDvuyDAAAAABiEwdZEgWWSUlJbOhUIBCILMASqPb2dm4J j8ejJUuWLGF3Weqxfv360LVYXFpTUzOplR8vnU6X/UnsW5J58+bR7fLy8sHo mjPmNpk+KF9gF+aETitBC1mi4ff7B2daF7NBknPmzIm8kmjMxmo0GrqRmJgY mhJev3595cqVtJxyqClqAwAAAEBcm+lZ0o0bN9j0ZRs3bgwGgx99EitTXFxM BbZs2dLf3z84NAke++C9qKiIFaAolI3O4q7ub25uprtLly6l57wt7Ype5HVJ 0TRnzG0yrbBvgt555x02Do06PTMzc3bItUgzq4vZFA2Uow376OiNpYYsXLiQ 7qalpXEzn7MpzSl1wog7AAAAgAkx07MkPp8/ygg09oMyFAazCHn16tUpKSns 8qXk5GQujSL5+fns4/39Q9hlIDwe7/a1LFqRWdJgFM2JZptMH16vl9IZqmFS UtKxY8e2bdvGvj7jLqGaWV3MfuL26NGjwz46ZmMbGhrY7k151qlTp9jEFKSs rGwKGwEAAAAQzyiyovjqzJkzt7siMSoqKholS9LpdKyYw+HYsmULt3zz5s2h F78PDn1iX1xcPH/+fFaAgtK8vLzb0aBxM5lMsyPmwY6mOWNuk2nF5/Pt3r2b pTastiKRKLTADOriEydOUAUKCgpGKjBmYzs6OtgQO2bZsmXNzc2TX3EAAAAA iEPBYNBoNHo8Hm6oUpjr16/bbDaKrsOunZ+homnOmNtkWhkYGKCscJRxZfHU xWM21u/3U2PZ74IBAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAfGr5fL5Dhw7V1NTc7orckuvXr7e1tWVlZR0+fDgjI6O+vv7DDz8M KyOXyzMzM48cOZKXl6dUKiOfpK+vr6Sk5MSJE2lpafQMwWBwSup+q6ieqamp 586dC1seTXPG3CbTit1uLysroxZReysqKq5cuRJWYPp3sUAgKBtBY2NjaMkx G8vn89kOn52dTU87hY0AAAAAiHMDAwNbt26dPXt2enr67a5L7Pr7+zdv3jz7 k9atW+f1erkyxcXFYQUoWg59EgqYly1bFlpg7dq1brd7ylszbpQUUG1XrlwZ ujCa5oy5TaaV1tbW+fPnh9aWmkw5PldgRnTxvn37Zo9g9erVXLHRG3v9+nVK jsJWpyW0fCrbAgAAABCXKD7ctWsXC7FmdJZ08uRJakJSUtKlS5dUKlVRUdHy 5ctpye7du1kBi8WSmJg4Z86c8vJyk8lEBVirpVIp9yQpKSm0ZNu2bRqNRqFQ JCcn093t27ffpjZFq7GxkYuiQ5eP2Zxotsn0YbfbqapUvdTUVOriioqKFStW 0N1Dhw6xAjOli6m/zn/S2bNnWVVpN46ysbW1tXSXyuTm5lIB+k9tpyXV1dVT 2RYAAACA+EMB1YIFC7gPomdulnTt2rV58+ZREyorK7mFLS0trF19fX10d+fO nXQ7OzubK0DRKS05cOAAu6tUKunu3Llze3p62BK6wSJPvV4/dY0ZJ6fTyXVi aJYUTXPG3CbTCu2fVDdK6rmvS/h8Pi2hrqcdYHAmd/G5c+eoDps2bbp69Spb MmZjDx48SHePHDnCPcnx48dpyb59+6a+/gAAAADxhH1Y/d5777GPsmdulmS3 2zds2LBkyRIu+iUUcLLcwWg0DgwMsMYaDAaugMlkYp/Gf/DBB3Q3NzeX7lLw GfrMbFDTxYsXp6wt4/LRRx+9++677DuIsCxpzOZEs02mD6otS2dcLhe3kDII oVCo0Whu3Lgxc7tYpVKx3M1sNrMlYzaW7lLqF/aezcnJoSU7duyY4voDAAAA xBnKj3g8HkVfpaWlMzpLGpZMJmMRMsWcer2e3WYRJofFot3d3YMfh53FxcWh BUpKSmjh/v37p7TqUSsoKGCjClkDQ7OkMZsTzTaZPhwOB9VqzZo1165dE4vF lBFQXtPZ2cl91TJDu5hyeWpUWJo2ZmMHh74IpjKLFy/WarWDQ81nQ0yn82Vl AAAAADPL5cuX4yxL6u/v37BhA/fFgVwup9vLli0LK8Yu5GfXrVDOSLfr6upC C9DdaXtpEoXHlBRQnOz1eg0GQ1iWNGZzotkm0wcbLLd161Y2rI6zY8eO3t7e wRnbxZWVlfTqCxcuDP3ybszGkitXrqSlpbHltA+wG1SeKwAAAAAAtyjOsiQK ICmepBYtWrSIzQkmFovp7jvvvBNWkmVSfD6fbrOJ/lpaWkILtLe308Lk5OQp q3yUqI1r166lurW1tdHdyCxpzOZEs02mj9bWVi5ZSE1N5fF4eXl57GK0lJSU wRnbxRs3bqRXz8zMDF04ZmNJIBBglyaFOn/+PLtqCQAAAABuXTxlSRQ9smCY okq5XM4WstF3a9asCStMQTUtpwCbbrMLfChADS3ApoCgJ5yaykePXd1//Phx djcySxqzOdFsk+mDm8cvJyeHW8hGnRGv1zsTu1ir1bL6W63W0OVjNpbu7t27 d/bQNOa0kweDQfq/bt06WrJnz56pbgYAAABAnIqbLMnhcKxevZqNQVKpVNxy i8XCFoaVZ0OVWMmjR4/S7aqqqtACbEAUN534NMFGl80emtOvZQi7QGnRokV0 WyKRDEbRnGi2yfQhlUpZk0NHlPX09LAZG+jRmdjFbMjczp07w5aP2dj+/v65 c+fSXWo1V4B2fnYRVugcJgAAAAAQs/jIksxmM7sIhRKlsA/ng8EgCzsHBga4 hdwkeGwmsaysrLBP7wc/njcsNTV1apoQJXYpzUg2bdo0GEVzotkm0wdLgsiV K1dCl7MvUPh8/ozrYqonm8K9s7Mz7KExGysUCunG8uXLw1ZkP6wc9mUZAAAA AMQmDrIkCiyTkpLY0KlAIBBZgCVQ7e3t3BIej0dLlixZwu6y1GP9+vWha7G4 tKamZlIrP146nS77k9i3JPPmzaPb5eXlg9E1Z8xtMn1QvsAuzAmdVoIWskTD 7/cPzrQuZoMk58yZE3kl0ZiN1Wg0dCMxMTE0Jbx+/frKlStpOeVQU9QGAAAA gLg207OkGzdusOnLNm7cGAwGP/okVqa4uJgKbNmypb+/f3BoEjz2wXtRUREr QFEoG53FXd3f3NxMd5cuXUrPeVvaFb3I65Kiac6Y22RaYd8EvfPOO2wcGnV6 Zmbm7JBrkWZWF7MpGihHG/bR0RtLDVm4cCHdTUtL42Y+Z1OaU+qEEXcAAAAA E2KmZ0l8Pn+UEWjsB2UoDGYR8urVq1NSUtjlS8nJyVwaRfLz89nH+/uHsMtA eDze7WtZtCKzpMEomhPNNpk+vF4vpTNUw6SkpGPHjm3bto19fcZdQjWzupj9 xO3Ro0eHfXTMxjY0NLDdm/KsU6dOsYkpSFlZ2RQ2AgAAACCeUWRF8dWZM2du d0ViVFRUNEqWpNPpWDGHw7FlyxZu+ebNm0Mvfh8c+sS+uLh4/vz5rAAFpXl5 ebejQeNmMplmR8yDHU1zxtwm04rP59u9ezdLbVhtRSJRaIEZ1MUnTpygChQU FIxUYMzGdnR0sCF2zLJly5qbmye/4gAAAAAQh4LBoNFo9Hg83FClMNevX7fZ bBRdh107P0NF05wxt8m0MjAwQFnhKOPK4qmLx2ys3++nxrLfBQMAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAACAGaSzs3PHjh0fffTR7a4IAAAAAADAtLBz586EhIT+/v7bXREAAAAA AIBpAVkSAAAAxKuKioqDBw/29fXd7ooAwG023qMBsiQAAACISx0dHbOHeL3e 210XALidYjgaIEsCAACAOHPjxo26urr58+dHGRfNBoCZb0KOBhxkSQAAABBP XC4XhTehsVM0WdKYS+Ib2vtpEB+tHqkVwy6P4WjAQZYEAAAA8aS4uJhioTlz 5mzbto3+022pVDr6KsiS0N5Pg/ho9biypBiOBhxkSQAAABBPUlJSWDjEobsl JSWjrIIsCe39NIiPVo8rS4rhaMBBlgQAAABxw2q1JiYmUiy0e/duHo83d+5c LjpSqVQjrYUsCe39NIiPVkefJcV2NOAgSwIAAIC4UVhYSCFQUlLSpUuXjh49 yj5GXrVqFf0/d+7cSGvFR/QI8GkW+S6O7WjAGT1L4vP5dXV1E1BvAAAAgMl3 /vx5CorCxtjQEvp/6NChkdaKnCarXudmf43d9Odp1Hua9d5WvbfN4JPaerSe PoOv3+jrtwSuWHuuVMgCs08YHtumeeY9wcGsPEP7ThN/R2b5+SXnmxKzBVsv Cw1WWdAnd3u0SrtTZA3I7F63z9gb0PX5u3qdbQFzacBUEjCV0g2/if6K/Lps nzbDoS/VWPVyR6/CbDC3v2trWOBsWuSR7qMCVNKnPmOvfNla8lRV+tLE1VW/ fNP46GzDI0t1j6zSPrddebRAbDO19dhq3cIdtqqXLeUv2JpXebwWb/81f1+w N2jt6+02ObXvVYgWXuKvLBDkdEpdbqXDqTpZrXj5aNcf94oXp7Y0Np3zKo57 Fcc8XYc9XYe8qpMBU1Gvhx/0SfoC6r5eA/0FA+oeZ3OPvb7H0Uht6XW197hF 7h6Pq7ffGbzq7hvwfnDN98HVQNBJ7Q341Ha3rtXgpA0rsATsvVd6rnzYbR1I uRB46m3rH940HEvlqYUlAUMe/fmNhfRyAWtV0N0R9MmCfnmwR9fXo+/r0Qa9 4psv6mikBlKBgKUqYKvrdXdQ3aQ6weF68cpCyeKL0ifeVz6arH52h+poiVJr 6HK75XKD7P0q6ZoS6dYKxZFmbXq74YLQoHca/AFDX4+m1yMI2Go8lhqxTlzR ZSyRmeq7lI7uYr/+kl+X5VEc98gPeJXHe2w1Pa72Xjef/bkdgiqZ9HBD155a xZlWSZeORxXzO9tdAZejp98R8LmdUr+xKGDIv7miu8PrFhut8ssy4yWxpVoq 1nVd8ipP+NSnex1NVAF6wh5rTYBesZs2/gm3ZI9bvOPmn2Qn/fnU6T3OtqBf 2edX9bo6WNt7HM1Bj8DjEisMkr018k1l8mPNuhaDR+ftM3j9Xk9XwC0MeMQe n8ne06fz9J9ssrx1Rv5KimjNBb5O19RjbzBbOls13eVKR53KaDI0eozlAXM5 9Wmvi+dzCUw2SbnCXCCzValdYluAdn692+u2tfks1QFr5VCxtoCr0+zUKR2e LmdA6/YFgra+PkunTr+hQP6XE6J5ZwXtsha7qZF6jRoY9Cvor9fd2eNs8tga Dab2WoU2V2ItVTjMgQ8CVz/sv9J3c2+x1XssdXK9tEFtrlU7eN1mj6cr6JPS 37Dv3MgsKbajAWf0LGnRokUvvfTS2IckgFu2cOHC+++/n/5P5Yu6XK77h1RU VExg+a1bt/7sZz9btWrVjRs3JqimcQg9/ukx3s0OcCtof5s3bx5FQQcPHpRK paEBUm5u7khrRcZXyJKQJSFLmv5Z0ujv4tiOBpzRs6Snn376C1/4QjAYHPN5 bsWFCxcyMjJaW1sjH2pubqaHeDzepFYgsjKc7Ozs6upqhUJx7dq1KavDp9Nv fvMb2hX/8Ic/TOWL2my2hCHU7xNVXiQSJXysvLx84iobb9Djnx7j3exTibrv yJEjb7zxxooVK/Lz8z0ez+2uEdyqkpISCoGWLFkyMDBAd0OvRKCz+UhrRcZX RTLHx3/OYpmzROa8LHeWdbnKFe4GjbdDHxAYewTGXqG5T2Tuu9jheeuE/vfv qZ7bKdiVVSRqOSRpO5hSkrcws2V2hnBDgUjULTXaZEqzukVrr1Z5mzQum0vr 9ykCHinlMpquKrW8QqeqMGgrDboKvbpUI8tRSy/Iuyo7tVpKzXiabnHjPlnV alXtGrPoqNdYEjCX+bRZzvp59qrXarI2JK6v/eUc/aPz9I8s1z6yWvOnbao9 OWK5iqfX1es6DqqrFqgr5qqbtqpNJrWzX+f0W9wWu7u7y6TdVCKZmy1MuihM a5IqTF1dBsWh8q6Xj8mePSBOSm9raMl3qTJcqrNuZYpHedylSrPqSvRGvt4k 0VtUenu33tbdbVVqda0abaNG26LR8TS6TnW3WG5xyawBuT3Y5ehTOvtVzqDW YTPYNXqrUm3R8QzOZr2n0+zTuCiM7xd396Xmev+20jprpTHjnEAnq/aZylyG Up2yTN1VqlXXdBs6DRYpZRYOt9bl1dN/s0Wq1bVptM1abb1OW9OtqzXqG9xO ASUgfI14b4V08QVpYobs8XeVv9io/uP7ql15ynZ5l7Jb3iiXJV+WLsuTrivu 2lWtOdzQfbpNr7IaXT6j16ezWsQadUOXsqFRJimSGArEpgqpUqUqM6gLdYoc tTBd0XlCKUhXqxuomZpuoUYv0OoFCp0wny/ZV9u1q6YrvVXapWvvcbb4nJ0W j9Po7TW4vSar3NpdbtWVGrobtd18pU7CV8nPtRtOt5kKOiQicYFBdtqoyHZY Wl0OscMutBrqDaoCveKSRpSuaD/S1XZQ0XZI0XFI2XFIL8/02tp7vYoer8Ju 6dSo69SqOoOhzWajTFzSqZG+VylfWyzfX6+t1rikjp4uh9fqVLqdEo9L5vKa HT1BvSd4qtk856zs1VThqgudAkWzwdCo7O5s02rq1JZGlV6mbFYpqtXKKo26 SavrUOmEfJX0fKfpTLs1R+SoVXt5hgDf4LJZ+V5bk8/W4LQ2WyztJnOn1KBp 7XY2dXvb9S61xaKzmisl+tWXZJQlzT0rqBW2qrTNJlOLxSqwOrssji6DSUD7 jELV0tnVkS/Qnmk3XxBYpbYevbff4gs4XDqvrdVpbRFoZKVSY5HEUqMwmh1K t5uy3a4o38WxHQ04o2dJlCbQo+np6WM+z634zGc+Q6/yla98pa+vL+yhhx9+ mB6iRt36q1CqtX379pqamtGLJYzg61//+oYNG3p7e2+9JjHYtWvXjh07rly5 Mq61qHO3f6y4uDiaVXp6ejIzMzdt2tTVNfweOHniJmaWy+XcbhPlJCqfTujx T4/pmSXREfXFF18MO9TfeeedBQUFE/USUZ53psa4KkNn/7y8PJ/PF03hjz76 6N133+VON1VVVbdW01ulVCpzcnKamprY3dCRNqP8TkpkfLXwnPLmX7YyMUM9 98zNv3kZ6vlZN/8WZmkWZ2sXZ2kXZmrnnNLPTtXPOqn96wnZX46LXz4pWJDR kpxbvTG3eu7p1uePCJ85RAvFycXCd8sFGwok88+qZ6V0L89Q8yVNDkONQVN7 qaZlKa11Qrj5kuBQleBYHX9/Rec72W3Lz7atPN+xvlCysUy+oUS8Ortm5Zmy 7edLixrKDNrygKWCEqWAsThgLGpqq154RPjf61W/XK96ZKOS/n6zveu1Y9K1 lwSbCzoPlDWmV1ecqa44XlG/Ple5Kke7pVhzokmT1ak+0aR6LUXxhz1dT+/t eitdtr5QvKFIvOCc+NVTotfTRasvScrFKrVZozXJXfpSnz7XoMgvbKxdd0G0 Ilu6+oJsTY6c/lafly05LV6cJl50SjIvRZZ4Qp6Y0rXgtG7hGf3Cszf/FmXo kzK71+QqtlwW76ySpDQrhVaf2h1s1ntTWvQ7qzV7qtTZTbrGTkO7yKjVW10e u91tE2qN67MlC04KlpwWrssVbSoSv1cqvSTQlckpedEfrlQtPyul111zgb+p kPd+afuxOiGlUl02exHfsuik9sktqt9vVj26Xv3wOvWvklUv7FMsSJcnZcjm pcueO9j19P6uZw8qXjqueC1FuShTXSm3Syzuzm7nmVp9Uop0wXFJ8iXF/lrN kSbN4UbFger2A1Vt7xe3rspoWXSyedHJ1nknxYkpsnkn5QvTZEvOSJdmSFfm yN4pkm0ukx1rlMmNsqBf5vZ0NWrMZV22Yrn1slRXKpEXCmT7SuRL0xSJxxRz jivfPKWYdUq5JFu6t7zzVEPrmab2IomqrKu7WKY92yrZW9Gy43Lz+vNNS9Ma Fp+sX3KqflVm3TtZtUfLmuQagdspNpmluU3iZemiRadEO0tkmW3K852qE42K xGzJG2fESTnyg/W6sx3G8wJjk9bMN1hkFovBbfX12mw+S1aHKilH9MZZfuK5 zu2lHbsqOs408wUakcUmkevEu/OFq84KF58SLkoTL0yTLvj/2HsP8EiOOu9/ H5vD5DMYTDjugIMDbGPDwQFHMM5er9e73pyDNuecc5R2tdIG5ZxzzjnnMKlz nOnpnpw0M5I22cC7/9+oz/PXSVqtJC/g5339e+rRo+murq6uqq76frpSCrE9 iVwdza2M4tfGarelandl6E4V6dWC1Tpgt7isvTp9Vj+Z3EUE19JHC7h9Ofze TO2uRN3OeN2WeG5NAr48AYqTJqgaCW9CYlqR+E48uYdK6KQvlZL7kvAdcfj2 eHJbGrU1g9mVzZyrYoIamIh2tpoQBbtFZ7PkK/XnKpmjJXRQLdtIi906A7gp vsUzqw38NjklAbZAk/H6668/MpyPYzIlgUHtOubUY6SkM2fOQFBr1qyZ3Jsc kxdeeGHBiL3xxhvPP//8E088IR8HRffXv/7140dmWvbnP/9ZTiKtVjutC//p n/7pF7/4xZwRu379+iP9syz7ta99DW703HPP9fb2zjS+M7T/azQzWFRU1Lx5 80A2fDr+ahL7NMf/37FPICUNDg7+5je/kWP1jW98Y8mSJS+//PJnPvMZ+AkV fkRExGO5yxTbnb+PTSsy586de/HFFyEpACQBgib3/OGHH7777rtyW/PMM8/s 2rXrccT38VhiYqJfFEEKTOJz/NSGFfE4uOVx+JIoYlEECQ7+WRrjc0uiycVR 1KJIakE4PSeYnX2Vm3edXhaPrU5G1qVpNmUot+f0bcvuX5mgnheBzQ3DFsVg O3LVewpVmzKQhWE0+F8bQXYpm8zaao6qTazsXhRCzr1Kb00lz1RgF2uxE2XY 6jhkaSSyKk6zORvZUaDZlqdZl6RcHa/Yk96X1dyu4xt8lGRpHnL2D7mU3Zh6 TzL52mX61cvMyxfol88zr12GuBEBacjWLOR0OXqzEQ1rwoKq8YAUAmhuWyYZ WMdEdLBXG9jF4YBUzBtB8DjEpkxkS5ZmQ7pmbapmcyZ6soyqJkyExUlbTHap 2S1WaKnKjIb2gHh0eRS+Kg5fnYCBWxGLL7hFzr9BvhdKzblKzw5i3rnKvHed mX+TmX/rf9yCMHpNErYtV3OoBAlupFGLV3Tfade5IAIHi7GT5Vh+PyOZ9Q6H NOB1eu7cNntud/PugFh2bgi5KJxYl4RuyUT25mPx3bpctZjaqz9Twi+JoBbc IlbHwynV/gL1pWqig7cjxoHCXvuWcPG1Y9pXj2tfPsH/8ST3pzPs7Cv0wlvk 4nDi/TDirWD6jav029fouYCHt6j18Ww5Zu83uNv4geuVxgVX2feCmN1pbFAd d6OVvdZEn6lCTlao9+VrVsag829g865jc4IpCHBOMP1eKLkwjFgSSWxOxw8V Y+eqsZh2ghDp217W4eQaGLEQk3I14MQ8jZCh0B3P45eEsLOB4IIgWaj3w8l1 ScSpMvxaI3ajhchUCgUaKVupv9VKHytXHyhSbUhRLwnXLLipWRyuWR2vWpuo vFiqUrEam02jk5DkenwJPEIodSiHDW/hYzu0IY1sQBq6IhHZloUH1rGR7dqE bqGBsXYJdpXBLjgs3iGDwy3mK+mDJciGLNXmHNWBYvWhEnV4s1rDKd12JadT Hc3CVkTi828Q865DnvrcnGuUr5AEsm9fZeZCzt6iNybyPVo3ZJPRPdTMWiLa mWuN1MF8cl0iuQzejghqThA3+zI/N5hdHI0BJQF3HynFzlRhF2qxwHoiuIkK rKe2p9ELQul3gpi5wfQCeL9iiOXx+NZsdFcBcqqSyFGJrMPNODypPdL+fHJL Jna0hCpBzfA44CZ8c8dT0sxqA789co27tWvXQj1pNBqnEtrMzE9JX/3qV71e 7+hTj5GSIJCpU9K1a9dGH5QkCYhJPpWSkvLxIzMts9ls8q1nQEnV1dVT979x 40aQDXa7fZoRfDz2f5Nm/tSmYp/m+P879glMxsuXL8tRAj7685//LB/EMAyI CQ4+9dRTZrP5499liu3O38dmEJm0tDS4JCcnZ+qXQFv5SaCk4eHhurq6ixcv jpZPWVlZTqfzYZeM11db0lBwG5PRBeH43JvEuzeI927h70dgCyKxpTHkqnh6 VTyzNJp+8yr9WiA15zqxMhFdnw7KU7MpW7M1V7M5B1mVhC2KxhdG4ktisYAs 1aYc5apk9ewQ8k8X6cXX8fquFoauU6ONt0p63gmiXrvMrEuijpVTp2vIQyUE aMt5YdiSWHRzFrKrECgJWRrvu/XGJCShvpdg2myGVru5yz1AuN1MB0HvSmVe DaJfAXeZfuWSL1ZLo4kdOdi+AuxCDR3ero1o14HsB+6DQDakAiWx0V3a683a tXH8+zc4gJol0QTA0fp0zYYMzeYsza5c9HQ5nae2tvLuDt5Ksx06tlapqYus 6Fwcgc69ic8PxxZGoz4Xhc27Rb93k34nlHk9iP7TJQbiABjy5jXqnRDq/Vv0 yih6TQy1LR05UNB/olx5rQHv0A6ojUPluP1kBb0lB91XiKT1ULxVZ3SK0oBd dA9RtuF6yr0mjn/3OrMgnFqVjAVkINtysOtN2rhOfUSr/nihbnkUtziCWRmP r0vVbM3SHC8jKkhrq9aZ0m4NCNO/cox//QQ37xKzIIhefI1ZdtMXjZUxzJII +o1gX1q9fhUi7MvW1XFMKWrvk9zt2oHwWvOqW9plN7hD2VxoIx/ezl9vZU9V 48crsd35+PJYEsDq3Rv07Gv028HM7GB6dgj9znVy7k1ybTK+twA9WYHebMK7 WEqyMqyJrcQBjqRstSFPYyxCjNkKw9Fc4b1g7pXz7GuXaCgzc2/hK+KIg8XE uRrycj0d3yOkKsXkfv2tNuZ8NXqyHNmchgHtzgVguYUviUaXxqKni1AFg9ls GC9isbXku0B8QfTWNDawlg1t5i7VMlsy8PWp2O5cIqiOjWjjYzt1xZipmrY2 shalaNJZRNaoz1UwZ6qx/aXIwVIE7nKqQhPRrOlAlbxW2Y+rD2dRq6IBbJm5 N+i3Q6g3r5FvBlNvXaXfusK8c41+/xazKILelMxXYwOIcVgpDRZqLFCczlbS B/LIranYhkRsXSy5MFQ7/5puwQ1ucSS+NBZZGY/szMf3FeMHS/Fj5cSZSvJk ORmQRM8JoYG/3g2hV8QyAcl0QCq1NQdKu/poOZamEAnbIG4bSuo27M0nNmWg BwrJ1H5jvsYMbopv8cxqA789kpLq6+vBw5UrVx4Z1IxNpqSvfOUr8PfcuXOj Tz2SkoaGhnp7ewEfHvkJF5rCGVMS2N27d5999lk49TBQvX37dn9/P8dx0+1s +vDDD3me7+joIAjC31iPNjj+96GkL3zhC9AmTtEzJDhEu6+vbyqjEEF7jC6N 8L9Cobh///5oP7JmnjdvnvwTkkWpVE7xkadeDMBMJhN4lsu81Wp9pHibrv8Z GGhIuMUjB7d88MEHkJIqleofsnT/pzk+lYhN0T4hOT7jWsvlckGGQmGY5KPK tJIRYkJRVFdX19/nKw00ms888wzE59e//vWYWhceSo7q/v37P/6NptjujDFI Lki6xz4jeAaRgYLxox/96OjRo1O/5BNCSfDWbJ3INBrNwy4Z37intarBRdVp 1sbh790kQAzPj8CWxgP7oIcK2au1Uki94VyZfkUsAei0LA4DnNmRj+wuRPYX Y0fKcHB78oltWeSWDGJzJh6QqQ7IVC1PQGaHksAR80PJxEZFpaKnqKf3fIH6 3VBydgi1OY06U0VfrKOPV1DAXItj0YBU7Hg5FVhPn6mkl0VR74ZSKyLI0Aq8 FUcwASVFQmsRdDapSiNsTeL+BHwUCOKfeiOYWhBGH81j0rrYnH4uX6UvQIw5 KmNUu351ErEgCtueRSd0ShWEpUBjvlwlnSgSD+Xp9uTRW3OQrbnI/iL8eBl5 uoI+W8FdrhCvVJhCq/QpLX3Z7e0Jje3Hsvvn3sDfDiHnheGLY5BlkCCJ2JYM enumT+UuiCDfuQ5p5aMPcEsiyUPZdEw9m9TMpLepczq7cjs7MzsUsc2GqCZL UJW4LZNck4yCAA6qIYrVTAXGlaL6ItSYqzYmdEuHCvhdWdyeHGZvAQmxOlBE HC1mjxfzJ0q0Z8uESxXihQr9/nxmQzoakK7ZlYeFtfNx3brLVdqV4eybF5j5 V6iLWWhYsSaqHA2vYm/UisHVhpOF4ns3mDeuULNHHgEQb0MqVYE7lEavUvJU a2wJdfrYGiGlTczslzKVUkqfGNLCXm1izlbRO3OogBQS/AOVbE/jNqeyK+Po 98JwcCsSoQCgu/LR46VYXDue30/kK6hCRIRnKcXNNYytVeesZxyXK6RlEezs a9ScUBIAc3EMujoJ3ZGLHRgpM9ebfb0/8V18Rj9brKHyldSlSnxjCrosbsTF +1L7TCmm5Gi7g9UZmLgGZu51gD56TQJ9oIA6XkadKIeiwpypZAPruMgOIbFX n9Cjj2zXhbXqIlp1cW18ajuX2s6ndOmSe/XJffq0fm2REitRInndSFI9FlOJ RVUSgRXaC+X6E8XCtgx2aSwBWL04mlifTG9MYbZnsIcLtCdLdGfK9LfqLFGN tvB6y/ky/Z4cam8OGViOxzUiyS1ofCMZUmG6Wm45nm9YHcvMvUXMCydWJRLr UogNacTWbGJPAbwd+KZ0YlU8uSYBCJq9Ui1Gt5rCmw2nKqnDZdiZaiJVIfYZ 3L2SO7lXAqTaX4gdKibPVuguVArgpvgWz6w28NsjKekvf/nLt7/97eeff/6R Qc3YZEo6deqUzEqjBwpOQkmZmZk/+clP/P1QX/ziF5cvXz6mha2qqnruuee+ //3vf/Ob35SHUgA4/PP/tl/96lejL3kYJT34SNT993//95jjZWVlo0flQUyg MXrYd8jS0lK4qawM7927d+zYsa9+9auzPrKvf/3r/t60uXPnQvP0L//yL+Bf PvvlL395TOQLCgomSdjpUtIU/YPAW7FihUy1YJAFP/zhD8csOw+Z+LWvfe0H P/gB0N/3vvc98Abpf+jQIVB9ixcvlvMCyhUoQP8lcvIuXLiQYZjZs2d//vOf 9/cwwoUT8uODqRUDv0VEREB6yj6ffPLJP/3pT52dnZOItyn6b2ho+OeJ7Omn n54wGkuXLoWzWVlZ8FDnzp372c9+5i8Av/3tb0Gyjr8EitPatWvldJPTfMy9 pjjvbGb2aY5/MnMcFD78M7pRgJocQRCCIEYH9cILL4C34ODg0QenVWtBpQch AMLA/62trZAs/vhDxC5duvRxkr2oqAgeZNYoe/bZZyEy0K5NGJnHYrGxsfK9 0tPTx5+V2x0okP6poJs3b4YUePnll8d73rRpE5yCwuw/Mt12B8dxKDmQHfLs Hni55LhB7vz85z9XKpWjb6fVauUQxs+A43lePjV6WtAMGsExNl3q+YRQEhRL WQjt3bs3NDTUr4va29sfdsl4fWUzK8DhrHJ3Cv7edfKtEBL4Ym2aZksOGtWh b+Gd3fqBMsK6pxBbn6nakqPeV4IcKkOOV2DnaongZgbc5Xr6fA19uoo8WkaM DGNTL41DQJy/cpl+N5Q+VU5fa6Kv1NN7c8kFEfj74fjuHDqwjr3WxJ2vYQIy 0FUpyJ4CIqxVm6kUY9r1y8O0b17k513lT+bpcxWGOtrYyFr69E6F5MrtswbE 6V6+wLxyiX4nlHz3Jr4+gUxo4SS7ZB8wMBZrv+Rs5p25KvO6ZN9THMjjKnA7 YR3UmDwVhLUAMaf2SyHN3IFSdH8pcqGOutXG3WrVXqjkV0bzbwVx74YwG9I0 23OVm7OUS2PQ14Oo1wJpkL5L4pBVycjmTOxSLXutkTtdAcCCLY5DFsUiIK0X hGObkvDEZlpv5u121mbqt0mNAl/frmzZlEAsDKcXRlALI7GFUT4E2JqFnqjA TlViR8vRA6XIQUjJSqAGbVib7noLf6WRuVRPna6i1iSSiyPJtQlMULVYTTlq KHtoo25rDr46VbMuQ7OjQL2rSL0pC1kUjQGjrY3DmhU9PNdhEHtokVWKjjbO ndtvXxbFvHUNUolYEouuTEJ25BC1tBMxD3IOr+iwWx2S2S5iBnO71lbP2iso a0q/lNCrj2jXQcqcqCBOVZHwsDdatFfq+b159KIo7P1wX6/fimTEF410zZYs zY5c9cEiLEMhleGWBtbXUcU5hwnrUHSrYUMqZAECgLk6GVmTopH77wIyfEMr oajAY4a3MbUEa3ZojXY+T0EeK0W25Gg2ZWlWJmtWJCLnygiVoHMMiIJFn9jM z79FvxlCLYjCISM2ZqB7C4jYLmAfKV3h68PK0RhS+sUjJcSOHHRzJrI+GVsV R2xIoqJbDB2CC7e6GZvNaiMddrQNwfcnU7MDmaU3uYR2cylqK9SYgxu0W7Px dWnI1mzsfA0V2MDcaOVT+6Q8tSmp27gxUbvgJjc3lJl/k1oQjq+KxRKbUFKH mK2YZGE0klMpefP67ZuThNeC6Nev0LNDyTk3yLlh5OIYbM3IxLdtOcjeIvRY OX6tiW1knaRtuN/gierkAxuoq400xLyBczRyjtQ+8Wojc6aKPFxCrowjFkf5 3BTf4pnVBn6byq6yoFjAz5i6+jGaLHhAeECLAP8AOPhPTUhJ9+/ff/fdd+V2 5KmnngLJ8d3vflf+CZQBLY7fp7z6xOT2H//xH6MDlw9OSEmgZ+DUO++8M/rg xo0b/UGBSvzsZz8r/w9NT21t7fhAQCnB2V/+8pd379596623xkTm3//93/0+ AZEeGXkIbZKE/VtQUnl5OTTo8t2/853vvPjii35xOzqb/KME5Q442UAvjXlk 0Mb+S2TN/LnPfU4OEEoFUKHf5+uvvz7mc/fUi8GDkS+i/jGTs0Z4U9aHfrE9 RrxNyz+Q78MyaMI0lJ/0+PHj8FCyty984Qv+S0Bejulzcblc//qv/zprRDWB 4ITL4XnH3KikpGTyjJuxfZrjn9gcf//99+EfYEb/hfL4KDB/R54/X0Zz9HRr Lagk4SyUBAhkfExGg8Z0kz0nJ0c+BUXlpZdeAhyTu3hkg+ZswvT8+LZv3z45 FyZcEichIUGOgH8Rm5UrV8LPX/ziF+M9y+s/jP56Nt12x4+Q/nlSAK2QIPL/ 8ILU1NT4PbMsKx+HQjgmJgzDyKdGZ+IMGsEx9kmgJGDDaY36IwhCVkHbtm1L SUkBnSP/D38n5GLZxk9tsJn7weGcwkdJN8jZodTSWHxrNrKvEEvpk/oNbsTi ARW9rwTdkK3cnq88VqY5U6m5UINeaSButLI3W7nQFuZaM32lkTpfQ47IVHR1 IvbeTeqNK/S8m8yZCia0mQ1uZA4V0qviyJWx5IF8JqiODWniLtYy23KwjZno wSISMAFEb2SruOyWj5LeA0rKl3KVllrG2sDZOwVXtziQ2WtbF6v743nmTxfp uTfJ+eH45hQytZ23uIxur4m32RQGPyURC6LQ/XlsGWpHzYNKo6eKtBajpkyl dKONO1KBHS5HrzQxMV266E7hcrVuSQT7xwsMYFFAKlCSamuWamks+voV8rUg al44DqJ9fTqyMxe70cxGdXBB9cz2bAy4A44vjcWWRGPb0/D0DnpkNT/Oae23 iU1arqFV0boxHn8/jFoYQS6NJpbF4msS8B05+IkK/GQldqAE3ZEPvKM5VIpG drIJPXxUJxfcBJREn6mi1yeTy2PIjclMaJ3YyDkbeWdYq7A7jwhIRwIykY05 6k25qrXpmoXRGEDc+nisA+mRxA6ruYc1+CiplXPn/P+UhC+LR1enaHbmEuW4 bWSBd4/osLk9ksstkiZTK2+rZewVpDVDKYFij+7SAvaeqsLP1hChLeytNh54 9kgRvTqBWB5HrErE1qSg61IR31ywbM2OHB8lZSqkcsIC2dStdwMiqY2Dka2G TenE4jjNsngfH61LUwek+/hoaw66Iw84kTxbTV1vpitR1mjXGm3aIhUUHnRf MbKzwEdVyxORM2VENy8YHRJjlJJadcuj2flhzJJYYmUSGpCO7sn/H0qCOOeq jfmIMbVfPFiMb85C1qVqlicAvaIrYomoFqlfdOmcbsll8w6QwwNoN4bvSqT+ dIF5P5RL6zLV0dYqwnSzWbs9x0dJUBqBXEJaGMiLPLW+Apey+vRrYrjZwQwQ 0OwQYn4YtiIGTWhCgJJsNsxsYzj7AGMfKkec21L+h5LmXKeg8M8PI5fE+Jhu bRqyIw85XIqcrcbD29kuwS4ODOKWgZgubWA9HdTAJPToK0kbuIQe8Uo9e7qS OlREQqGaH+5zE7654ylpZrWB36ZCSRqNBvxAs/LI0GZmcitZVlZWXFwsNxD+ L8MTUtLhw4fluv3IkSP+MQk9PT0yVoz+3AckYvrIoAmGs4sWLTL9bwMVMTpw OeTxlAQNELRWcOrkyZP+g5Dgsv/f/e534OHByAgZIJcvfelLs0aWxRs/okam pG9/+9uycHrhhRegFYM4wIWQlYCKfp9wUI6hX/l0dXWNiTw84CQJ+9gpyWg0 yoIZhFlbW5t8EGSGP0f83zD92gwEWFVVFc/zsuoDO336NOTv3LlzZ4181fSP mJITRFYFt27dcjgccJCmaSBK+fiYzoupF4MHI5PrZc9A4lCe4aZwSVJSkr+f box4m5Z/SIEx+XLhwgXZ24TJ6H9SsLVr1+r1+gcjY2xk0QsGIm20/4CAADlZ /N89QLzJMAI3miS/Pr59muOf5BwPDg6eNQIg/g7oLVu2yAH6qbmiokI+4uem GdRaMiUtW7YMMvczn/nMzp07EQQBshsYGACtPrr3f7rJ/uMf/xgOPv/886On vnIcB6kE8Dimcn6MNn/+fLjvT3/60wnPQrLLsfWj5bQoabrtjp+S5GLc19f3 l7/85fbt22FhYTIrQfr4vxhMl5Jm0AiOsU8CJUGLDOU8Li5uiv5PnTo14QAb sN27dz/sqvH6KqtDCS6uUbUuDnvvFr4okjpUwMd3SZkKY6vWyTqHGedQq865 v8jXl3S4qD+5taOop7Wsr7NIgeaqxVyNVIYJtSRfR/IVmDapWx/XKQbX6XZm 0aviia1pZHYvV4fz9aQ2XyHGtpkiW0zXGnQX6+gLtSSoxPBWIapdH9qoP1Ws O5ir252pXXiLmXONWh7BhtQYmzmX2uQBUmvgbHWsLa7dtDpa+4dzzKuX6WWx xNpkbG8ekdihRSQzabJ06SzVtKUQNSd0SxvSfIMGt2QSgdX6qFZjZIshtJm/ 1swFN7Mhzez1Fg7Ef0qfCJ7z1eboNsP6BPbNYHLhLSKsVpXZoUhuU5wuUi8Z mYsEav9QEX22gg5pYGpwtpXhitVsaANzsow+UUofLqIPFtLnytmsXi1mkEiT 2ITj+Z29aS3diU2KazW6oCrpSrV4tVYfXKe/0ahP7jGW4pYizBzcyO/IxbZk aY4Vq6pUve14T1Gf4mIlAvofRHJEPZvYzKW38/kKqYK0AIMkdkvBDfz5KvZE Gb0uFV2ZBACCvB/hy7LVsVhWJ1KPIg0o/MPdrDVcLjMdy5MW3KLl4XYAdOsz 1EAoV+t04S1iUpehHDEoBEkhiCWIPrZLF96mi+nUZakNuRpjXI/+bDUBwHKg BDlZiZ+pIi7UUMEN2vAWA7jgeuFCNXehmgmup5I7sLxeTX4fVoaLFaQJniuz zxLVbL7ZYDxWzG3MwNakatamaaDYBGSqoPyEtwnp/UYoWofy2W3p9K4sKrCS yuimcvrofKWuEDHkqozR7fp1KfjiGHR3Np7QyRRruBKET+vW32owhtQaz5Tz +4uIPYXYviLiRCl7qoyDJ0rrNxRj5iyl4Vg5sSMf2Zyj2ZAx0oOTiaZ3MaRB Z3Xq7E7W69vySdFHqo7nou+H4xsSiSoNrRHYbpaJaCADkvFFMdi6FAIQGNg/ uZtowfo1TFezqmdrIjY3lHzrGvVeGA6Fak2y5lKlJrMHqUCwJooVnEMm770m 2rMvW/fOdUhtakcmdyCfP1DA7sgh1qdpAA8PleAX66gbLUxKD43oaKuNYkTq Wj1xuBjfX0icqmCu1uvAHS9ld+VQWzLIjWkEAPiyBJ+b4ls8s9rAb1OhJLCf /exn0EY/bAzMxzSZkrKzs6FJlQWSf0D4eEpSKBTyx0lQSmPCgcZFDqq1tXX8 Xf7rv/5r1ozmJUGDBZJDbs1B40EE5OODg4PyYLnnnntOXobdbyAd5XB27Ngx JnyZkmQDjTSV6QbQUsv+/9bzkh7pf968eRCNZ555RpKkMadkDfzqq6/KP/2a 2T/RDGQP/HzllVdkkZycnCx7kLXxg4+U5He/+10URUeH7PF45C4GyEH/wWkV A9CQstT//ve/PybB4V7jxdt0/Y830PyP1MwQyTHraIHa/OIXvzi+2Mi4cejQ odEHZTn6ne98Z5JofHz7NMc/yTnuF/N+VoW6SNbVBw8elI/Iy5oBSMo/Z1Zr yZQ0a2SV7Ak7m2SbbjJCBOQ8nVBUP3JdtY9j8qDH0cPkRpsgCHJs/fk1LUoa bVNpd/yUdODAgTFPffr0afmUn9emS0nTjcx4+yRQEhgkDrxB8fHxj/QJRPkw USQPuXnYheM/R+/O0YDbnoksjgTJja2Oo2/UG3pFD2Yd1A3csQzdEz13uvUD IOfWpavPlfa0KOtpupLn6tWcupm1NHM21GCU7HqTU9DbJYXk7BUHyjBLYB2z twA9U4H1cazeojXZ9ZzVipi8fZI3VSGdrSWPV2JXGulqytbKO9N7rDtSdHND 2LmhzHu3fMI+IJkGpEJNQ6L7DmUbqqIsxZgprElaEaH9w1n2tcvMxjR8dz56 ogKP6dTWMxANaxluylZJyb1iWKtuYya2MhlZnYIGpILmpDalkzvy0N2FyKFS LLiJSVPoc9RSJWlt4Z0NrDOr37Ijg5t7i1gXj9crFRTfp6F7k1sVmzPV69M1 x8rIkCYeICJHqdWaOYuDw0Q+T6WL6hAi24UbzVrgr/A2bY5KauEtTZw5po07 WqjZn6s+W4YXqM2VhN23ERVpqyAB9OwKg1d03+Uct5O6jaCiA1LVpwr7MLLO pKtuUzWdKepflYTuzMaaCZ6WBNao7+SlQtS3KkK2ypDRLyX2iNebdKsSiPcj MR8iheFAScuiiZB6Or6TjW5ljxdwq6L5BTf5+Te5OdfJ2dd9s8nWpKk35yg3 56i2ZOBb0ol9uUB8fBEqgbvVyh4rxw6VoudqyByNsZQwJ/XpT1TiW/NU2/JU OwvUuwvVxyvwhG59h3agW3BX4NZ0hSG1TyjU8BotrjdpOAPWyPgoKVtpDqwW AxK41bHMumR8Q4ZmY5bK53KU4M7VkrW0nbQNdWrde7J0iyOZ+eHUmkR8dy5y tBjLURowyyBjH26iXRt9077wtUn48VL8Yi1+vZnKUopt/EArP5DaJ12uo0+U E4eKfXN/ViUS+/OYhG5DGW7JVRvPVJH7S9BdhcjOQs3OAs3BEk2hQq03qL0O ldeh8Np6B209SqrvYoVybZp6X766l1IZTCpKq46pQ1dGk3NuUMtj6Sv1XFgb n92DoGSthS/C0PK9yYoFN7E514lF0ShwX0CGel+h+lSVJqgBS+jhDJ47nnt/ 7hEGjxZo54djy2PxwBpteJseCuGpSmpTthrcqUoSqDyhhy9UkbS2e8DUzPNt 50sVWzLUa1OQgFQMCjM4yNZF0cSCKIBE35DXdRlqcBO+uWMoaca1gd+mSElX r16d9b/HbDxGkxVORkbGg5FB1LNGPqKaTKYHE1HSpUuXZo2MlJiwGwWa/tFN 22ibFiWBkHhhxEDC+Uc+jAkZJJl8MC8vb3w4b7zxBpwCthpz3E9JL7300phh Ng+zvw8l3bt3D5509NCOMQaMLPemTdh/AYkgy2n5p18z+/UbiOdZI7PP5J8t LS2yBzmjH0y64hkU5lkjg6z83RDTKgZNTU3yvcb3rk649NZ0/Y+3qWjm0WPP /CYPuQEA8R8BJSm/IElJSaN9+hPwb7cJ5qc5/jD/4+0fkuOQFPCOz/poDwU4 DhcCnEK6/frXv5b9vPPOO+Bh9erV8s+Z1Vp+Spp81NMMkvH3v//9rJEpbCdO nJhu/fZxTJ469/7770941m63y7ENDAyUj/x9KGn8sHZ/uvm3yfj7U9LChQsD AgKm7v9vNy9p3759UMLHdLyON5IkJ9FFk/S/j/8KvToFaMI372ZZLLE0mtyW xiZ0GFmH1+Adsg7dcd2+Z/beUUgDx0qwrVnKq5WdKrTGqi1z6Gs4UdUj2nol p9ZmHBjQDXp0A26Jd3oYx1Cbzh7Rxpyu1IQ0opSBcw3ohryS02s3eod41yCI 88B6/Gw1cqOF7NJZEaOzWGXZlczPuUq9G0z5lteLQjenU0ldJso6bB68xzmH qkljMSqGNwmrIrk/nWfeCqK3ZeEHitGzVUR0pxZ4p4ayFiGGrH4huUcb3sru ysc3ZWEb0rG1yejqBHRNErIpS7MlV7O/GA1togs1QjkutoxMd+rSOovV5iN5 9PJoZFeaRkGqbGalKCkKe1V78zXbc5Hz1XRkhzatTyhDdTYHN+zh9FZtDaHP UErpCjGxR5fQpU3u1RYhIiBSHW0ObeS3ZGEBaeihIrqKtLfrXG06Z6vPOTpF F24dMnk+EJz3MnpN+/OJHZmqy6U9OqbSK5UpkLpLJb3rkpEDeahaxztcgmNA rxTFPETMVEt5iLEINeepTbEdEiAk5NTSaGJ5FDh8XTxxrpK91qgNrNXuyuQW 3WTmXmPeC6UXRtCLIinfiui5yN5i4B3N2lRkZSKyMR2/WM0BdgEqBjex+4p8 S1KfrMALEX01CSymhazZVdAPDhBpT5HmVBWeqdCTlgHW5mnX2otxE+RgPS3o zfjggNruQDt5oY6S8lTSpQrt8mhqUQSxPgXbma/ZU6TeVajekqvanKu6XE/3 6J3WwTuEafBQtrA8kllwC6gEX5/mG8qYpzLqXUNW722FMLArnVgRjWxIRg4X +yZwXW2i8hEDavFiFm8JbrrVzl+qp09V0GuTiRXxxN5cJr5LKsXNuRoj3AL4 7kg5JrvTVWi5Si0aFB57v8fW57b1uq09KqYvtK5/V77ydLkS4frsll6d0J/S iAbE0QvC2LXx7NV6Higpt1fDUJUeXa6WKDqR1b86GlsSSaxORDdmaTbnaCAl D5UhZ2rwyC7e6Bkevn9fJbrPl3Gr4pDNKRiwM5SNVIU+uJHdW4juKUQv1zFx Xbosha5SQ/K6dq+xRuAaLpf27chUBwAljbj1KchK36r46JJIdGUcui0H2Znv c1N5i2dcG/htipRkMBieeOKJ5cuXPzLAGdhoSgL73e9+N+sjMpLVxWhKghp7 1siovLUTmfyZdMJOtGlR0hgDKTJnzpzu7u7RPsPDw+WzVqt1fDjyCBxItDGi zk9JU5/n9fehpMTERPA/ieT2fw0GYTM+5V999dXREu6Rmtmv1qaimcPCwmTP /tnl0yoG/sv99/LbhOJtuv7H21Q084RPKvfd+DtoZJMRYIzwuHHjxqyRSSWT RONj2qc5/jD/4+0flePyugegSx+MLMgA/0dGRv7yl78E7pA36Zan+fj5cWa1 lkxJj1yzfQbJKAiCf7GCWSO9h1evXh2z+sTfwl5++eVZI/OAJjyLIIgcn8zM TPnIP4qSwOSyvWXLFvnn35+SIEe+8pWvjNmkYxL7m67esGfPHhAM0FpN4geq mhn3JY05sjEDBbc5E9uTze7P1Z4rFQoUBsFuNbvtjkH3wO1h2+AgbrLdqFNe KOtObGoliCq7rswp1uoNGtTkxExuySYOuEjvAO5ysYLTwzuH+0RHVh8Z0axM 6dLwFmHAIw0OmVyDNovXLbpcDZQ2vl0d2aJI69KodVrKIDQgfGARsTMB3Zbg k4hL49CtGWRCpxExD+ndtxm7u5XmGwg6uZXemEC9dYV67zq1Kxc/UoqdryVB f1ZRVt8EE0xbrsaLFEhmN3apBhiNPVVGnyzGThQgxwuRIyXo4VL0TBUa14HX 4WQLRfVqdYhkUOkNjYQ2tEJzJLMnsKiP4FGnnTBayEo1caYCP1aGX21go7t0 H1ESO+zxTaVp44xlhAXEeZGGL1SzJRqmjuC6eb6N5SOaqd056OYM9EgxVYL5 1jQAV8vYwNUx9hbW3ckNtdLetA4hsFx9ubQ3rq5D4qu9hgqMrI+r7zlVpL5W jZB60umk7U5WIehyNWKmSipETeWktQS3pPebjhZqfRv+ZnOHcqgjOeTRPOp0 MXe2lIe/R3LYPWnM7hR6Tyq7L1u7P0d3tIi7VE0GN2KX67B9+fj2LGJ/PhXa qAPmKsbM4e38qUr8eDl6pQ6rxal2mqrQYFHNvUHVnUE13XDJ5XryVgtdjvKc Sa+ziH2CsY621FDmDk6UzJpBR6/T1q8SOAClGlyIa+WO5lIHs6kzpXhgLRpU j5yvQYCVdhSoghsJuNbucbAW6/VK/mgWvS+dOpBLHSohT1dS5Ygg2cx2lxkT xKAS1dGs/rOFqusNWEQrkdTDVlNG3OoFV8fY0pVSXJcQ1qo7VsQdzOcuVOqS uqVCxJSrMkZ16m628aEt3LUm9kojc72FqUEJzoBZrAg4s1ljNqtVrCq5w7fD bGRzP63tcVq6JbG/pAs/XaDdk6k/nC9ca+RvtHLZvQjN1HvEUpGpDK9Un8wj D2bT+3LxPQXI3kIEWAxI7Xork9anNbtsw8NOUjJF1xPHc1Xni9BcpVRJWMoI c0qPLrgev1qPx3XyOWpDGSY1Uaxe7PKa6yWhKb5JEVihOVOCnilBRpzmeJ7y UEY/uGO5qkvVVGAdA24qb/GMawO/TZGSwF577bXPfe5zU68qp25jKEke+AGK XRRFedz+aEqCdmrWo2zbtm3j7zItSoI7to8YkBEQ4oTL5PpnSUwYTlJSknx2 TIvvp6SprNMu29+akkCogGYAzxOuWeG3SWasjzZZfT1ezeyfSUGSpHxkWsVA Xn5kwpyaULxN1/94m7Fmlme+j9HMR44cmTXyvd3//b+jo0Oe6LFq1apJovEx 7dMcf5j/8faPynGo5Gd9NAxPvqqnp2fTpk2zRtbDgRpDjpV/nPDMai2Zkh4p fWeWjPfu3bt+/fpPf/rT0VkJ1XVhYeHfbgvvdevWzRpZN2PCs/4qV17W78E/ lJLkbq8lS5bIP//+lORyuV555ZWnn34aCs9U/M+AkoaGhs5MzU6fPv3kk0+C ZhjT0zradDodROBhumiSjUXGU9LxChzcpTq6nnb0S17c5BItWre112vrGXLT w0PGoSGze4AT+VodUyZxZU6h3K0v9xgaPE7Kc+eu584dj5PwGOs8UqXL1GZ0 uw2eewanXTSqRH2LwdA9MGj33hkevHt7EHjLq/W6Gbul3yC0iNoGg7bOoa9w 6sutQrXANXFsWz/WfrZMsS1bvacAi2rXN3FOleTCJMmkbwI/nZrGsyV9q32j npDjFeS5Wiq4iUvpE+toWwtrRth+PVtu5Er02kbCZMPMg6TRykuEoO+ndaoy DZelEPKVXBfRZ9RWWYUKh6EJRLLT0mM1tOrpUg7L0dFlA27D0N1B5/BtxDwA EJSPGEGWJ/TqU/qEEh8lMcNexuPRmwD3PHdEt9dsYywW1GpW2owddrFO0tYX dHeerlDtL0FPVhFJvfpMX5eTlKYQU/vF6Hb98UL9snBhVQQfU6FQaBp1XK2o rR2A1DPVOQ2NktipE3pEscdl7vRYOy2m3h4Gy+gX4NpK0tqlH1BI7j7J3aZz tWpdPYKFNugEE61gmD3x3IIr/KrrXFgZ00dSWonmjDrG6qVttzmbR++0GJwS YxFzlEJspxDfrQdE6hIGwFUQliyllNmvLVPjAldl1ZaY+VKRLReYCi3bgBtE xOTCTVadpHYIkPVlklHDWuyMza2zGO2mTq+hCrLeZe132jU2G2qyEIKRFoyM aNVKDoG36Xt02sA64lg5FtWqQbl+j61nwNonGQGKcEZP9XK6KtJcQxoxXmUT G9xSlUNfKTBlPFUm8E2ShTA6dAanQWe3U7ZByjYEAC547gju25xzWCF5e/We Rs6ZozIk9Pq6b5QmN+scglOI2dupd3Xond16R6/e1udz1l7R2qu3qEU9p9eI Yq9B7BmwdHitbQPWHqOFYy1uxDjUyrtCm7mLdVRyF04KqkFHv9uuNFpY0awj RKEK1Ya1+ZYryVIZAHg7dHalINnNXV5Lq9PYIgmtWr5FEHpsbqdz+K7z9l2r x2VwiJJd5G1WyuamrG7eZnFYNYO2Tre122TBJDOpN1N6E6E3onqDWqdt5alK nqzUcm2CwyUM3AE3lbd4xrWB36ZOSYmJieBz8m84M7MxlAQmf6YGzQOSYNb/ piR5NsSPfvQj5cNtwj1w5QbCP/LkYSa3MpPzgmzQssueJ+SdwMDAWSMzEcZs LvNxKInn+SleItsUKUmtVj/11FNf/vKXQYlN4g2Elj9xJkl8Wdg8Xs18/vz5 WSOfuP0LUk2rGPi3jxwcHBwT8oTibbr+x9vj1cwffPDBCy+8IAcIuu4HP/iB /Mr88Ic/fOSGOx/HPs3xh/kfb/+oHM/NzZW9QTx///vfw4t8//59GXYg/eWz n/vc5z788EPZ/8xqrSlS0sdMRo7jbty48dprr/kXxPN3oDx2kwsnGEDH+LN+ 3INoy0dkSnrppZfGe54KJU3e7kxCSfByyYMqjx8/Lh/xU1JxcfEYz1OkpEc2 guPjcOzYsVnjFkh5mM2AkqA0vjZlk59xcmQDmJKH04zXReOn3flttLf/Caea ABfczPRLbq3rttntcTg5r7V90No2PIDeGRJvD0nDbtprqPToizxisUcsB+c1 NQ4N0Lc/uH/7/r0hF+YxVLjFYpex0TzgNnnvWQfsA7Z+r7nRa+scHHYN3rs3 ePfu0G0bIMawhxiy93hNDR5DjcdQ6dUXeoR8r1jiNdYOmpv0upbAqr6deco9 RWhEu66esfeKDkQUHGKNRyxVojWXKrsDslSbc9WnqkhQsyHNHBBEHWNr4cwk 123jC926XKdUaxpwGb13rW6n20UMOfrsVmUTpStFpQpUq6J7nEKJV18Ed/Ra 2uBJfZHR57v5NI++EJBw+P49150PSPtQPWurpCw5GmMiUFKvUIJofZTkoQcH RdeQ13H7A+fwsNdND7o0g46+QXOLVyy3aStKe9rOVSsPlaGQqvHdPsBJ7tPL G/rcaBZ2pWtfu8TODqSjS3tYsg5S1Wus8prqvKZ6r7lpEJLd1jVoBdc2aGm1 GDu6aSS9X0jpl6ooW580gJi8PmfxohYva3M4B4RhD6OVmB1R/Dvn+MVBXGwF rRXJIQ816NW5bt923vnAfef24G377dsmu8dYjovpCn26QqykgFXc/ZKnjrUX osYCjVCH4Wa+xK3LcQt5IxldMiDVGhxG/YBXdDotJsWANgec2aTQOxyCyyvZ jU5Tu0cqg0cYtPcMulRDA8iwG4f0Gfayt4d0t28bXF4DadIHN5InK/HYVg3O dcJDwTMOuzTg0+WiWbMeWKNda6F1/XahwiMUeIRCyAVwg6a6ITc5PCh4h0zm AQdtH6TsQwbPXXiigTsf2ofv887brGMYniJXY4jt1gHMMo5h2/B9yBfBfQfS R2Py9BncgJYdgqtDcLbrfE4tGi0WZMjeO2Tr9qW2pdUL4DagdQ4NQrkFvLre wp2vJRO7CEqPDrmQoQF0eFB7e1Bvc+k7OV1cFx/dqS3BzEBhaqOTNEpOY7Ov IPlysEEu8HfuDtz98IO7H354557n9rBhaMhg8zr1A0OCa1By2lwObNDePWjv HXZjwx5y2EMNuwl40YZc6kFLs+9FAGdq9twe9N77M7gJ39zx3zpmVhv4beqU JLe8V69efaTP6dp4SgK5PmtksSx5OMpoSpJns0LDMZWtLUebTF5z5syZ3Jtf Fj4ywIaGBtlzRUXF+LOgiGb975W9ZZsBJfnlZV9f3xQvkW3qI+68Xu+ePXue fvrpSaZKgV6Vt9vwt9eT2OPVzG+//bYsEf1HplUM/EpyPAZOKN6m63+8PV7N XFhY+OSTTz7zzDN//OMfv/SlLwHPgtq5ePHiI1/b//N//k9PT09bW5tfJE/L Ps3xh/kfb/+oHJckya+ZAZH+8Ic/PPhISM+ePVtONPmgbDOrtaZISR8/GWWj KMq/zuGEFPPgY5ftrq4uOfzxi4EAon7jG9+AUy+++KL/oNw9961vfWt8UPJW rQ+jpKm0O5NQkn+LW/9UL6PRKB+JjY0d49m///jDKGmKjeAYg6Yf2ujR669O bn/TEXeZmZlPPPHEzp07J/cGVZZ/RA3A+/nz5/26CFq6h101Xl9drKfAhbZw TZxTafBiJhctsYLQoRfajWbM4hRtLslio426WgNfLvEVAlMlMNU6rlkw0HrH sN4xJEiYlqvWMeWCrtnsclg8wxaX1WpRWY1tFlOPYLNrHcOCc8g6YB30sINu 0m7pM+ib9No6gavhqTKeLNHSFQLfAIikodrOlvdvzVbtyMNCGrWFGksdaW0l dRTVwFOVjf31Z0p712Wqt+Qil+qY0GYusl2b1i+W4qZKQuohemiiREcW6bgG yTUAitrioyRqyKmwWdX1BIQmliJaBa106mvdYrVFbNbrO3VCp5ZvZvESBslh 8TLB4psgwzvvKo2eBs5ezdgKMBMo8AylvgTV0QZGNFOSTW/xeKxD963eIclC i0aNXurT61oFtooiq7Pa20+WK/cXoycqiLgRSgJEiuoUwtuFoDrd1jT+tcvs 24F0WEk/ijdbxQa71Og2t3kt7U5ThyR0a7U9el2Py9zmsbRaTZ39LJqnFrNV hnLC0sQ52rROcO2Czykkq2DVWh0MoWOOJOtWhgibw3VpdawgkYNuCpQ/b/cy 9tuc3SM6LAaHxFmkUkzMVIoZSqkEh9CcEGAJboafmQptmcbXl2TRlhi5MoEp h9RmmUZUlDQGF2qwsjqVxBaDo7Uqld6q0Ls0gkRznVqmSsvWCmKfIKklI2K3 Y4MectBD2128waZnzWI3r7vSSJ6qwiOaNR1oN8+183wnr9doRZz+qC+pijD2 kgqaqtZRpQJTatFVWIUKs9iiNxKCSceZJFSyduvdPaKHcww5hu+679yzD90F biKsg0ArOWpDTLcuTSkC4wgDt0X3HSDcfgP4H2gXXC1aRxPvaOQd8qDHNlYk BVTQ9+l93UmdBrHdaOw12nUmz6AwcEdh9AAlXailEjpJtRa32TCrDbc5tVan XmvW1xG+KW9hrVwxauwRHajJThhFSd9uFBoNQqNO26LVtujFHs/wwPD9D25/ 8OHgHY/ba3R6jFqbXWN0q6UB0mi12qlBp8rjUJusuGQhJCtptRPA8h4X5rG0 ewyV4NyWdseQ1377A3BTfItnVhv4beqU9Prrr0PtNGE3zce08ZQEBk38rI9s NCU1NzfLB9etWzetu8hDLL773e9OvlKfHPhUKMm/PhU06GNGhvT29srhrF+/ fsxVM6Ak/4JLMTExU7xEtmnNS3I4HJARkLyT+JF3k/z85z//yIkDM9bM/iXR /ObXMxs3bvQfnFYxoGla9ixP3xht/iFGo8XbdP2Pt8ermZ9//vmHydpJDMok KDc5GtPVRX77NMc/+Tku798qh+PX/M8+++zTTz8tr3c9eq28mdVaU6SkGSRj fX39hBWyfyGICZcMeixlW0aGf/7nfx690/GDkWVj5ZBHL1Uhd25CDQm18WjP gCryMn0Po6SptDt+Shpf/cLTzRoZe+kfbv2Xv/xF/nYxZkUFu93uR8uHUdIU G8Extnz58kdOSRttfztKgtbzySefBD0wpsYYb4BRDxtjM2bpyNE2/nN0ap8I LqFbf6mGP13OnSilT5VgZ0uUF0qV4U1EWi+XpeQz+5nULk1KpyqmWRlY2Xeu tPdUkeJILrk3Xb8vXX8sFz9T3HelsjuxVWGw8l6vALJTKWhbaL4K5S+VCYdz 9edL9GUKncPFuwa4FhKLae0Pru0NrOg9lt13OKPveG7f+ZK+y5V9Z8v6A1J8 a1wvjcPWJpFb0tjt6eyuDOpQtupItmJvpnJDqnpVsmZ7Dl6CWjp1LkCGTJV0 qZ48V42fq1CcK+kKKu+ObUYIq0f03DF5BjxuYdhNWe10noqN6eQSe/gmRm92 GY0OQ6WaCa1BzhSrTuQr9ycrdsX170lQHsljjhdpL1bqk7tNzZyjXedsAY3N 2mtpW5HGeKWWOF+BRrWxbbyDsg8pjQPxnWRIgyaoRnGpqv98Wc+Zkt49Ocr1 qZr1qdjuXDK+S5+hkCLbhMNF9I4sMiCFXBBOvXKJeTOIOZFH5fRQ9TjVxXI2 j8077ERFc0QddTgbuVyixuk+h6lnAHDEwqmNvr2iMpSGwHr2VCV1qoI6WU6D u1LHpPRwhWq+UK0rUlqLlI5qxIYIksPJOZ00qWeu1nAnSiBP2YtVZGAtHtJA 5ml8tFWMmaM7hNPl9PFS6kQ5cbISP1uNX2sgKjREA07k96JXy5XHcvqO5ymO F5EnSphzFXRYI57do87qVl+rpA5ma3el6fZlssfyVaeLek8X954pUZ8p0UBi 1mgQhwNx2NFKJXGlgj5eSJ8soc/VkBfqyNPlxIFMfHcS5nPJ1O5kek8qvT+H OlRMHi4hjhWqT+T1nS3sDalUVGqIRoLK7aYvFLEHM7i9adyBbPZwIXOmnK1E RYPT4h6ymQdsXYKjmXdUUtaUPoBQbWy3Lg8xFmMmeDrgplSFmNLvc3JfXnyP ENLKBjczQVBUytHTxeoLZZqYViSpE8nsIypwoU1rb9M5q2jLzVbuSiMd0c7m qbkqQluBa/PVfI6SS+lhr9RRu/Ox/YV4bj/LmgWzU08btQUKLL0biWtVB5ar ThcqQqpRwuyxD9933r6vdXr7RWs7b07uFM+WaU+XaiObRdLouHPXY3W7kjvp qzVoSB1eoNQiRittsZhsnNuBepwam51USbZe0QVuwjd3PCXNrDbw2xQpCYQN tAVvvfXWIwOcgU1ISf39/bM+sjFDB+XPemArVqwYTW0Mw2zYsMG/EPEYi4uL k686fPiwv424f/8+tFCjvcl+pkJJD0bNVgamk2dnQO0NYPL1r3991shCeeOn SM+AkiC28srA//Zv/zZ6zWSdTidvvPIwe+wrgUMKy7thfvvb387MzPQ3VR9+ +CGooN/+9rf+ja5mrJnB/vjHP2o0mgcjiVlTUyNPP//Sl77kn8gv27SKgfy9 d9bItl/yIC6IqjyPY0LxNl3/Y+zxamZ5+vbatWv9a2hPxfzLRMvmz5pp2ac5 /snPcX/MZ42aq7Jo0aKHxXwGtdYUKenBNJNRRqef/vSnxcXFo3uxBwcH/VvT TriVz2Mp2/6Vz7/5zW+Wl5dDIwjVqTy0bNbIQn+jF+UuKSmRj7/99tvyciU8 zx89etQfh4dR0lTaHT8lAcBev35dXjrD6/WuWbNGPu7fGkM2efv1z372s4WF hQ9GljqE7Bu9C/nDKGmKjeAY+4SsBJ6dnQ2ItGXLlkci0oOROXqyCjp79mx3 dzdoIb8umqRnfLy+ylUbwSX3SQcLqc0ZeEAaGpCGbEjTbMnQnCwnQpuZsDbu Vhsb2kyGNBEXqrFdecimDM26FGRZDDHvBgNuRQwO8LInV3WlRiNaqCEvY7Bx HbxYgRuyFdL2VG5ZJLs5kUlr4+1OnculLUfIC7XqfcWqnXmqlTGapZHIqjjN 5gzVjlzl1izVsjhkYZRvV9AFkfiCCBLcwkhieQwKbkUcsjwBWZEElEQ0MA7C Oqg2ebJUhiPl2N5izY5c9ZZM5e4czaVqCrF4RyjJ7fGIwx7WYmezFGx4OxvT 7Vs23DgwKLm8WfDIBcT6FGR1ArrgJvZuMPFeCAEP5VsULoO+1SS18K5u/UDn yJCtJt5ZiJj3F+CbM5FzlUw1ZUMt3k69M7CePFii2Zmv3J6r2pKl2pypXguU F4esTMC2Z5Hx3b55SbdadLtyyTXJ2JJY7N0bxCsXmTcDmYM5XHynrhDR19NG y+Bt7917fYL7dJF2aRS+IxVVEP02U6/brjLZtYzdrTF7knulw6XE1hwE3JYs FNzBIuJGM5vQrU3rFxpZZ5fgVkkuwWZ0uzm7g1bz1L4cfE0SDry2NRvZkefb tSdHZaimrKWYZWQfVXRTpm/D071FyKEy5FwNnq0QihF9XDu/NxtfGatZFYes S0HXp6Jbs/DT5WRkGxPewhzMY5ZGMfNu0ZAvkG7rU9UBqeoRKtQczNPk9Wjs vglKSFYHtisdXx1PbEwlzlT7OisPF1MrY6i5ISPuGg1u/nV6WYxvjbuAdN/l a5M0G1ORA3l4Vr+uFBWjW4QtSdyiW+y86/SSSHJNErYtE8/t40S75Bkymlzm Ft5aQ9tKMHNirz6yQxvVqUvq1af064GJ4nqEyA7dyEFtdKcW/rnZyp6tIU5V 4Yd9SzUi61OgeCPnq/HgRiK8nclU6qspSx1jK8HNN9t8e/tCgU/pE7JVkH36 mC4OoCmkmTleTmzMRLbnIDl9lGDROgZ0jIlP7KZvtlKXaohdOcj6ZPXBPEIh eSyD9+y371O2wVatHRL8Wr2wJZ3elEqdL9dppMG7H3xoct++Vsfty8MOFeIJ nUKf3oUZByS7YWCAdQ/QZjvfpbM083ZwU3yLZ1Yb+G2KlCR78y/483htQkoC 82+5OIaShoeH//CHP8in4Nof/OAHv/vd7771rW/JR0BcTbhWMLQF/m0uQQk8 99xzL774oqwARVH0e5M9TJGS/vrXv/ojOWtkpIq8XciskbkABQUF4y+ZASU9 GLVXIzQToFheeOEFICb4CeJwkqseOyU9GIm/nGhg8kAgiMxTTz0lH7l48aLs bcaaWd7aEgw0m6wV5acev03GtIoBSBp5FM2skQ+zX/va1+T/v//978uf1sco yen6H2OPVzPv3LnTX8ZA0f1wxKAYQOLPnz8/JiZm/DSQByND/f35AiYLqhnY pzn+Cc9x/1SjWaP2opIXxJPNYDCMDnwGtdbUKWlayQjVmt/DE088ATofePmV V17xF4kzZ85MeJfHVbaBGuR+mTEG4KbT6Ub7hET7yU9+4i97MsjPGhmVJ5fz h1HSVNqd0bvKyun2ve99T+6ikm8xZtEkaIj9niGt5OX6wT9kunyvh1HSFBvB MfZJoCR5DOrGjRungkgPRnah9Quh3bt3j/56PHpf+DE2Xl8l9xjARbWLWzOJ FYnokjhkSSwKbkUCeqQIvdmERrehkW14cDN9tYnxbQGTia1MQhbHonNvEa8F Ma9foZfH4NszwTNyrQZVc76lAzCBbaN1NYQ+TyFsSqbnh5GrY6mYRlZn1umt 2jwldbIK2VGo2ZSteT8Cm3sLXxCJrUpRr89QrU1TL4xG50eg74Vhc24Ss307 /pBzbhALo/DF0fiSGHxxLAIx3JROlKGOXr23XTcQ3yPuLkK35qmBuYDUDhci QTVUh86pMXlxo4Mz6QUjiwtMfCcd3ETdbOVACZO2QcwyGNth2JZJLPVBGfpW MPXqZRqeZX4EtjAGxD9xtU5XS9mbOV9HUgNrr6KsGQrD1hx8VarmaCldhJn7 DAMgYoMbsRMVigPFfdtzFOtS1WtSNCuTNMsTkVXJKKBcdIeQ0qcPbdRuSiOX xGDzwvHZIeQrl5g3gphdmdz1Zp+wz0eMhHVY67pbSw4cyNfOi8A3JGFtiEoQ lSazhjXqFNJAh86d2COdqiD3FWJ78rEt2cjmLORAER7SzMV265L69DUQVd7V qXMgokFr5BiRbsPpIwXU1gxqcyaxIcO3teuOfCS+S8pTmTP7TJdr+PVp6Opk 356nW7I1u3w7T5G+ddgIc2yHCJA4PxxbEIECk65MBopBDxYTVxrYoHpuazoz 9zrz5lV67g16bSIBZWZzOrEiAQAQ3ZqOprZhJgtutOCJzfj6BPz9cByQE8rM xToKGHxBBP1q0IgL9Lk3RlZ9hxRbm+pbonxtCrYxjdibR8e0i6m9huBa/dp4 7r0bzDsh9Hu3oACgaxLR9E5ab+E9Xp3JqW/hLUBJpbg5pU+M7dL5XI8w4nTx 3T54SeiiI9uZm61MaDMb1EAfKsX3F2M78uCOyNJ4ZHUycrSMPFtDBTUw0Z1C jsqYrzal9UtB9eyFWigqXEKPLkMpQvYFN/GXarlTlczufGJVCrI+HUloI9U8 x0p8D8OFNDHnaugjpRQQ5ZJYZEs60cQ6afsQ6xgCvi5CLdlKU2CNbkcWvT2T Pl+h6+A9Rs893DJ0vordno3tysNvtujqGHur1qEWDbyR00Jx1fMVmKkIMYOb 4ls8s9rAb1OkpJ///Odf+cpXJqSPj29yOzWeklAUlQFq/NBBqCqTk5PlxX/8 BjHcvHnzmNZttBmNxtdff330JRA+CI/Ro8Hl3ZGmSEmypaamfuc73/GHCSG8 8cYbHMdN6Nk/dH9alAQGotEvHmQDgbFv375JLvlbUNKDkakQixcvlptmv8J5 6aWX0tPT/V9f4enkU37NLE8M969OD0Qve/B/B5bHWIImj42NffbZZ/2B//jH P37YshLTKgZms/mtt97yaw9IzDVr1ng8Hnkt5fELRk3X/2ibXDPLT7po0aLx pybUzPAiTKjl/AZleMIJZQkJCb/61a9kP9HR0ZNEeHL7NMc/yTne09MjHwSW 8Yfg74t/2L7D06q15AXopih9p5WMgHsA2lCWRlPPrJFNbyffFudxle26urr/ /M//9G+KB8C4fPlyiO14nwCAv/3tb/0xBBjcsGEDxL+8vHzWyGr5D7vFI9sd PyWVlZVBUfEvXgFvHKS5fwETv8FbAGjsTzFI6t/85jclJSVwSh50V19fP+PI jLdPAiVBNoG8mfqahykpKQ8bYxMeHv6wq8brq+NFOnAH87UrYqkF4fi8W/jc m/i7ACZh2KWi3uLOlnplc1lfR3QHHdbOX6hhN6TiS2N9IPNOKPFaIP32VfpQ FhHfgmZ3oxkdeGwdHV3DZLaSbRiq4dWtmGZ/NroiDl2fjF+toetIvoH2KdgL tfiRMmxPAbY4GtgBAzICuABEAt0OxPRemC8Cb4eQb1yl3gym5t4kA1LxzZn4 2hQfLr17i1gcTV2ulMKbTGGNhlPl7KYsoAb1pUpFcmtfVqcis1OT1K1P7DYk dulT2riUVjq+mT5bSRwtw09XkuHt2kLUlKcxnavUrknybc8Kt3vzGvlaENyL BLUPDLgxEz1TSSf0CCn9+sRen/CO7NQGN7GbM5E1qepDRSTI6VraVk0ai3r7 87vaU1r+v/bOKziOK7335Re7/OCyn+yXW+uXLVf5xX645XK56trXknYlLiUx ZzGKJJhJgCApMJMAMwki58hBzjnHQZ48nacn9eSASYikqF2K9xv29RSWFAmQ kigQ+n51aqr7zOme851zuuf79wktjW+eAMWxs1i755F2nyTc3XOyhrrXxydL 9Zdb+e057OfJzO/vsx/fZv8rXvfxTd2efC6uSZfQqbvbaygcdUgmXMl91qMV zLZ8bVSptkFBjbD0CMc2qQ25Q9YsqS1nxJo7ZskZFVKlpkst3DdNTHyHLrxE ucJSKLOk9tmTuu3J3dasPkORlH00xDwa1hWP2YvGnEn9wqlaClTSgXLtNw2G y03mC/XmI2X85hxyQxY49uSWbGpXIX26Rt+lmxy3BiqV7iOl+tVJNGg6kBJ7 JJr9ZZqjVWRsPR1dy2zP4X5/l/vdHX5Tmv5qs5AhtT7oFnbm6taksDtz2Iwe PWcXIKR363dksasfMpuzmPMt7PUuNqaW3ZDGfnwrHD66HQ6/v89syCRBwR0o 10AOv2nkLjTxF5r0FxuNFxuMsdX83iJ2Rx4F2VuXToFS3phGFfSqTWZ50Ktw uogRk7PP4G3n3FUau0RhLZZbskZNGcOm3FFDjZLtp7W9lLZghIzvoL9pokHl fS0Jv4h2ZyG1LZ/Ykhd+QdjxGuJ0A3GuiYTyjG83xreZLrcYTlaxxyqZi806 KG0QsI/ktrO1+sMl/NeF3Fd54QxvzibjapjMbi6/X5fSyZ0o10VJuH1FLORz fQYJLSpzWBD7Z7OGrTfahfhW4XaHAKWU1GtJH7SWyB3VGqdEYb/YwoLmja4j r7VzqYPmzCFz/jBfPMIWDzN5Uu5el/FWpwnCEq/id7sbRFiKSqIoCtKAH7Lo 2d4/wWAQHCpwzF59Q8frAHeur69PKpWq1eqfcIkwl8s1OjoKzsnPpCWfvxh6 B3UhZt5oNC76l/EzqSQRcI95nm9ra9NoND+HyRzHdXZ2/uBbXV5l6c1gdnYW qgnyvMRJ32+b/icHHOa//du//cu//MucnBzw0xQKhfwF4AiBcxhZH/t1smJy clL0AMHN+JE5wRp/P/zIGn8rfr671tsWI9zcoAEMDg7KZLIljqD7Cdv2zMzM 0NAQlPyiN1VQGT09PWDXEjs1IrzhfyeikuAr2AWNBg1bpVK9ueigyiAbcMm8 w7/YW/0JLgeV9LaA3he9oMgiwKdPnxY3oKJfd9RC90mM+SKJhfB5EvvpvbAk +eQ299EN7r+u6VbdZFJr+0hltY0rZ+jmcrmuYEK402XcU8iuTaNWPaR/f5f5 7wTd6ru69A6GMlGCgxwk6K/SdJ/e4k8VEEOKIZ+tm9f3XmuUR5WqD5QScc10 klSXOsQ/GNDd7uXAcwZXf0s2uTaN3JAZ7rbYURjuhfkimVqVCBJJFBSQGd0X iVxMHTjb9LFKek0SF+70ucWtT2U2ZzBbs+itOSQoi32PNEUDMlY/IggjY9R4 TI12Xwm5u5jankNtyaS3vXh76b5SzcGK8ItlL7dTF1vp/SX02jQa3O/PHtBg OPwc6LIdRZq9ZeqDVeroeu2lNhJSftNCxjRqTzZojtVqwKXf/Uh9qoZKlRpL FNYalYHlB+2mTpbtLO4f+lqi/qpIe7Bce6JWG91AxDaSl9voK+30ySrmi4e6 /77B/5/r/H9e1/1nPGxz69MgS+Exb1+XaPcUk7uLqL2PyK/L1AcqVCdqNdkj hjKluWjCHN/Og0LcL2Eyh2xSk19hDw4YJzNHTIkD+owRY6nCWgue/IT1sES/ JVO3IY3bkkV/lUscLKYuN+pJ56w99GTIGAA/PNxPV6xel0Z/kcx8/pBdncT8 IYmC8OkD5nd32C8f6o49MivtM5bQ4y4ueKLUBPGfJ9N7SjRRlWoI+0rUOwo0 UEfhEYPheVX6vdnmCvmk0hbqZgP7csyf3tavSzTcbrFPCH6Z4L/bZtuQov/d HW5dCohB5loHAwprYxoLsjcc7obDqkRmSy55uEpzok6b0M1ljZiyR4W73abt udzaNGZrLh1VThyv1R6uJMAoaAar77BZTaM80xmwtDstQ3JLeKG5fuNkK+Oq Jx2gaKBYkqWG3BGujyaczgm7fbxiXHGmXgX1vqtY+0UK1DUDnyCEw31kBZp9 UNqV6v0Vavh2W25YLW7OokAHbcggj5ZxuSPWZtpVKnfuzjH84R7/yS2Qh5Dn cINZk0pvygiH9dB+HrCf3Wf/kMh8kQzx4R7Po9VETCMBzWx/Cbkpk96ezdzr tLRSk728v1bjutLGHa4MD3Q800SeaybDzwpqCWgz0BL2PtLuKiR2FhBf5RPb 8kgx/OCV+6pKere7QYSlqKS4uDhI8+ahy8gy5G1V0l//9V//4Hgb5Bfk0KFD cPW9br3KyJoG4pPkVxFnIoDL/YPzO5BlyI+s8V8PK6Ztv/l9Sb84//Ef/7GU laAiLAeVVFlZ+bqnx294vPCqf/XpPQbCZ/eYLxLpNQ/pLx4wn94GlcSvusmm 1PRrlFUCW0bSTeWysEq612OKkui2ZrPr07kvk3Sr7vHrk/i8Hk5vZRxuephk NiXr/iuBP5xLdI9JHaYumu290SI/Xqk+Ukl800Tf6dPd6w9LpJvduuud3DeN LKikdWGVRG7NBcc1PO0IVNIfEl90JN1m/28C9/FN3Zpk7nwTF9/Jnq5hwesG ifHRrfAorLWp1Pp0cmMWsfWFSiqRyo3mMZd9TMFOxNRo9pYQu4qJ7dnklgxy WxZsq/eUqPeXq0/WaS+0knEt1IESekM6/UUSBAbE15qHuvWp7O5Hmn3lqoNV KtBEsaB0GgiQPEeqtVFVYYW1RwIqSXOqhk6RGiQKS7VKT3NSq6GLpruL+4cP lIT97agKLbj90Y3a2CYCJNLVDvpUNfNlEvfRzbA++s8EKB/df9/k1qRROwq1 cLadxZoteZrNYH4h+O3w08oTdeqUIUP+hDlrxHypRberiNz7iE4ftEqNfqU9 OGj05Y0JaUPG7PC7bm3lSnv+qPVgMb85HSQJC347SM6v8qhztXqtfdYWDKuk +A4d6C/I/KYMcn0KtS6ZXpfCrE+DxOzaZJBI3OY0PrpMUNqmzcH5bi4QWyms TeE2pbNHKrWn6jUn6zT7SzRbcjQbsrRQNR/d0n12V/91nrlaMam2h3q5/6+S 1iYa4hvtA/rAoD5wq9W+JU2/6r5uYxp3toG91MqAVNyQzvzuLgPi+suH9Nqk cN/QznzyUKX2eC1xo1uXM2bKHRfu95q25bAgQ0D8gj46Wa8F0bE1m/oogVt1 i02tHyW0HXZDm9EwOGxw9Bt83fxkC+1qIJ21WkfWqDllyJg/phukNZOuEbdj uHJs4nSdcrdEDZpI7J38MoXenhcWpPskxKEqzZFaDdTsbol2ay6xKQvaUrhz E8Lxcq5wTOhgHZVK265sw2d3dB/f5FbdY9al0OtT6DXJNEjI1Un050nhSwYC FCm0JdBEW3Oog+XEkRpoMNqvCokvU8iN6fTtdksL7e3VT9ZqHFfamMMVqqNV 6rhm8lIbdaGVjqmnDlWEJ5rtLCI252g3Zms3ZmnXZRHrMsNhiVfxu90NIiyq kr7//vvf/OY3v/3tbxc9FbLceFuV9C//8i9bt279+fKDvAPiYvivm4AWef3o D44RMplM4uC0X9xpQZbOj6nxXw8rqW0vZ5VEkiRkLC0tbemHLAeV9PDhQ3CB Dh06tNAjOnz4MHy+OuExwqv+1e/usBDWJtI3axQZLRPJTbKYR+ovw74f9U3Z WF5nT+lAV/HAYOogkyTVp0pNGYP2rEFnRr/rYbfzZqv9brutXmlVmy2EIHQT puOlup153NFH9J1aRWbzREaL/EEXey88pYW/2s5900ydbaJO19MgNE5U0YfK KPBL16WT6zPILblhiQRaCbYhZk14vhL9WbhHg91dwBeO22q0jtQBy94CHiJX JzHb8imQD3seEftKiK/LtMeqtJIxWmvmDA5eadLnjwipg0JGvz6vT13YI8vq lJ2qVu4vVx2q1lxpZ7LHTNlj5vh2/clq7lgFd6bWkNTlyBv0ZvQ5zzTRx+rA c1aDF70jj96ey2zJYjZmgA9MbQIpl6fdlq8B1VM4YW1j3c2kvbhXldc+ntYy Ht+gOFVDHK8mj1QR4PmDsDrTSGWMmEDs3Okx7H/EbsqiQBmtTqY/fcCACWvS KVBGWws02wo0O4q0XxVrdj0Kzx4CoRRVGc7krR7udo/uZpf+epshvt2Y2GfO HLLkjljzx6xlSlu5ygbb8S3CuWpzXI1wrUlIaBGuNJmPl/M78qhteeRBCZvU 48wedCf32eM7+AutzJU2OrVTk92lzu3W5vXp86TOPKkrr9+a12so6DOUDJvb GHcX760n3FlS54Mue3KPpXycrlUQklHt5XpiQxr5ZTK1MYMBFbOngIMfSh4w SxTWrGFhdx4oWXZ9KneiTP+gR3jYJ1xrNp+pNp2qMB4v1x+QsPslzO5CemMm tSaV3JapvV0ny2gdT2mZiKtQ7sgltueRp2vZO736hwMGKKvYWl10lQ5U1fEa 4lhNeNrU5mzy4zvMp3eZ6GJFeutQYY80r3c8qdfwoNecNig8UtgqNfYKjb1U aQHpWqPWy9gJn9DiNjVVjfSfaZDvK1PvfKQBUQMlvzNXd7fTUhQei2jLGTND M0gbNt7q1p9v1MXWcadruega9lQNe6eL6yY5yqIbYrkTpczmDGpLBnW2nMju 0mZ3as+UkZsywiJ9dw6R3CTPbZdltKvvtvE32sxXm01RZeGOS5DA4ZloydT6 NOpUJX+vx5QuNWUP8ZJRdcWYrGxMkTeqyx41Qh7yx4ViGVSr5Uqrbn8Zub1Q uyWPWJ0U7lGFsMSr+N3uBhEWVUninOuEhIRFT4UsN0AlXbp0qfkFCxfHex3i pOB//ud/Pnr06Nu+wRb5mbh7967oQR08eFChUMzOzj579szpdDY2Nq5atUr8 CtK8emBGRoY4x/xf//VfxVW5kA+Cd67xXw8rrG0vT5UE7e306dN///d//0// 9E9Pn/7wq0kifP/9962treJ/zb/927/94ipJfI9kBNB61dXV4vbrlgR5/kP+ 1cd3WAibU6nW0SGC7pdpBzLbx7dnazdmEPsKVcfLZTHVsjN1iuudTEI3mz1q GtD7NbZpwj4zbgr1cIFu1j/Ae0eMnkGDq4m0X2vTnWtkT1QwOzPpjYnU7gwm bUAoVdiL5dbbPXpQEAfKtXsl4bnzOwrCC0SszyTXZZDwuSlbuznnRcgNh005 WogEWbE5izlRYQj/qGOqQeM5Walf86Lv6UCZ9miN9njt/w+xjUTRhGnM5Cbs k2rr5EB41YXJEd5CG5RGw5CKHr7WIj8c7iHS3h/QN9HOFsaVNy6AdrvZxYOf P2oKGSefqG0zd/r42GbyUJV2Wy71+zvsJ7e4j2/oPornPk7QfXqH3ZAZXt0i upqrVIeHe3Ww3thyfk8WtTOLPFJCxTVTcS0k+PZfl2oPlBFnGpgqjb2VcRVO WM420AfLtLuKNVtyCcj/mlRqbTq5LpPYkBU2dk+pel+5am+Zak+JerdEva9U c7KOON1IXGilH/QbHsltxTLb7W7j6To2upYBX7qWcIJGyxtxRBUZv0jkt2ca sqWOapW7TOG80W7c/4gCsbkth9yeze/I0h+W6OPbDclSQ86ITsYoOMO4waw0 2M0G74xhctbk8pgdZr3VoDIaJQpT7ripVGUFu6TGwKjJy9s5wUGo9NrUDnJN Mog79qt85ng1fbqBPttIQ/Yut9NxzfTOfErsgtmeR+8rZg6WMBea9Il9pof9 5vg249Ysdk0Ksy6N3phDbM7TRklUXRODJNU7ru5LbpxYk0T/IZE98Ii90BJe 5OFmty6x3/Cw33ijS3eqnjhUrQaNsymH+PQ+A2FTBrG3QH2gWL2/WLOvmNpX TMfUcGnS8DILZSpLHelspFydjFnDDfv0pS5OUi1tj2uYOFgR7kcTRfcxibFR 66dds6RzZlQISM3+Lr23Um3PHDYnDRjDMq1PD3qteJwjTcSkR82ZNRcbtHuK tYdKtDl9atqgovSqtA7tzhxqYzp1UqKRaYcM+kGtTtZJmuq0LsmE43AJ92Iu VbiWQSWtSaG+yqeiysiYGuJ2m0ZKyszCmM4s72ZMcL200G6pyUe4plX2UO6Y cKqOhDawvUDz6V3245schCVexe92N4iwqEo6cODAX/zFX1it1kVPhSw3fvvb 3/6v/yEmJmYph1AUFRsb++///u99fX0/d/aQpfDdd99t3br1pdnWke1/+Id/ AI/x1aPAafnHf/zHv/qrv0pISPil5tcg78a71fivh5XXtpenSoqOjl69ejV4 EUuZKvjtt9/+5je/ifzdXLly5T3k8A2IfpE4+wC2xciTJ09GvXiz5OuOenVq w+/vsRB2ZFCj6mGnMGAyDpRLxw5J1HsKtfsl2qhyzdFK7alabUIHeaubLJww aOw+a2DGEZrVT85oHVNqW2hc8A0Z3X28s5myJQ7wN3u4c/Xc1nR21W12Swqf P2xvIt31pCNlwHCyRnu4Qn2gRL23WLO7MLz4846ccPgqN7wNMRB2FISXidtW oAUNBa7+nkLmfJ1BYfUbfVNdrOdsDQ9SBbTG0SptdL329IuBbWeaQFCQErl5 xOSB7GnsfvA5lfYQ43B4nIpp16AgSO90Kk/WaWIayLQhfS/v6Nc7K1XmzGEe QqVK0HlD00+eWEOzGSOGKx10dB21K5/99A7/yU3+4wT9R9f1H8frV93R7cgh d+UT5+t0jaQHzt+r85+QGDansZvSGFBJ1zvI+C4ipi481Wh/CXGmnm0gnT28 t1xlu9ROH6/THKgIv+xpXRqxNpUEh39bthYM31OgPlqlPF6jOFqtPFKuOlSm PlKhPl2vOduoudxGpkr5Go2tWm2/02M4Vk0fqqDON+maKHevfrJozHWowLj6 rm5rqj5/2NVIeOu07gc9pmPl1J4i7c58YkMatzaF21/E3+02FkyYKxQGnaBw OcZ8HqU/KATmHwfmnwRnfKGQddJv4u2G4nFdxhBXpjCNmF2026/zeL0+Phig DFaysJfamkqtS6KiJPT5ZuZqJ32lnbrQpo1rUZ9p1ByUaPYWaXYXaHcXELsK iK+LyettfP54eH2/h32mzVnc6iRmXTq1E76SaM5Uq5Rkn9vcped689snNiSz nyfqoh7Rl5rJm13EvR4qZ5TPH9MnD3JxzcSpes3hKs2uQmpdCrs2mduUyWzL o3bkh8NXBeFwopJ9OGgsVgglKmsL7ezk3AM6geRHvXyZgy2pHe683CIHaXyw nNiYzqxNYc9WmaU6v3tqzhmcZdxBwhmQWSc7OFep0lrw4rVKKUNhRVmp0Omt mhmf0mJTgbQ5Uak5W6epHFc7nCq7Q1U0oD1YTO/MpS5UaUyGwWlnn8M2roQr R+9qIRwxlfzuPHZrFrMtk96aSW/LoncXkF9LiBOV2tttajU3Me0ZnXTKxo3m Pp2j3wAtNmgPzpl9M5Uqy4UWzZFKxX6Jeu1D7vN7PIQfvHJfp5Le9m4Q4c0q CW59f/d3f/fJJ58seh4EQX4+jEZjSkrK559//r9fsH79epC95eXlb3jG29bW xrLs+8wk8hPyDjX+62GFtW2/319TU1NbW4uV+1Nx7ty5iMskvo0dpKi4e+jQ odcd9ap/Jc5POVDAjJIKu01mtky0yuTxDcpLtaprTdqEdupmJ3W3m8geUOZJ FQ1KkrELDr/DFfQIPj/nmWbcU1qbS2E2jRv0/ayhUGbKHjM96NGfLaOP5BJn JHSdytmn9/XwnhK5+U4nkdCmudoYfv/mxRrFxWpFXOVEXMVEXJUMti/WKr+p VoIjvbMIAnGghD5Wzp2t0d/rNBAWm8Vr72OEuDpmcw6xPZ84WUucaSTiWsJL McR3sXd62ToVrzQaWKuZsVm1roDWNaVzObxu9bR7yGodSetXX2wlwLcvGmPH 9HqZQd9BsBVyGkI7qTO5LNPTTofPEZZOI8a7PYYzNYZd2YYd6YYtqYZNSYbN yYZdmfy5KupiHZnSAzrLS7qnR4XA7SbD2VL2TCl1o0mbNajMGlRcblYdrtBE lRPnGpkaraONdZcqbfFdzNlm4niNdlch+WUy/WUytS9Xc7pUeb4SykF+q23i dsf4rXZZQqs8vkWZ0Kq836V82KNM7dcUjOqq1NZKtS110Hyljb/cwt/rNXZw 3kGjr1rpvl5vPFaoiynhU7sd+YOuPKnzYZcxvpm63EBcqKOjK0wny8xXGs1F Y0IDIbRRJq1RwwsKk01j81pd03Pu6XlHYNLitRhdRsLCNyg1FROqVg2pEgx6 l8XoFjy+8FuATXauboQ4+0geXSS/1kA97NenDRnShrjkPlVy70Rit+x6k/py veZinRYK52I9dbWJzpbq67Vm+NGCUdPeQnZTJr07nzxTpb7SoExsl1PckMcK enyoclB1pFi/v9BwqR5KVZE7KC8cUtTItXUKonScSOkH0UTc6CDP17PHSoxH JcaYSsO5OkNcveGbesO5Wv5sLX+1RZ8xHF6KEGRON++RmnwTZidj0jgMrQLf 0iwbvt8LLYSJa2aPSPioYv3tFmHC6PaG/K6gz+BycQ47YbNLDfYm2lFN2EvV VlBJiYP6cgXHW7UzfpXDqS4YIu+0kw+6yDo5wZi1UIqSYSauVhdbqXvQSlmM g9OOXrdthDKzMqO5nzU/6DRfqjd/U2M6V2U4V8mdq+JAVl9s5m60czlSitbL QSX5XDKNYBw3WWWCg3V5XMGgze/vIHVpfRM3WkevNSpOSYzHiy0QlngVv9vd IMKbVVJdXR18K5FIFj0PgiAIgiDIL05iYmLELzp16tT58+cjMxGOHj36uqNe fRx9tjE8eiqhg5MJXrMv6PBPWl06s3kYgsVBWj1mm1ewuVibqdNmaHaa2nzW rqCtN+QcCQUMocePQ/OP/T7Ga+12C212q1Tv8nGeWZ3DrTMpeX7QYBx3BiY9 0/Oe6Vmn3211cxYnJVgmTIZ+o77byHca2SYj02DkWk36HpNxUM1IrzbLD1Wo jlYTD/oMdVpXN+ce5c0uoT9g7ZQRffEtst3hRRg051uYq50gjvjcMXMD6Wij rWqdzG5omTQ1eSz9Vr/fEgQJ4A1OUjPecY9T0U7yFUpzjVI/Rsudpnav0Oq2 9Dmtow7rqMsqDViaQ0Jt0Nbm8jnswWmDb25UCNZo3WVKZ+6oLXlAAJHyaNzI CDrBwVndgiMUcs08cYZmzHbOZNGazHLBKLXo23muvWJk6FKrMqaRuNBO50+Y S5TWIrkla8yUPmK602s8XKL/5Kbus1vsg7qJcVWvQddp5LvMxt4Xod8sjJiF MSEchgRByplGB2gtqMtHcksD6ezlJ6VG34gZfPeQ2hkiHZM6u8lo08lZXUye ccs9055kY2arTqVjwPU2OY2ce4r1zBm8IXvA4wnahUlbJ22qURvrNOYe3qWy h9SOqUGjt42xt1KmPoY28m02fb3T2OQ1t/ktnT7rgGPSbg0ELJMek1mpJ+p4 oo7RK0m7m3T4QYoaTSMWfZtg6BQsMsGqNtu0gp2Cn7Y4dA6vyR202nw2hSDE dzCxoK16NMPEqNkktQrDQa9y2q/xTZKEYGwlw/0vE4zSoGu38o02faPD2AzB LvTZXbQNftltpWyuUVNgxBTS2Kd1njn95By0Mbl1asIS6uV9ZWpbzrhZorJS 7mlraN4xNW/yBWmni7A7R83uHt7dofO0s55m2tNMeYZ4q8VBT02qQl5lwD0e cI363ArXpMkaCOl9sxPWwMMBPYjuwjGaFYgZv2bKRzi9vM1thKIe5AylMv7R hL5Oa+tgvAO8R6YXvLb+KXtn0NHtsw96HUMep8wOV1Bw3hKYs/j8FrfV7Law TrfaHtDYA4zT6XFrp71jU15ZwEf5/RyEYICbCjKhAOmxDVn1LWZdC1wIvCvA ex9D+MEr91WV9G53gwhvVkmbNm36m7/5m0XfpoQgCIIgCLIcaG5uBhfoyJEj 4AXBRnR0dEZGRkxMDGzHxsa+7qgfGKvTTkO426+TW4NG/5wzGJr06afcQ9Nu 6WyAmJ+xzM1YZ4PclK01JNSFLPUhEBSW5ilH70yAm3v67dy3T2b8ZMjWErTU ++29zkDQMfXEHfAGPPIpZ++UZ2R61j/95Mn048czc57ZKd1siAbZMuXoCdk6 QrbWKaE2ZK6esjSAqznt7BNMAzfbZEfD67wRaUOmbp13wjKptZgnLR0hS6OS 6EhoHdtbpjpQqb7YxoA3e79fXyy3dOk8A3onox/zGGqDpkqftdMR8NunHruD vqCfnpmUed3KPtbUSFhbCKOKG/eZG6aEOvjFKZcULA1nRqgOGh6BUJqZcc5+ +8Q//5TxznTznlbWVaGx508IRRPmBq3RMwn556anLf6Zqcm5p77Z2akgBw7/ 9KRs2jkwZWn2GFsax6VX25WxTQSUKig4yF6hTIAz5I0LD/vNxyTGjxL4z25y mY3jPNMFpTplb5tydE05uqecfdNQ7J7RaTcE6bRr0GUfHuO0Erm5SG5tYz0y a0DrmAoH1xThmuI9k76AeTakM1p1RzIMq64aNt7SZ7dwRgszE2Knp0z+uTnf /NPg/Nz0nHduzgFqqZmySBSCRGFpZd0ya1BuDXXx3lrCXqMxd5GU09AQNFUE zVUvKroBZKlt0i4Epiw+n8uhCBgrIDgdCmFy0uyfsnrtPsdQyNoEJkx7x6f9 qpmAdjZIQfnMTvFzM6a5OZt/ysY4hLu9zIVWKntQQ+lHwCiwcdavgZQgEHin MGzyDhldnEnuNbeEzDUhcy3UAoRpR9dMkJmdNk/NOEBoc95p1jtjCz0GiwLz 33lnvwUly0/OghWVGlv2mEmitOomZz2z30K9mIPzUD4aR0hmC44KgWGzf9js GzKFg9pid7m0M96JGc9YuLRdg1Oe8UDA6JuZhnardU4lDuivdTL5o6JK0s4E iNlp49y04PELI3pTzqghc8QIinVE8KvtPsZu9dnDKulFDfaIDX7+ceDxd08f f/fd/JPQ3KxtZsbmmfIJgRmzf9rq8/gnSVBJ096J2SA5G2JmQ+xskIYLbcav nnb1hy8ECI7+0Nz01JM/QljiVfxud4MIb1ZJcM6vv/560ZMgCIIgCIIsB54+ fSpORgAfBpyc3NzcU6dOiQ+Ts7OzX3cUqiRUSaiSVp5Kere7QYQ3qySGYZbV fFIEQRAEQZA3IJfL7969e/Xq1YXL/544cQI+BwcHX3fUq/4VgiAfFq9exe92 N4iwlLfKIgiCIAiCfBCAXyQ+Op6cnHS8oLGxUVzh6ttvv33dUaiSEORD5wdV 0jvcDSKgSkIQBEEQZMUAzo84xubEiROZmZk3btwQJyYIgvCGo171r35tugnt /TWwMqx+nRWvxr/b3SACqiQEQRAEQVYSbrf73r17kQE2586dk8vlbz4EVRLa +2tgZVi9dJX0/J3uBhFQJSEIgiAIsvKYmpqyWq1er/fZs2eLJkaVhPb+GlgZ Vr+VShJ5q7tBBFRJCIIgCIL8yolCEOTD56e9LaBKQhAEQRAEQRAEWQiqJARB EARBEARBkIWgSkIQBEEQBEEQBFkIqiQEQRAEQRAEQZCF1NfXf/bZZ0+fPv2l M4IgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIgCIIg CIIgCIIgCIIgCIIgCIIgCIIgCIIgK4pgMJient7f3/9LZ+RH8ezZs4mJifLy 8oyMjNLSUqlU+sc//vGlNAzDlJWVZWZmNjQ06HS6V08yNzfX1taWm5tbVFQE Z5idnX0vef+xQD4LCgoqKytfil+KOYuWybLC7XZ3dnaCRWBvd3f3kydPXkqw /KtYrVZ3vobh4eGFKRc1VqVSiQ2+oqICTvsejUAQBEEQBFnhPH369MaNG1FR UcXFxb90Xt6d+fn5+Pj4qD/n0qVLgUAgkqa1tfWlBOAtLzwJOMyxsbELE1y8 eNHn8713a94aEAWQ23Pnzi2MXIo5i5bJsmJ8fPzYsWMLcwsmg8aPJPggqjg5 OTnqNcTFxUWSvdnYZ8+egTh66XCIgfj3aQuCIAiCIMiKBPzDBw8eiC7WB62S 8vPzwYTo6Oi6ujqe51taWs6cOQMxiYmJYgKHw3HkyJFDhw51dXXZbDZIIFpN UVTkJHl5eRBz8+ZNg8HAcdy1a9dg99atW7+QTUtleHg44kUvjF/UnKWUyfLB 7XZDViF7BQUFUMXd3d1nz56F3fT0dDHBh1LFUF9Vf05JSYmYVWjGSzR2YGAA diFNfX09JIBPsB1i+vr63qctCIIgCIIgKw9wqI4fPx55EP3hqqTvvvvu6NGj YEJPT08kcmxsTLRrbm4Odu/fvw/bFRUVkQTgnUJMamqquKvT6WD38OHD09PT YgxsiJ6nIAjvz5i3xOv1RipxoUpaijmLlsmyAton5A1EfaS7RKVSQQxUPTSA 5x9yFVdWVkIerl+//u2334oxixqblpYGu5mZmZGT5OTkQExycvL7zz+CIAiC IMhKQnxYfefOHfFR9oerktxu95UrV2JiYiLeLwAOp6gdrFbr06dPRWMtFksk gc1mE5/GP378GHbr6+thF5zPhWcWBzXV1NS8N1veij/96U+3b98W+yBeUkmL mrOUMlk+QG5FOTM5ORmJBAWh0WgMBsP333//4VYxz/OidrPb7WLMosbCLki/ l67Z2tpaiLl37957zj+CIAiCIMgKA/SRTCYD76ujo+ODVkk/CE3ToocMPqcg COK26GFGEH1Rs9n8/H/cztbW1oUJ2traIDIlJeW9Zn3JNDU1iaMKRQMXqqRF zVlKmSwfPB4P5OrChQvfffcdQRCgCEDXKJXKSFfLB1rFoOXBqJdk2qLGPn/R EQxpTp06ZTQan78wXxxiupynlSEIgiAIgnxYtLe3rzCVND8/f+XKlUjHAcMw sB0bG/tSMnEivzhvBTQjbA8ODi5MALvLdmoSuMcgCsBPDgQCFovlJZW0qDlL KZPlgzhY7saNG+Kwugj37t2bmZl5/sFWcU9PD/z6iRMnFnbeLWos8OTJk6Ki IjEe2oC4AekjCRAEQRAEQZAfyQpTSeBAgj8JFp08eVJcE4wgCNg9f/78SylF JaVSqWBbXOhvbGxsYQK5XA6R165de2+ZXyJg48WLFyFvExMTsPuqSlrUnKWU yfJhfHw8IhYKCgpkMllDQ4M4GS0vL+/5B1vFV69ehV8vKytbGLmoscDU1JQ4 NWkhVVVV4qwlBEEQBEEQ5MezklQSeI+iMwxeJcMwYqQ4+u7ChQsvJQanGuLB wYZtcYIPOKgLE4hLQMAJ30/ml444uz8nJ0fcfVUlLWrOUspk+RBZx6+2tjYS KY46AwKBwIdYxUajUcy/0+lcGL+osbCblJQU9WIZc2jks7Oz8Hnp0iWIefjw 4fs2A0EQBEEQZIWyYlSSx+OJi4sTxyDxPB+JdzgcYuRL6cWhSmLKrKws2O7t 7V2YQBwQFVlOfJkgji6LerGm39gLxAlKJ0+ehG2SJJ8vwZyllMnygaIo0eSF I8qmp6fFFRvg2w+xisUhc/fv338pflFj5+fnDx8+DLtgdSQBNH5xEtbCNUwQ BEEQBEGQd2ZlqCS73S5OQgGh9NLD+dnZWdHtfPr0aSQysgieuJJYeXn5S0/v n//PumEFBQXvx4QlIk6leR3Xr19/vgRzllImywdRBAFPnjxZGC92oKhUqg+u iiGf4hLuSqXypa8WNVaj0cDGmTNnXjpQfLHyS51lCIIgCIIgyLuxAlQSOJbR 0dHi0KmpqalXE4gCSi6XR2JkMhnExMTEiLui9Lh8+fLCo0S/tL+//2fN/Nti Mpkq/hyxl+To0aOw3dXV9Xxp5ixaJssH0AvixJyFy0pApCg0QqHQ8w+tisVB kocOHXp1JtGixhoMBtg4cuTIQkn47Nmzc+fOQTxoqPdkA4IgCIIgyIrmQ1dJ 33//vbh82dWrV2dnZ//054hpWltbIUFCQsL8/PzzF4vgiQ/eW1paxATghYqj syKz+0dHR2H39OnTcM5fxK6l8+q8pKWYs2iZLCvEnqDz58+L49Cg0svKyqIW zEX6sKpYXKIBNNoPfvtmY8GQEydOwG5RUVFk5XNxSXOQTjjiDkEQBEEQ5Cfh Q1dJKpXqDSPQxBfKgBsseshxcXF5eXni9KVr165FZBTQ2NgoPt5PeYE4DUQm k/1yli2VV1XS8yWYs5QyWT4EAgGQM5DD6Ojo7Ozsmzdvit1nkSlUH1YVi6+4 zcrK+sFvFzV2aGhIbN6gswoLC8WFKYDOzs73aASCIAiCIMhKBjwr8K8kEskv nZF3pKWl5Q0qyWQyick8Hk9CQkIkPj4+fuHk9+cvnti3trYeO3ZMTABOaUND wy9h0Ftjs9miXlkHeynmLFomy4pgMJiYmChKGzG3Wq12YYIPqIpzc3MhA01N Ta9LsKixCoVCHGInEhsbOzo6+vNnHEEQBEEQBFmBzM7OWq1Wv98fGar0Es+e PXO5XOBdvzR3/gNlKeYsWibLiqdPn4IqfMO4spVUxYsaGwqFwFjxvWAIgiAI giAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAI giAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAIgiAI giAIgiAIgiAIgiAIgiAv8f8Agby92Q== "], {{0, 192.}, {561., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{561., 192.}, PlotRange->{{0, 561.}, {0, 192.}}]], "Output", CellChangeTimes->{ 3.763383247129921*^9, {3.763383327082108*^9, 3.763383364069449*^9}, 3.763383411150867*^9, {3.7633835325686483`*^9, 3.763383537987918*^9}, 3.763417363739852*^9, 3.763744000253457*^9, 3.7637446346951933`*^9, 3.763744911664227*^9, 3.7638100685550327`*^9, 3.763815816322093*^9, 3.763914842693679*^9, 3.763989532107759*^9, {3.764005727000289*^9, 3.764005753814575*^9}, 3.764081111243669*^9, 3.764589445488948*^9, 3.797848690118957*^9, 3.797922485352477*^9, 3.798178777791087*^9, { 3.7981860837266693`*^9, 3.798186094757482*^9}, 3.798186202495075*^9}, CellLabel->"Out[24]="] }, Open ]], Cell[TextData[{ "For each subword token, these three embeddings are combined by summing \ elements with ", Cell[BoxData[ ButtonBox["ThreadingLayer", BaseStyle->"Link", ButtonData->"ref/ThreadingLayer"]], "InlineFormula"], ":" }], "Text", CellChangeTimes->{{3.7637376968398438`*^9, 3.76373771634305*^9}, { 3.764589512230421*^9, 3.7645895175421877`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetExtract", "[", RowBox[{"net", ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{{3.763383493635799*^9, 3.76338350654666*^9}}, CellLabel->"In[25]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzs3XlsJOd9N3g5UWwnyB+WnexrI8gCyT8ODMEIBMR5N0YMr7OLF3ayi1iy EcebjWwp1jk6LMkeSbYljkYzmhnOfd/3wbnJYfO+72tIDu/7bt5DtklPx7Es jfbr/i2ft1TdXV3dXX09/f2AILqrqut4+qmq59t1/cUTLz361O898MADr34a /x794fr//ZVXfvjGY5/Bm++++OqzT7/4o//45ouv/ejpH73y35/4fXQ8+4kH Hsh+8IEHfvf6IyIiIiIiIiIiIiIiIiIiSkrvv//+8vLy7Ozs9PS0m4iIiIiI 0hLiwNzcHKIBAkKiMwrFw/379/F1z8zMLC4u/vKXv/wVERERERGlMYQCRAME hKWlJYSFROcViiF8vwsLC3fv3sX3fu/ePS8REREREaU9RAMEBMQEhAVGQo0h 8i8vL+O7TnSNIyIiIiKi5IKYsOyT6NRCMfH+++8vLi4yDBIRERERUUAIC4gM vJZQSx6PZ2VlhaeJEhERERFRQAgLiAwIDonOLuQ8uWww0VWMiIiIiIiSl1xI mOjsQs7jlYNERERERGRNriJMdHYh58nTJRJdv4iIiIiIKHnJEygSnV3IecyD RERERERkjXlQV8yDRERERERkjXlQV8yDRERERERkjXlQV8yDRERERERkjXlQ V8yDRERERERkjXlQV8yDRERERERkjXlQV8yDRERERERkjXlQV8yDRERERERk jXlQV8yDRERERERkjXkwpS0tLS0afPjhh6oX8yAREREREVkz5UF0MeYLvE1E yiFbfvOb36z7uNHRUdU3JfJg10RDdstBm3+rv1qJ/xz+anWl7cQWm3+TzZXx n0MiIiIiooiZ8uC5c+eM+eL06dOJCDpky3/913/hO8rPz59a88EHH6i+KZEH kfKeOf2Izb9fri7Hfw5XPcsXvvG/2PxDJIz/HBIRERERRcyUB1dWVlS42Lx5 86lTpxKQc8geyYMtLS0B+6ZQHtyQ952dlT+y+Hv54t8nNg9e/R//rebf/leL v6LH/ox5kIiIiIhSjsX1g7t372YeTGba5EEkvjN33rb4e+3KNxKbB29+6/Nd z/6lxR8iIfMgEREREaUc5sHUxTwYB8yDRERERKQx5sHUxTwYB8yDRERERKQx 5sHUxTwYB8yDRERERKQx5sHUxTwYB8yDRJHJTahELz0REVHKYB5MXcyDccA8 SBQZ5kEiIqKUwDyYupgH44B5kCgyzINEREQpgXkw4Q4dOvTaa6/t378/3A8y D8YB8yBRZCSX3Y475kEiIqKwRJwHP/jgg9LS0oKCAux//ft2dHSgV3t7e7gZ J3qDg4OYsZMnT2ZmZiJquVyuxcXF+M+GfVu3bkWs27lzZ7gfjEUejPOP88yD RLpiHiQiIkoJEefB//zP/1zn89JLLy0tLZn6btq0Cb0QysLNOP7u3Llz8eLF tra2kEPeu3fv8OHD6/xgDouKiqKfkxhhHmQeJNIP8yAREVFKiD4PwoULF0x9 HcyDly9fxqgOHTpkPdjMzMzPfvYzmZ8XX3wxMzPz9OnTW7ZswWvp2N3dHf3M xALz4MsX/x6Jz+LvubN/k9g8eOkffhcJLf6u/o//xjxIZMQ8SERElBIcyYPI XAsLC8a+DubBEydO2MmDmFuZmW3bthmXyO12Z2RkYCT379+PfmZigXnQ5l8C 86DNP+ZBIoV5kIiIKCU4kgfhzJkzxr7WefCDDz6Ymppqbm7u7e0NNnX48MMP f/Ob3+zatQuj2r9/v//MqyE7OjpkNrZv3/7++++bxoMumKKpI+IhxvDrX/9a dVleXu7p6WlsbOzs7FxZWQk25xMTE+3t7Zj50dFR4zwY/aePmvrQ0BCGxyL7 z8ZHfnkQw2B4pDxMCIsfrHA+imUedBlY5z6Xn7AaY6u/WkHKM/3dXV4sqygp KMpramn0rCyp7uEui1MQCeVvZXmppbGhKN9VUVq8vLiguqu/X62uJGomiZIN 82D8DQwM3L17N9FzQUREKSb6PIjQh/8vvPDC7Oys6hssDyIc5eTkvPzyy8Ys uXnz5snJSeNgtbW1pmH8rV+/Xg2PGCgd+/v7LQKUaRIY/r333pPodPjwYSyC Gjlem+5CMzMzc+HChZ/85CfGecBgWBxTAh0eHkavH//4x/fu3UNMfumll9Tw r7zyin98kzy4b98+fBcnTpzAMGr4N954A5E52CJokAeDmZubKykpKS4uRtJH MUY5NkdgNjAzmCXMGGYv0bNDlOyYB+MMOwsp8IWFhZUV/jZFRER2RZ8HkUfe fvttvECWUX0D5kHsoTZs2CCfQkrauHHjm2++qc44NV7fV1lZaR0G4fXXX1fD S4ZCsAo2t/6qqqrwkV/84hcej0fNlTFsGs8vrampMabFV1991ZjaTJdPdnV1 SXcVHpFtjR/H0hmHlzyIOZFilMJRAyNXBjtYGes8aJ0ExS0f/09FXzMnJiaK fVpbWxP+4AzMAGZD5gczltiZIUoJzIPxhO2S6Zc6bLKS5Mc0IiJKctHnwYaG hvr6enk9NTUlfQPmQXXzz6ysLO/amZYDAwM///nP0RFpSJ0eiaTjWfPWW2+h 7969ez0fp1ISXgfMZdYkD0JGRob8x1KgKN5//30sBTKdceDx8XEM87Of/QxR bn5+/r5PX1+f5FlkvZmZGTWwyoMSUYeHhz/44INf//rXhYWF0vGnP/0pFlAN L3lQHDt2DOP/yHfyKhZZOubl5QVcBL3zIExOTspRQnw1Nk+CmptbrqsbsvmH ge2ME5PGDMiRQcxSdMtElC6YB+NmYWEhPz8fS93R0YEd4tLSkpTD0NBQomeN iIhSQPR5EMEKeUcOsSHxSV//PNje3h4stSFYSa/bgR5lKHnN4n4ySJTycQQu 1fHevXt9H4cuxk+pPChhU13uF8zg4KD/lYltbW0yhurqatVR5UEs6W9/+1vj 8HKvVKitrVUdVR40zv9HvkgoN0c9fvx4wFnS4H4yISFol5eXI4uVlpaOjIyE /Lm7vX3iqR9l2fzDwNZjw+QwUUwaM4DZwMw4tVxE2mMejA9s6ktKSrDI2Emp jniNLsPDwwmcMSIiShXR50E5+7G5uVnejo+PfxQoD165ckUOpZlymZADbS6X y79XyDyo4qTxPEzVUbl27ZrxU8bjg8ajdXbcv39/aWmpt7e3oqJCRnL9+nXV V+VBzIPpg7Ozs/7DSx7MzMz0n5Ck7IC9PopxHsR3gTZGqW0Y2HhI0cEq6vF4 ULvkXE3kaLfbbTGw5MGXn7/11msVFn8YIGQexIQwOZkuZgCz4eBCEWmPeTAO sJGvqanB8mK/ozrK4ULAfiqB80ZERKnCqTz44Ycfvvvuu+t8NwJVGceYB3fs 2CHD7wxEegW8H2nIPDg1NSUfz87OVh0HBwdfWSN9EUiNn1J50D+1BTM/P5+X l4cF+fGPf2wKm8aDnhZ5EKUkh/yOHj2qOlo8b0LurSo3vfEX0zyIEDQ3N+ex DQPjI7HIg17fcbrh4WE5UCinj46Ojq6urvoPKXnwzR+XHnin1+IPAwTLgxgt Ri4niMphQUyal+EQhSusPLhnz55/s9TY2Mg86K+1tRULi8JR2yhs+UtLS9FR 7oANeFFWVra8nLBbNBMRUZJzKg9+ZDh5Eu1nCTLGfCexzprxjjSmD1rkQYlF wT6uxhAsD5qenBjQ/fv3XS6Xms+XX345MzMTS5eTkxNWHvxo7dY3e/fuVV0s 8qA8VDEheRAtCqS8/7INA0sjJHaNsZWVld7eXjl7E9DC6ejomJqaMi5mxHkQ I8GoMEKMVsaPCWFyvE0fUWTCyoOPP/74A5aqq6uZB/1hd1NTU6N+HMN2rLa2 1nS4cGRkxHQ2qT/0bWhowN4wtrNLRERJycE8iNCE5CKHCA8ePBjw+OBrr702 EdzS0pL/hCTNYYTB5uSjtdNNf/KTnxifJ2gaQzR58MaNGzLwli1bWltbjVOR Y4U28+C9e/ek16VLl1RH5sGwIKANDQ2hcVi8BtNFIfT397vd7tu3R+znQQyM j+CD+LiKmYCRYxJMgkTRiCAPPvLIIxlBNDU1MQ+G1NbWhmVHsjOe0oB9gcvl qqurw+vFxcXe3t7p6WnTB1W5YWPIJxgSEaUbB/OgMQrJoxOMeVDdSkXun2mf ZMxt27ZZDHPu3DkZOVr1/n2jz4M/+9nP1gV52H1YeVAdQq2oqFAdkzMPJtX5 ov7Q2pmZmenu7jYGQzh/Ps9+HsTAxs9iVBghRsuzQ4miF0EefOyxx2wOzzzo b3Z2FguOnYvpXHps0LCbRiQsKyuTwmltbfX/uNvtxjZQLh7HrornlxIRpQ9n 8yBkZmaq8yoD3l8UqeqDDz4INlp/8pSKV155xbv2iAp/2HNJLnvttdf8U1iU eRBLKk8PvHHjhqlXc3Oz9AqYB+vq6owDI0vKJZbrDA/m+ChZ82Dy3E8mpMXF xeHh4Tt37tTU1ISbB/ERfBAfx0jiOc9E2mMejD9sx4w5bnJyEtGvqKhIyqSw sBBv0dFi1zAxMYFEiYHz8vL4uAoiojTheB403tjTdH+YY8eOqbMu+/v75VkM CDUdHR0IiegbcELFxcXyKQRDNavY6yFtffjhh2owdSnfiy++iNcYP3aLaOdj HqR7xHnw/v378mT5119/HSP8yJfsRkZGDh06pJY0YB6Ec+fOzc7OIv9iJ7tt 2zbpePbsWeP4kzMPRikWddWO1tuj9vMgBk7UfBJpj3kw4eTWo+Xl5Xfu3MGL 5ubmkB+ZmZlRRwmtb+ZMRETacDwPfrR2V0z/PHjv3j0kQZWVENzWr1+v3r7w wgvqEfOmCf30pz9dZ6Bu7zk6OqoGQ+ZSV/n5e+211zo7O42jjez6QZM333zz jTfesMiD/jZs2GBazKTNg/afSp8MeRC1a3FxEQ2fKO8vSkSOiCAPfulLX3o+ kOPHjzMPRmBubk5dDIg9dUFBgcVOYXl5uaGhQbb8ra2t8qyK1dXV3t7exsZG JErGQyIiXUWcBxFG5FRJ/zw4NDQk2cf/0fMffvghgphcjqe88sorZ86cmZ2d DTYtRLbt27ebghVS0uDgoGnI7u5uxKuXXnpJDYbIiVjqv4y1tbX28yBmOzs7 2/iYiZ///OfXr19HVpVLF40PN1R5sKamRgKdwFxhYP9nHcqiyXM6TJgHQ5qf n+/r60NzRe4LGtb5osyDRLETQR4M5sknnww3D44njYmJCSQp7OCwsUI683g8 Kysr8b9IGTtHFMvU1FTAvkiOclppXV2dipDYOZaXl+f6TjRFlsSLkZGROM4y ERHFScR5MErYGyLNoSW/tLR0//59Ox/BPhTDY/c6OTkZ8KH2ym9/+1u5Yeny 8rJD8/s7v/71r8fGxrBXtR6tyoP9/f0f+Za0t7cXsy3nxzoosXnwlk+i8iBa Vp2dnZWVlcYbwiAS3rxZyTxIlAwiyIMPP/zwq4GcPn063DyY/LB1RchC4Kqv r29ra8PeDfsXxEYExlh8HYuLi9hmBrx3KPaV+fn5mKWOjg4VVFdXV+WqcOzI 8Bp7k6qqKgyD5BiL2SMiogRKVB7UmPXzBx2UhnkQbRU0mRoaGowZEE0pdEQ7 B32jfx49ETlCNgVh5UEHrx9sSBqIezU1NZWVlYh+SFiFhYUSvizk5eVh4MbG xt7eXrfbbbpfqLMmJydlaz88PGzsLo+uGBgYUF3Gx8fRhTeZISLSD/Og4zTI gy4/IX7pjkseRLsFzSqJgWhWtbe3T09Pm5ZR8uALz95c/1KhxR8GYB4kiqnE 5sFEL31osvOdm5tDzurv78cGDeGxoqJCzsw0QffW1lZksfn5eQfPNV1aWkI+ zfU7jxRzJdPFlhbzg2R69+5ddCwtLcXcOjV1IiJKEsyDjmMedLwxhkZLU1OT JEG0iwYHB7F0AYeUPGjzj3mQKHaYByO2srKC3Dc8PNzW1iaPDjRuXfPz8+vq 6rAZtPOIQIu9AHKl3Eq0p6fH1EsODmJCvb29yKG5vksIjZc9jo2N4bOYt46O DrnzDBERpS7mQceZrh+MnTR53sT4+LjcK6a0tBQR2/rUqeXllcHBGZt/GNjB +SQiI+ZBp2CjNzMzgx1KU1NTSUmJcUtbVVWFNLewsBDssxbbTIS7oaEhxD3T AUfsOJAECwoK1KWC7e3tuYZzRxEGc30/G9bU1OTl5WFIpFeHlpWIiBKAedBx KM+2trY7d+4YH48YC9rnQbRSuru75bBgY2OjnR/DiShJRJAHH3zwwc8EsWnT prTNgyYej2d4eLihoQFZTG1yS0tLOzo6pqenTeFuZGSktrY2rCsQse+QL051 mZiYQJfOzk55W1dXh7fy+An8RzCsr693YsmIiCgxmAdTVyzyYPJAqwaZWsLg wMBA/G/PTkTRCCsP/uAHP7B43gS8/fbbzIMmSHnj4+NYZLkGUAXDvr4+dUY9 XiCvIcHZ3x0gb2I8ra2tqktXV5fx+CDSH8aptskVFRUlJSXOLRYREcUb82Dq 0jsPShhEM4O3LyBKRWHlQQelTx5UEM2mp6exzVRnk+bl5aEoZmdn0bempgZd Ghsb7e8R5OH1chbo8PAwPo63yIlIoENDQ7W1tejS1taGt4ODg3hdXV0dw8Uj IqIYYx5MXRrnwYGBATkyGOzpyUSU5JgH4w/BcGJiQs7nFEh2zc3N8hovbJ5o gSBZWFjocrlKS0tzfZcKut1u7FBkzMiGxkdm4PXMzEysF42IiGKHeTB1aZAH u9zzOV0jpr/LLT3bbxTi72xtm7H76q94yihRymAeTKDFxcWOjg7/R1egfGxG wqWlpZ6envr6+ttrxxl7e3vlUCB2LhiJFHVbWxuv7CYiSnXMg6lLgzx4vKnn a8fybP55YvlQZiJyFvNgYt29excJrqyszBQJkeAiGyGyIT7u8XjkLfYvxcXF VVVVzs0yERElBvNg6tImD/5jVv23b7ZZ/H3jVDHzIFFqYR5MlMnJSblmMJja 2toI9g7Y1+Czxme/FhYWlpaWOjrvRESUAMyDqUubPPjd3K4fVk5Z/P2f5yqY B4lSC/NgYi0vL09NTfX397e2tiIeGu9BKtcADg8Ph3Xf5tHRUbkIEXsWfLCj owNvm5qaYrcIREQUH8yDqUvyYGNj4wdrjH2ZB4kogZgHk43H43G73YiH6rrC 8vLyiYkJ+2Noa2uTG8jIfWbwgk+iJyLSgCkPfvjhhypc7Nq1i3kwmUkeNBof H1d9mQeJKIGYB5PWvXv3RkZG1MMpmpqa1FWBIU1PT3d0dNTW1ra3t/NOMkRE ejDlwYsXLxrzBfNgMrt//35fX1+PwW9+8xvVl3mQiBKIeTDJYQfR398vT46Q 00cTPUdERJQYpjw4NzdnzBezs7OJCDrkAOZBIkog5sGUsLS0pJ5XiBd4m+g5 IiKieLO4fpBSGvMgESUQ82AKGR4elosK8/PzBwYGwrrPDBERpTrmQV0xDxJR Alk87yAOEr30qcfj8TQ1NUnpVVdXLy4uJnqOiIgoTpgHdcU8SEQJxDyYisbH x4uKinJ9BwrxOtGzQ0RE8cA8qCvmQSJKIObBFLWysqIOFHZ0dPDcUSIi7TEP 6op5kIgSZWpqKi8vD4GitbWVgSIV9ff3u1wufIO1tbX2n0ZBRESpiHlQVymU BxH3vnmp1uLv6ycKmQeJUgXDoB6mp6fl3NHi4uKZmZlEzw4REcUK86CuUigP 2vxjHiRKfgyDOlleXq6ursa36XK5BgYGEj07REQUE8yDukqJPDi3/MuB2UWb f2xbEiU5hkH9YD/S3t4ulxO2tLQk/26FiIjCxTyoq5TIg0SkDYZBjY2Ojubn 5+f6nlm/ylM1iIj0wjyoK+ZBIoobhkHtzc3NFRYW4iuuqqrC/iXRs0NERI5h HtQV8yARxQfDYJpYXFwsKSnBF11eXr68vJzo2SEiImcwD+qKeZCI4oBhMK0g BiIM4utGMEQ8TPTsEBGRA5gHdcU8SESxpsJgTU1NVlYWkoKz43e73Rhtfn6+ zQEKCgrwFnPl7GyQEXYuVVVV+NILCwvn5uYSPTtERBQt5kFdMQ8SUUwZjww2 NzdnZGScOnXK2UkMDAxgtHv27LE5wN69e/G2r6/P2dkgk9XV1bq6Onz1SOLT 09OJnh0iIooK86CumAeJKHZMp4m2tLQkQx7Myck5e/bsxMSEs7NB/rB/aWpq QgVANeABWSKilMY8qCvmQSKKEf9rBpMkD1I84atHBWAkJCJKdcyDumIeJCIL Ho+noKDg2LFjW7duPXTo0K1bt+7evav6Zmdnnzt3zu1237x5c+/evbt3775+ /frMzAw2LC6XKycnByngypUrg4ODMrzKg7W1tUePHt22bdvJkydrampMd5jp 7e29cOECxrZjx46zZ8+2t7eb5mpychJzdeDAgV27dmH8Mlpj3LMeQI4PYhiv LyridWNj4+3bt/EiMzMTn8JiGm+Midlramq6dOkS+h45cqSsrEwG7ujocLSw tcVISESkAeZBXTEPEpHApgDpz9hldnYWoSzDZ8uWLfICmUiSFCBhGXuJ/fv3 nzlzRsIg4h66bNq0aX5+3ruWBxV0lxeIWmqiSFvScePGjWoAjEoNMDc3hyAp 3d999101NhX3Qg5gvH4Qyc40mEAqVCk1Pz/fNMOisrIyVl+GdhgJiYhSHfOg rpgHiQiQ10pLS10uF/KRykGIaUg9iIRyqR2GOXbsmPGET8mD0NjYuLy83NTU hNc7duyQMIiOQ0ND7733HjrW19d7DXkQ48HYMCEMs2HDBnTp6urCAOPj4/K2 qKgImyYZ4J133kGXgYEBmejhw4fx9vjx4wsLC3LkTgZQcS/kAP55UGYJ84BE XF1dLV36+/sxwJ07d/AaY2hubsbYFhcXkXaZByPASEhElNKYB3XFPEhEgBCE ZIT4hkgo0WxmZkZSD/KUGgx5TTpKQpQ8mJ2drQY4dOiQhEF1zWBWVhaGKSws 9K7lwY0bN87OzqqPXL16FR3Pnz+P19euXZMEajyD9OLFi+iI/17fIUu8RmbE 7KkBbt68qeJeyAG8gfIgZkmOYIp9+/ahIwIgXl++fBmvsVCq79LSkoRc5sFw MRISEaUu5kFdMQ8SpTmPx7O8vFxQULC6uoq3PT09aK4jVXV3d0v0Q9O90EA6 yjV9kgcbGhpkVGjh37p1Cx/H2FSgw1sMI8/+kzx48uRJ4wzIATgJa0eOHJHX xinKQckDBw5ggM7OTrw+ePCgcQxIdmoMIQfwBsqDplvNyBHAuro6NXBvb69x AImozIMRYCQkIkpRzIO6Yh4kSnN1PohF8hYbhIqKivLy8tra2ozgqqqqvGt5 EBHSa7ibKOIePqvGj46mPGi8WhCGh4fl0jy8Vtf9+du8eTMGwLz5J0pMWmW6 kAN4A+XBEydOGIe/cOECOmIpEF5k6qOjo8YBcnJymAcjxkhIRJSKmAd1xTxI lOYmJibQMpdjYWJubs7lckmw2rJlS28gcsKn5MH+/n4VBuWET+PY/POgnPmp dHV1qbAmI8R4/KdojG9HjhwxjgFhTY0h5ADecPKgdy2imh5eL1dWMg9GjJGQ iCjlMA/qinmQiJqbmxEAFxcXVRdkNHRBGNywYYPxwrqVlZXs7Gw048fGxrxr 8a2trU09Z/D8+fMh8+COHTuMlwcWFBSg45kzZ/D63LlzpgsSvb5TQBEzJXy5 3W45mLi0tKQGqK+vV3Ev5ADeMPPg8ePH5aRZ1RebzczMTObBKDESEhGlFuZB XTEPEhG2A4WFhY2NjaoLNgvIbhL3kNEwgHSXJy9s3bpVwiMGQLhDclQ3kLGT B6G0tFT6IlfKQxzkrjUNDQ0ZvrvBqONxSKNy85aioiKvL0RIFrtx44aESgyw c+dOFfdCDuANMw+qB1LIbXaQiOUGOCoPYvxIxHfu3FEf7+npQRckU6e+IF0x EhIRpRDmQV0xDxKR15fL0CwfHx83dpR7swDy1OXLl9XTJeTGm17f7V9MdxO1 mQczfM93uHjxooRBjEc+jv9ySxk5Ynj27Fl5MuC+ffvkdjcwNDS0cePGDN9N Y7KyshBOZXgV90IOEFYexCypB0xs2bLF+AhCyYMyBmRY9XEpKJV5yQIjIRFR qmAe1BXzIBGJhoaGoqIidShQ9Pf3y8MX1E1dSkpKpBda7xIGq6ur1fmfcuNN dcdR79rpoPjvXYtOly5dkpMwJUZduXLF4/Go4TEDt27dMj4gHsnRFFQ7Ojrk IYNyaiiSl2RGmwPs378/Y+2Bhm1tbRmGJyoKuTxQnpno9e0Bi4uLkS7feecd ZD0siyREGQBxJsP3xAr1ccmbZWVlEX0PaYeRkIgoJTAP6op5kIiEPHVCzoo0 WVpaGh4edrvd6iCduoGMOjIYrsXFxZGREVP8VDBOTA4DGK9qNJmfnx8bG7PY goUcwI6+vj7EWONFlJg3ycjIktGMmRRGQiKi5Mc8qCvmQSJS0BRfWVmxM1iU YTCFyPMTT5w4oZZUPYnDGBIpSoyERERJjnlQV8yDRBSWtAqD3rVn2ctddM6e Pbt79255q86bJacwEhIRJTPmQV0xDxKRfWilG+8mmujZiZPbt2/v2rVLXc+4 f//+ioqKRM+UnhgJiYiSFvOgrpgHicim8fHxNAyDAsu7sLAwOjq6vLyc6HnR HCMhEVFyYh7UFfMgEYW0srLS0tKS69Pc3JxuYZDijJGQiCgJMQ/qinmQiCwg CXZ3d+fn50sYrKioYBikOGAkJCJKNsyDumIeJKKAJAkWFBTkrnG5XDxbkuKG kZCIKKkwD+qKeZCITPyToOjo6Ej0rFF6YSQkIkoezIO6Yh4kIiVYEoT8/HyP x5PoGaS0w0hIRJQkmAd1xTxIRMry8vLimuHhYbmbKA8OUmIxEhIRJQPmQV0x DxKRP/XQ+aqqKh4cpIRjJCQiSjjmQV0xDxKRiQqDaIH39vby4CAlA0ZCIqLE Yh7UFfMgERkZwyBa4MiDPDhISYKRkIgogZgHdcU8SESKKQyiC/IgDw5S8mAk JCJKFOZBXTEPEpHwD4MwMDCwsrKS2BkjMmIkJCJKCOZBXTEPEpE3SBiE1dXV BM4VUUCMhERE8cc8qCvmQSIKFgaJkhYjIRFRnDEP6op5kCjNMQxSimIkJCKK J+ZBXTEPEqUzhkFKaYyERERxwzyoK+ZBorTFMEgaYCQkIooP5kFdMQ8SpSeG QdIGIyERURwwD+qKeZAoDTEMkmYYCYmIYo15UFfMg0TphmGQtMRISEQUU8yD umIeJEorDIOkMUZCIqLYYR7UFfMgUfpgGCTtMRISEcUI86CumAeJ0gTDIKUJ RkIiolhgHtQV8yBROmAYpLTCSEhE5DjmQV0xDxJpj2GQ0hAjIRGRs5gHdcU8 SKQ3hkFKW4yEREQOYh7UFfMgkcYYBinNMRISETmFeVBXzINEumIYJPIyEhIR OYR5UFfMg0RaYhgkUhgJiYiixzyoK+ZBIv0wDBKZMBISEUWJeVBXzINEmmEY JAqIkZCIKBrMg7piHiTSCcMgkQVGQiKiiDEP6op5kEgbDINEITESEhFFhnlQ V8yDRHpgGCSyiZGQiCgCzIO6Yh4k0gDDIFFYGAmJiMLFPKgr5kGiVMcwSBQB RkIiorAwD+qKeZAopTEMEkWMkZCIyD7mQV0xDxKlLoZBoigxEhIR2cQ8qCvm QaIUxTBI5AhGQiIiO5gHdcU8SJSKGAaJHMRISEQUEvOgrpgHiSIwkFA6hcHE lmSil95JuQmV6KV3gIqErJNOYZ10CuskJQnmQV0xD1LaQqpqbGwsLy+vrq7u 6OjAumD/s4ndO2sTBr2JLslEL72T0qrtvbKy0hxIf3+/DIDV+c6dO1i1a2tr BwcH1dqNVQZv0RErfkNDw+TkpHG0EgnTqk7GqCQF66RTJck66VRJUpSYB3XF PEjpaXR0VNoMpaWl+fn5eFFWVra6umrz44ndO+fqEga9iS7JRC+9kxLZ8o57 2xt7roCz0dTUJH1LSkrwFv8LCgrwoq6uTj7Y2dmJty6XC+u7fGRsbMw4ZqxW aVUnY1eSXtZJH0dKknXSqZKkKDEP6op5kNITYqDaWSAG1tTU4O3Q0JDNj8te cibuZLrahEFvoksy0UvvJGn/3I67hLS9vb7VdmXN4uJiXl4e2oGzs7Po1d3d LUUhg9XW1uItvnSsNfLjz8LCAnph9cfr6upq05jTrU7GriRZJ1knk60kKUrM g7piHqQUJbEIe4q2trbKykr8x2ubn8UupqSkpKysTHXp7+/HvqOzs9P+1BO4 d45zGIymqO2MPH3aOTEtyZRreztYGg0NDZiHnp4eeVtUVFRYWKgO98/Pz2MF x8jRBa1KNBplDfJ4PLm+MwT8Zyy16mTSliTrJOtkspUkRYl5UFfMg5SiKioq sLUvKCjALkAO9mGvIb3u3r2LvvJbIszNzWF/ZL0zqq+vxxiGh4dtTj3l9s7R sCjq6LEknRp5yrW9nVqF5dzv6upqaQeiWYi3NTU1WJ2bmprq6uq6u7tV61HO BGhvb5fLh3N9p16bRphydTJpS5J1knUy2UqSosQ8qCvmQUpF8ksgtvZtbW0e jwd7isLCQux6pC/2FOiVn5+PPQ72NXKJgcW5oDJ8cXHxysqKzRlIub1zxKyL OnosSafGn1ptbwdXYWl5ut1uebuwsJDrBw1O2dOh2WnsXlJS4n8jqdSqk8lc kqyTTpUk66RTJUlRYh7UFfMgpaLp6Wls7bGnkN8MUYex98FbNUBvb6/sceSC ArwNNirshlw+YT10LLX2ztEIWdRRYkk6Nf7Uans7tQqjSYm+5eXlpjHn+g4T YAc3Pz8vTUqs6ZiK3IkCw6OlWlxcjNctLS2mcaZWnUzmkmSddKokWSedKkmK EvOgrpgHKRXJ5X5qJzI3N4e3zc3NxmHkUWK5axeeB9TV1SV7pXBvTJ1ae+do 2CnqaLAknRp/arW9nVqF1RMDVRc5RpCXl6cO9w8ODub6zhybmprK9Z1sJt2x 7ysqKkKX5eVl4zhTq04mc0myTjpVkqyTTpUkRYl5UFfMg5SKsGfBdn5+fl7e yh5BPZzI69sByVkouWvnpZjGcO/evfb2dvTFLgMDhzsDqbV3jkbIoo4SS9Kp 8adW2zv6VdjrW4vlEIPxTG+5wgjrteoyMTEhJSN3Juzr61O9pME5OjpqHG1q 1clkLknWSadKknXSqZKkKDEP6op5kFJRaWlpXl6eus2mNACw85K3al/T29uL vUPAPU5PT0+u75mDS0tLEcxAau2do2Fd1NFjSTo1/tRqe0e/Cnt9N6ZAr9ra WlN3OYVMDS/NwqGhIblwyXgKmdzpwvQtpFadTOaSZJ10qiRZJ50qSYoS86Cu mAcp5cjDao3PFSovL3e5XKomy4OH1Mkqsscx/ki4vLyMvVWu70KDeoNUef6g QwUZWsiijh5L0qlJpFDbO/pVWMihgfb2dlN3+XhhYWF3d7fcZrCoqMjj8WC6 cgpZc3NzZ2dnZWVlruFmFEoK1ckkL0nWSadKknXSqZKkKDEP6op5kFKOXCZw 584deSsnkGCzbxzGdPNq09vx8fHcQPx3PcGk0N45GnaKOkosSaekUNs7+lVY SEtycHDQv9fQ0JA6Ia2iokKduoYX0lAU9fX1/mcIpFCdTPKSZJ10qiRZJ50q SYoS86CumAeJIpBCe+ckx5J0Sgq1vePj3r17aGd6PB7/XnJnwmDPl2GdNIm4 JFknTVgnnRJxSVKUmAd1xTxIFIGw9s6nTp16+uPWrVu3fv16dJ+YmNBj7xyx sEqypqbm7bff/s53vvOVr3zlu9/9Ll6XlZUZB2hsbETxvv7662lYkmG1vffs 2fNvllCSqd72jlhkbe+G7pa6rqa2/g5jx/GpCXTEX+9wP+uktczMTFS8n/70 p/69Kioq0Oupp55inbTDtMd59tlnsUnctm0bijHcWq1fnaQoMQ/qinmQKAJh 7Z2feOKJB4L43Oc+l52dnc57Z/slifbMpz/9aVMB/sEf/AG6q2GuXbuGjn/2 Z3+WhiUZVtv78ccfD1YnRXV1Ndve9iHu/cPW7+Hvh0dfMXa/1Vgo3Tff3Ms6 ae2xxx6Turdnzx5Tr5ycHHT/zGc+wzpph8Ue55vf/GZzc7P9iq1fnaQoMQ/q inmQKAIR7J2/9rWv7V6zffv29evXP/TQQxIJu7q60nbvbLMkKysrpT3z/e9/ /8aNG0grN2/e/Pd//3fpePjwYRmMedBmg1ny4COPPJIRRFNTE9ve9l2vc63l vn3G7luy90v3q7W3WCetqTz4p3/6p1jfjb2YB6PZ4+zYsWPDhg3/8i//8olP fALd//qv/3pyctLmqPSrkxQl5kFdMQ+STXfv3u31wYtEz0viRbB3fuqpp0zd W1tbP/WpT6HXoUOH0nbvbLMkX3/9dckvpu5f//rX0f1b3/qWvGUetNlgljyI FrjN4dn2Dqi+u/l85XX8vXjmF5L7Xjm/QbrI3/cOPCfd9+afwNvStirWyWBU HoR/+qd/MvZiHnRkj3Ps2LFPfvKT6PXqq6/aHJV+dTIabAV5mQf1xTxIIXk8 nuvXr2/YsEEOHOAF3ga8jjtRlpaWBgcHl5eX4zZFR/bO8NWvfhW9nnnmmSTZ OydtSX7ve99DQf3rv/6rqXt2dvYXv/jFf/iHf5C3yZMH41+SuubBpK2T8GbW Fol7Nv8wPOtkMJIHv/KVr0gk3L17t+qVbHkwmevkjOUeB6WKXkiFNg8R6lcn I5P8raC4YR7UFfMgWVtZWTly5IhsAA/4yCYRHZPq/l3j4+Mul6u6urq/vz8O OxdH9s6Y589+9rPo9cYbb9gcVRx+rU3Okty3bx8K6tOf/vSmTZswY8EGS548 6I17SeqaB73JWifh23v+I6w8eK7yGutkMJIHX3rppX/+53/Giz/5kz+pqKiQ XsmWB71JXCdnLPPg6OjoH//xH6NvVZXVoWqN62QEUqUVFB/Mg7piHiRrZWVl 2O5lZmZiJyu7htu+G8GhY3l5eaLn7mNktyItgVjvXKLfO09NTck5kHDjxg2b o4rP2TtJWJLDw8Nf+9rXpLh+//d//2/+5m9effXV7Oxs06/cSZUHvfEtyQjy 4Je+9KXnAzl+/LjN8cSn7e1NyjoJWdXZZ8qvvHDm5xL3Xj77Nt6qP3Wy6KYb e6RL74jVXUbTvE6qPFhZWfn5z38er//xH/9ReiVhHvQma52cscyD8PDDD6Pv sWPH7IxKvzoZgRRqBcUB86CumAfJ2r59+7DRww56wABv0XH37t3jSaazszP3 47Alx6Z7aGjI2QlFsHdGC+era7785S/LkUG54ZvN8ai9s7PLElASluTExMS6 detUuYmHHnpo//79aphw86CzyxJQ3EoygjwYzJNPPmlzPKrt7eyyBJSEdVJs uLZTcl9OY6Gx+/97+CXpbudhE6yTKg/iNVZqqYq7du26HWkedHZZAkrOOmmd B//+7/8efd988830rJMRQCUM1gpCAynRbbR4Yx7UFfMgWdu4cSM2er29vcYt Id7KuRO56SqCvbO/L3zhC2+99VZYjyCU6SZ66Z0Ubtvb7Xbn5eX95Cc/+bu/ +7sHH3xQShI5UfqGmwcTvfTOs9lgljz48MMPvxrI6dOnbY5Htb11Em6dVMcB B8aGVMe+kQHp+P8cesHmeNK8ThrzIHz729+Ws0bLy8sjy4M6cTAPfvGLX0Rf 42N60rBOhkXODg3YCkIDKdFttHhjHtQV8yBZw14DG73W1lbjlhA7XHTcvHlz Q5Kpq6vLz883bczRpaysrLa21sEJRbB3fvTRR8vWYD7HxsZsftx/7+zgggST nCVp0tXVJW3I3/u938PeeSb8POjgggQTt5KUMdtsMDt+/aCDCxJMctbJzqGe gE8evFGfJ93fvbHH5qjSvE6a8mBVVZWcNfqtb30rsjzo4IIEk5x10joP/uEf /iH6ZmVlpWedjACaOsFaQWggJbqNFm/Mg7piHiRrV65cwUbv1KlT/f39shnE C7xFR/RK9Nx9jMfjKS8vV7uSioqKnp6excXFWEzLwb1zWGS6sVgio2QryZaW FrQMP/vZz46Pj/v3/Yu/+AsU75EjR2aS7/rBeJZkYvNgLJbIKNnqpHK5Jkdy 37acg8bum27ske4IhjZHleZ10pQH4cCBA3ICwKuvvpqE1w8mbZ202ONgPqVI sVFNzzoZgRRqBcUB86CumAfJmtvt3rRpE7Z7hw4dKvDBC7x95513uru7Ez13 /5PaocRnV6JxHkzOkkTEC3YPBMmD169fn0myPBjnktQ4DyZhncxvLt1wbSf+ Hj/y8v9/XujBF6SL/Klbj75ybgPeniy7xDppzT8PwqOPPqpOsE+qPJiEdVIJ tseZnp5+5JFH0OvrX/+6zVHpVycjEKwVhI7olei5izfmQV0xD1JIvb29W7Zs yTDAZvDChQt5eXlTU1OJnrvfwQ6lpaUlnrsSXfNg0pbkD3/4wwd8z5vYtWuX uqfo6OiodEdDEVVxJpnyYPxLUtc8mJx18qcXN4X1pIlfXMlknbQWMA+qs0aT Kg8mZ51U/Pc4+Gx+fv43vvENKcnkuaN1/EsyMv6tILxFx0TPVwIwD+qKeZDs WFpaqq6uvuKDF3fv3m1tbcUON0kiYfzrsK55MGlLsq+vTz1v4lOf+tTDPoiH EhLlZNGZZMqD8S/JCPLggw8++JkgNm3alCRt7+Ssk4/u+VFYefBS9U3WSWsB 8yAcPHgw2fJgctZJRfY4n/zkJz/rIw8cFJ/4xCd27txpczxa1smImVpBeJvo OUoM5kFdMQ9SZO7du5dUkTDOwto7P/nkk9gRP/300/b3wonaO8ef/ZJENVu3 bt1f/uVfqrbNH/3RH/3t3/5tQUGBGubGjRvo/ud//udpWJJhtb1/8IMfBLzn rfL2228nSds7/uzUydt97fVdzVuy90vc23BtJ96qv+dOvSndT5Reki6jE6Hv H5XmdfI73/kOKt7LL7/s30ui4kMPPcQ6aYfscYwQpf/qr/7q+9//fn5+vs2R 6FonKUrMg7piHqSIpXMkDGvv7CD99s4RlOTQ0BBaNVVVVW63myWphNX2dlA6 t73fupIpuS+vucTY/fsH10n3/tFB1knWyehxj0NJgnlQV8yDFI20jYTcOzuF JekUtr2dYr9OPnH0VYS+/2Pbvw6Pj6iOY5Pj6ILujx95mXWSddIR3E5SkmAe 1BXzIEUpPSMh985OYUk6hW1vp7BOOoV10imsk5QkmAd1xTxI0UvDSMi9s1NY kk5h29sprJNOYZ10CuskJQnmQV0xD5Ij0i0Scu/sFJakU9j2dgrrpFNYJ53C OklJgnlQV8yD5JS0ioTcOzuFJekUtr2dwjrpFNZJp7BOUpJgHtQV8yA5KH0i IffOTmFJOoVtb6ewTjqFddIprJOUJJgHdcU8SM5Kk0jIvbNTWJJOYdvbKayT TmGddArrJCUJ5kFdMQ+S49IhEnLv7BSWpFPY9nYK66RTWCedwjpJSYJ5UFfM gxQL2kdC7p2dwpJ0CtveTmGddArrpFNYJylJMA/qinkwrYyNjWVlZWVnZ8dh WsZI2NPTg7dxmGjccO/sFJakU9j2dgrrpFNYJ53COhmxqampoqKi8+fP7969 e+/evWfOnKmvr/d4PBGManJy0tmWzMrKitvtdnCEccA8qCvmwbTS0dGRkZGx ZcuW+ExORcKcnJzR0dH4TDQ+BhIq0UvvJJakU3ITKtFL7yTWSaewTjqFdTIC aIFUV1e/++67GX527tw5Pz8f1qiys7PxweXlZadmb3BwEBE15Soq86CumAfT SpzzoNf389fJkydztTtxlHtnp7AknZKebW/svyYmJsJq2oXEOumU9KyTscA6 GYGioiJJf4cOHWptbcWGorOzMycnB00gdNy3b9/S0pLNUa2ursqoHMyDBQUF GGHKVVTmQV0xD6aV+OdB2YrqFwm5d3YKS9Ipadj2Hh8fLykpkRlobm52arSs k05JwzoZI6yT4XK73Rs3bkTzAwHQ1ModGRmRg4ZHjx612QBmHlSYB3XFPJhW THkQ2/mzZ882Njbevn0bLzIzMw8cOHDr1i21xZMB6uvrq6urjx07hgEuXryI 4VWd6e3txQClpaXGqRQVFaEjPjs8PHzmzBnZil65ciVXr0hIRAk0PT3tcrn6 +/uxverp6cHmZWJiIvrRYgOFzRTG1traqtlVz0Tp49KlS3JeKFq5/n3b29ul ZYJNhzeclsypU6fQRkLf7Oxs9B0dHb18+TKmgrYT8p1xE3T9+nUMgM2U6oJt C7rIDRywhcGnMMIdO3agIyYRm2JwHvOgrpgH04opDyIG4q3/2fXYsklDSAYQ Gzdu3L59u7zOz8+XMTQ0NOAttmbGqZw8eRIdW1paOjs7jaPF9pmRkIgc0dzc XFtbq96iLTcyMhLl7oxhkEgPu3btQqujqqoq2ABbt27FAJLOwmrJHDx4EH33 7NmD1++88w7+Z2ZmSjsK45ydnZWPS3sJGyU1QkRLdNm9ezdeHzp0yDhOtM1i UQixwDyoK+bBtBIwD8KxY8e6uro8Hk91dbV06e/vNw1w9+5ddLlz545sAGXz Zb0V9X78LAvtH0JBRHFTX19fUVEhr7HtwlalvLwc/9Exss0LwyCRHlZWVqTh gYZNsGFOnz6NAc6cOeMNsyUjfVUelAHQ/dSpUxm+axVlAOs86OX5opRkmAfT SsA8uHHjRuPdGPbt24eOcjGODIAt3szMjBpA7rJ19epVb/hbUUZCInLE8PAw tiTYBPX39yMJFhUVdXd3o/VVWVmJ7sbTtOxgGCTSBlZ/aXiMjY0FG8blcqmD fRHnwRs3bqjhZ2dn5ddyaS8xD1JqYR5MKwHzIDZrxmHkPPm6ujo1wIkTJ4wD dHV1yTml3oi2ooyEROSIwcHBsrKy6upqYwDEHq2wsLCxsdH+eBgGiXTidrul 4WGMYyaIchjg+PHj3ijyoJxJpRw5cgQd29vbvcyDlGqYB9NKwDxoinsXLlxA R7kwRwa4dOmScYDR0VE5quiNaCvqZSQkoohgc4EGmJy7royPjxcXFxu7YPOC kGh/nAyDRDpBy1YaHtIOCUgaKllZWd4o8qCpAXPu3Dl0LC8v9zIPUqphHkwr keXB8+fPGwfo6enJ8D27x7u2FZUz8BU549QiD3oZCYkoHNhiNDc3Y4uBRhT+ q5s2eH1HCdFlbm5O3mKPhnh4584dO6NlGCTS0rFjx/wbJ4rb7ZZzO+vr670R tWQkD46OjvpPtK2tzbuWB403DkV35kFKWsyDaSWyPLhjxw5jJZFnvF65cgWv sanM8N1tRvVFm0putGWdB72MhERkW29vL1IeQh+2GwsLC8Ze2MigV0VFBdp4 8/PzTU1N2EYFvMm8CcMgka6wHZC2h/9PQ1jZz58/n+G7ubo8kj6ClozkQbmy RqDX5s2b0VHOXUfuy1h7noUoKytjHqSkxTyYViLLg6CeyzM5Ofnee++hi1ye 093djdcbNmxQP4Ll5OQYz9PARlXeBnw0GCOhCYo3YLsUzeBz587V1NTEf5a0 gTqWlZWFyODgODE2jFM9fiWehoaGMGk5MclBIZfINACaNHgbh5UXcU+1rDD1 a9euYROEzQ7aaegyMzOjHkxfVVW1uLgYcoThhsGA6+b4+PitW7fQjJQHkF28 eLG9vT3W0TJGX31KGxgYQJmohwvErVrGSAI3LOFaWVlxdqPqIKybGb4b4iGI yYbC67vri9wIFNQuNYKWjORBrPhyOz4McPPmTfn9XAY4evQo3mLHLW9RG9H0 MubB4uJi/+txkh/zoK6YB9NKxHkww3fZNTZcmzZtUqfce337AnmIz8aNG48c OYItoRpenbcvA+CDxh/fFEZCgXKQG7f6H0jFSirPKkIbOCHzpgf5tRbtRgfH KdeDmO7IFB/ygzYaNs6ONuQSmQbYu3cv3vb19Tk7G/4Q9+QZN9jmXL9+HS09 OXEUsUh+4ZfjhqZLC4MJKwwGXDfRvMTHMwLBptLj8USxrCHE6KtPaabrv+JW LWMkgRuWsAwODmK7mrRHuLDmoq0iayVS4YEDB+QR8MI42xG0ZCQPZvieOYip HDx4UPqqM0irqqpkABTR4cOH5fRUYx5EGpUu27dvl1NMUwLzoK6YB9OK3Bp0 27Zt8lbOZje1KxD6MtZOqlc3IJWz4mXTh8YYNp5q+JGREdn5yrYUmUWaSWho yQCNjY3qkfcBm0mMhN7gJ9bOzMwgsEsv5sFoMA/aEW4ezMnJQSM84NH/6GE7 09nZKXuoO3fuVFZWYluBVIgweP78eUx0bm6uuLhYNlb2hXtkMOC6ic2gSn8V FRXYDGJrefnyZeloujGFs5gH/ZnyYEyrZRykSh5M/jMesfKWl5cb8x2cPHnS /7mE4bZkJA8iCcqP5IAgaRwtJu1yuVQM3LVrlxyFxFRkgKWlJbkfKWCLFJfy cADzoK6YB8mC8QDiwsLC6OhosObT/Pw8NqfGnGiE7hjA4jdzRsKAbc7q6mq1 r2EejBLzoB3h5sHYwUZDHiQxPj7u9W1/0Ljq7e2tq6sz3qJhenradHsZaxFc M+i/bso9tQAB0DQw5ll6DQ0N2ZylcDEP+gt4f8jUxTzoLKzpaIGgATM2NuZ/ Bo6R/ZaMet4EmtAYbbBz1LH1wHQtNlAYIT6bQtcvMw/qinmQLAQ8oTRGUigS Zmdno+GBjTxag3LdEHaLpt+isYNAfDt69OjWrVszMzNPnjwpzyQSN2/ePHfu HBq3hw4dQk7JyspSVzTgxa1bt2Qw6YKRyIUJyZMH7ZQA9nFYECwgSuDw4cM5 OTnG24Bgy1NUVITu6Lt3796LFy8ar7sH7HMxzmPHjmEAjASjMp4KiBlAAbrd bpQMPo4yvH79+szMDEaL+nPw4EGUOUrS2CaXPIi2NEob84y9+dWrV01XvlhP 1Ou7iAwDYHl37dp15coVaZnbbLY5u0QqFNTW1qKGbNu2DXWspqbG1K5AHbtw 4QLGtmPHDnxlxkpoc4msB5ADMRjG62vE4nVjYyO2G3iBGcansJimZ800NTVd unQJfY8cOVJWViYDy7mgioTBQp/u7m7pKPcRxcwYJ4pywyT6+vrwApPDaM+c OdPZ2alGJXNVX1+PEaIXPouRlJSUqOuJUEQYQF0iLVA50RGfRfaUR7Ia101U 1wzfeV/+v3FhAXf7YDFVxzt37mAkqHXyLagjDsZCw/CnT5+W71GOeGKNwHeH Jdq3bx+WWrVRbX71jpBqiXoYrHi9odZ007YOhS8FHtYie33t6qqqKowKI9my ZQvWEcySaoQHPD4oNQTr0dlA5JZoIuRqEmvWa5l/GcrtkkLWK6n22OagSFFp Ueam9p71dydFp57s6fX9nIIuKHm8xmzI6ZcydeOvNOkg4PMH0wHzoK6YB8lC PPOgN3UioewI5DwQ7GflHBLsT9VvgG1tbRs2bJA2pDrDBNBokQGw08/wXWsg 3eUycwWNfxkMjT18BSgWuS9Z8uTBkCWAr0/uOwRyyzUZQD2MSW7vZiof9Qxx jEfim7FwMCFp4KkZMJXb/v371YnNAiUsF/t71/Kg6q7mDSnD5kTn5ubQdvWf bTt50PElkhaj/xIZ704g1SbDd/qTGsD4Y37IJQo5gPFCLdlcGAcTaOWqqJKf n2+aYVFZWanmSoVBuVmoMVhhzcL8Y0XA+iUTRahEg1Zdy6Oo63HURdBotUoY lMeKqZt1WD96DNnHOFqsm+rGgyjegN+1SppCrj2UbwHktVqXZfbUSWUKBjB1 VGeU2fnqnSIFq+qAf/GGXNNN27rt27fL/Ie1yChzuT7L9EGMXBowFtcPGi8Z M8LaJwOHXE1iLeRa5l+G6prWDMt6pQaQZx9kGKq918Z3Z/0EPbmqXTH9pKM9 5kHSDPMgWcBOHzvfYE/wiYWUiIQqDcmV5svLy3J0D/tHGQAN4AzfjcWQArBE w8PDcpkA8p0MIPv3DN+Da+vq6jAeiwdzeNdaLEmYBwOWABZZSgD/Je+43W5p PKBRhw3OzMyMLGxvby/eLi4uylWriD8yfnmLVoccc0QukFikzpFT7X9ESExd 3VocioqK0L5CmUtTR11cJnEM84yPyOlDcv+3w4cP25wohszwXS+2sLAgx7mk aWonDzq+RCoUYDwYG+YHH5RfIeQalvHxcXmLj6OQZQCZYXXSbMglCjmAfx6U WcI8eDwedfKktJru3LkjX0FzczPGhu9dHXpTedAYBr2+A0bGI3f4FLogAF6+ fBkzMzQ0dOPGDWQ3uYUgSgk1DbUow/ArlnpoDj4lp4m2t7fLUkgLNtxHUWO2 5a2dRxxiWmpNx3jwRWDp5HuRg1DGQsMWD/OvrhfGyoI0ioliGTMM+SXkV+8g taYHLN6Qa7o30LYugkWWmoNlRF/UkLt376p7ccjhY4s8iEbO8hpjnZRIa2c1 ibWQa5l/GYZVr+Q8BJShsdrb+e50faK6I1BKKM/YnRaetJgHdcU8SMkm+SOh tJHQYlFdkPtkV4ukg53vxYsXsYvHazWANO/VblT27zdv3lQDpGIeDFYCaGLJ shjPIJ2enpYBkAHRwMjw/WqtWhpo1qI9JkeCVFo0HhgyjVNmQE5bEnIUACWv usid5ZAs5K3kQeNH0P6RFtTo6GjIiWIBpUVq/FrlPN6QeTAWSyShAGVovDLl 6tWr6Hj+/Hm8Rm2RKmc8jVBOdJRxhlwiO4vsnwcxS+qYrHftmc4IgHgtt1vJ yclRfZeWliTkqjyIxqoKg17feXTGZ80LBPlz587JaZ+IAPi4OsXau3abLKxi 8hZzpY4MqmsG5dgKissbfh7ERknemh5FHZDcrtl05E7uRSN3KVTtdrXI6qcA uXDSOEU58hjyq3eQVMtgxRtyTfcG2tZFsMj4jpAZkT6M8yZjlu/I5v1FkZdl lVdP6wi5msSanbXMvwxt1it8EcbRGqu9ne+OeZD8MQ/qinmQklCSR8KAJ4rI EUDjhSfYEaP5UVVVhdwkYUQ9mUj27+r0SG9q5sFgJVBbW5thOOtVOX78eIbv kU/Y5qjzo/bv35+fn49mhtoQyU3YMnxnixUaSEcpYZkBdf4tICAYm3le3+Ut GYbzowLeT0ZOQkPbKeRE5bxB00KhwZlhIw/GYokkFCC2GCckh1FkfuTrwGvj FOWg5IEDBzBAyCWys8j+edBUGnIEUJ7aLANLU1ORtrfKg1gRjHdmwFu0OTFm 40eM45FyMw4wPDycYbiLMiZtCoPetTstSzmEmwex05RMYbqMLiCp56ZT6aQM t27dqgoNDW/VFzEzw3feoOqC6coMyFVjIb96B1kXb8g13RtoWxfBIgu8xtSb m5tRk1Gv5CRJGbOdPIipyCmXyE2qY8jVJNbsrGX+ZWizXpmu9TBWezvfHfMg +WMe1BXzICWnZI6E0kYyzZWx9Y6Gh7qPtJEpDxrzVCrmwWAlAHhx+vRp06fk 8JbEmcnJSXVzb5GZmSm3lEED3r/oFHnktMyA8RY0ckGi5A6BmpPhlwdN8yxp BWEk5EQrKir8G+FyFCNkIzwWSyShwHSAQNrqaPR611qMAW3evNnre8K79RLZ WWT/PBjseabqgc6mw2ry3Gfj9YMmaDBjwY23FjFO1P+nCTnwIYEFcysPqb9x 44bxAJAEEAQKb/h5EOScyWBPhHe73WhCy45VMovppDJ18AsLJYVmfDbr2NhY huFUSa8vB8nwxjxo8dU7yLp47azp/tu6CBYZ3x2ih+maU2EzD87OzmL8Up+N F3iGXE1izc5a5l+GNuuVqZIYq72d7455kPwxD+qKeZCSlrOR0MEnREsbydSs lZ+U29ra7t69Kw+uRUvj6tWraAkPDg7K5R6mPKjuZOJNzTwYrAQk/hjbewJt j4yP34UD7UC8xZBywAX/FxYWpCWzZcuW3kDkHDn/ZqrNPKhOSBMS22/7WE9U BsDwxo9L+ypkHozFEkkoMJ3SJr//y/zICOV2jibG+GaxRHYW2X4e9K61vU1H beTKSos8uLS05HK5jJ+ymQfVoyXQ2DadSCkPjNi3b593LUqYrpKW01yD5UE5 yTDYyZnyI4NcGSoZxHRIFCt+hu9cvoCF5h+OsO0yhqOQX72DrPOgnTXdf1sX wSKXlJTI2+PHjxcWFmIjMz8/L5fd2cmD+OIkwqM7qpP/AlqsJrFmZy3zL0Ob 9cqi2tv57iQPGm8cKmcLMw+mM+ZBXTEPUjJzJBKikqMFhTGYWgIRkyaEsaGO 9obcn216elr2mJs2bTLeuFt+BFYNHj3yYLASQCtFfl03FjgaeHLzTHwXAwMD N2/eNN4nZHJyUiIhvm5pC+Gt8TK0lZWV7Oxs1AQ0Hb1R5EHjOVdI7nLEAQ2e kBN1u93ytRoXqr6+3k4jPBZLJKFgx44dxsNe0jyTaCOHa40XJHp9x9pQiyR8 hVwiO4scVh6UU9GMT17GuiktW4s8CM3NzVhq9XgOO3kQ7X8Jg7LuoKCMe7qi oiJ0lCcOSEkaG8bqDqLB8qCkjwzD3YqUoaEhqcnSS34kweSMw8gsSaM64jxo 8dU7yDoPhlzTvQ7lQYnnxpqDZZeVV8rZIg/i65OjvVu3bjU+OkGEXE1izc5a 5l+GNuuVRbW3893JNtN40oJxEl7mwbTEPKgr5kFKclFGQrQ0CgsL8fGamhq0 akyPk4uMtJF27twpzXvMoVz+L4f/pGWyceNGdRkU5kFumoFdrXTx37+rs+lM T/ETyZkHg5UAGhXSyMcMy+ZF3R0dxYKvQBq0aDarpy3MzMxIKxplgvabtEPQ VFNXD8lzCtCik1KNOA9inuXmJJglOcqDUa36WE8Uw8tCqTMPsexyy5eQeTAW S6RuMqliNaqZtJDlrjVSD9VzGWSGpR5KMzLkEtlZ5LDyoHoghdwGE4lY7oKi 8iDG39bWZrxvJ9qi6IJFwzxXV1fLbITMg8YbyPgXFGqdlIP8OCBXd6Kg1HEQ OYVVtYr91018oerGmyht1aLGckmJbd++XU5IwHJJa18dTMdIpMkt6SbiPGjx 1QcrRtPTNu2wzoMh13SvQ3lQDlS5XC7pi6nIvVMAtcJrmQeljqHWGedBCbma xJqdtcy/DG3WK4tqb+e7kzswY6slY8D+Vyah8mBxcXFGbB50QkmLeVBXzIOU /CKLhNjfYceHT1VVVaGlh72hHGWI/gbR6tEAaMxnZWXJPUnUrhmNLrlFGxql aLScOXPG+EgpOcTgv38HOcsU4/E/hyc582CwEvAaWiNomF25ckVO1kKxyC04 8NVICaAZdvz48dOnT0sRYWBpEamHvqFddPnyZTU5uU2lN4o8KPOMBozcaz3D cFeQkBNFzZHLdrC8WGr5vjLsnaTn+BIZH0KHCnPx4kVJBEeOHJEyxH91ESsq IZrKUsj79u1T10+FXKKQA4SVBzFL6gETaFgaLweTPIi2/ebNm1ErTDUNbVqs +Fhwyc7WeRD5zngDGeOD2FDZ8NXLdLE4MjxiqSwXlhQlhtVWDS950Bto3USE kea0VGOUj/ogBpNjvl5fcpTKj5HjC0VpyLdw8uRJSTrR5MFgX72MM2AxWlbS AKzzoDfUmu51KA/KTzdYolOnTmGjanzcpITEYHlQbrEiMG97Pg7TtbOaxFrI tSzg+SR26pVFtffa+O6w65QBsPE8fPiwevKjyoPqqR8I7OqRlKQ35kFdMQ9S Sgg3Eo6MjBQUFBQWFpquccMuFR2j3NFLawQ7VtWgRYvC+OQv7GfVPl36Yn5k ny47TckmxusyAOlVJUfT1Y5yuqnx+Q6JFbIEvL5jJeqxWWjIHT161Hg3PHyJ aOOpx6+jVYMmjfHMJTRB5SQxgZhQUlKi+kovYwNJblNpvD+nnMuk7lEv84xk bXysszy8zOZEvb6nIchVSxm+Zj8a2Blr1+OE5OwSqftFyEmYUsho1BlrDjbv t27dMv4cgfhguoIy5BJZDyBtSLlrq5wprZ6oKOTyQHVeJXY3xcXF8ugufCNY FkmIMgBWcDRN5X4XQhr2cjWT6XYuMlFTuaFeISBgPGjYG8MRpiWn2GX4fhBA pjDeoAarp7q7EaaOmim3csU2RwYIuG7OzMxglTQWL74ChH3TSYmYbeOT1pF6 8FYdJvYvNAlcxttvqpuryDyH/Oox2xlrtw3xL8aw+FfLiYmJjI9nN+s13X9b F8EiY8sgJ3aqeohaJCdVSlWUJ1aoy+VUDZGzIoOR5bKzmsSa9VoWcH9hXa/s VHtvqO8OO0qsTWoSGFIOpqM6yQD4XlSaNp7NSxpjHtQV8yClCvuRcGxsDIM1 Nzcb71Xu9W3H5F7ljuRBNO8xQkzLeHt8BZPATKKdGdZ9bLCznp+fd/DWNzFi pwQEeqEZE2yJ8HEUEZpewbZCaG/g426328Hf6mWejY/lCnei+I4whsi2nLFY IhQyitFU2xWsOHLHS4uvKeQSRbPISl9fH5qpxosoMW+SOOR3EqzapkZvWNQN ZIyPljAejVpYWBgdHTVedmeEGUMpmRrMSrB1E98j6hJGi6lbfKcoOiQpx2+V bP3Vx5P1mu7gJBC3g32D0bCzmsRaBGtZsHplv9p7Q313qNX4uPFhl/6zLWfU 259tSl3Mg7piHqQUYj8SGtucqktlZaXL5YqmwSkCPn0vrbAEKAJy0A1tVNV0 lIegZaw9mjyaPBgwDHqDnMVKpDdWe4oR5kFdMQ9SaongWkLU8O7ubiTB8vJy ix857WMaYglQBNTFXFu3bj179qycBbd//36szgU+sl7La7lPiE3BwqCXDWNK S6z2FCPMg7piHqSUE1YkRABEDEQY7OrqMlb1sbEx4506wiJXP0V/X5rUxRKg yKCZqq5XkjBYWlo6MDDQ74P1VB5HAsanYFtT1wz6h0Gv72o11FXHH8RAlMxY 7SlGmAd1xTxIqch+JOzu7q6srPQ/fRQNTnWnESKKG6y8ckGT/6M2Eevsx0Ax Pj5uEQaJiMhBzIO6Yh6kFGU/EgZsJfb39zMPEiUEdjoBb14RVh5cWVlpaWnJ 9WlubmYYJCKKNeZBXTEPUuqK4FrCycnJ8fHxiYkJfDA/P39ijfFJB0QUO/Pz 86WlpYh+t2/fNoW4kpIS9eQ+C0iC3d3dWH8lDFZUVDAMEhHFAfOgrpgHKaUF i4SIeAGbiEVFRbmBRH/TUSKyo7q6GklwaGhIruoN67OSBOXOMwIj8T/vlIiI YoF5UFfMg5TqAkbC3t5e05Poxa/WyPmiqz546x8el3xiPvdEacPj8SC7yXqH tz09PVhtbd7y1z8JCuPjs4mIKKaYB3XFPEga8I+EyINlZWUWZ5FZ3E8GTdaq qippbVZXV9+9ezdW802UTup8bt++LW+x66moqCgvL7feBwVLgpCfnx/TB6AT EZER86CumAdJD6ZIiDyI1wEPEQqLPIhRtbS09Pf3YzxovhYVFWE1idmME6WL iYkJrJXG57zMzc25XC7EPYtPLS8vL64ZHh6Wu4ny4CARUfwxD+qKeZC0YYyE t2/fxguLQ4Szs7PIjCHHibWjuLi4vb3d6ZklSkfNzc0IdEh2qktXVxe6LCws hPyseui8HL7nwUEiojhjHtQV8yDpREVCdRDB4hChTWjBov3pyOwRpTnscQoL CxsbG1UX7IBu3769srJi/UEVBrGCy9F/HhwkIooz5kFdMQ+SBhYWFlpaWqp9 kN3Q4FRnlFlfRRjS8vJyfn4+IqGDc0uUzsbGxrBijo+P2/+IMQxidUYe5MFB IqL4Yx7UFfMg6QHVeHh4uLS01P+mEyEPEc7OznZ0dDQ1NSH3ocGJ1z09PWhz ImO6XC60PI2ntxFRlBoaGuxflmsKg17f3aJ4cJCIKP6YB3XFPEg6CZgKi4uL LQ4RejweuXVhbW0tIiH+V1ZWYgxor5aXl6MJyqdOEDlLnjph5+GD/mHQ67sZ VMjzS4mIyHHMg7piHiT9SCosKSlRkbCzs9Ni+JmZmfz8/Lq6Oq4LRPGBoBfW NYPGn3Tk8YVERBRnzIO6Yh4kXaFiDw0NIehJJJycnLQYmJGQKKkEC4NERJQo zIO6Yh4kva2urtbU1MgdR+VR9cEwEhIlCYZBIqIkxDyoK+ZB0h7ak/I4QnlU vcWQEgmNN8MnojhjGCQiSk7Mg7piHqR0YHxUfchIaD0AEcUOwyARUdJiHtQV 8yClCfuRkIgSgmGQiCiZMQ/qinmQ0gcjIVHSYhgkIkpyzIO6Yh6ktMJISJSE GAaJiJIf86CumAcp3TASEiUVhkEiopTAPKgr5kFKQ4yEREmCYZCIKFUwD+qK eZDSEyMhUcIxDBIRpRDmQV0xD1LaYiQkSiCGQSKi1MI8qCvmQUpnjIRECcEw SESUcpgHdcU8SGmOkZAozhgGiYhSEfOgrpgHiRgJieKGYZCIKEUxD+qKeZDI y0hIFBcMg0REqYt5UFfMg0SCkZAophgGiYhSGvOgrpgHiRRGQqIYYRgkIkp1 zIO6Yh4kMmIkJHIcwyARkQaYB3XFPEhkwkhI5CCGQSIiPTAP6op5kMgfIyGR IxgGiYi0wTyoK+ZBooAYCYmixDBIRKQT5kFdMQ8SBcNISBQxhkEiIs0wD+qK eZDIAiMhUQQYBomI9MM8qCvmQSJrjIREYWEYJCLSEvOgrpgHiUJiJCSyiWGQ iEhXzIO6Yh4ksoORkCgkhkEiIo0xD+qKeZDIJkZCIgsMg0REemMe1BXzIJF9 KhIOJFSii8FJLEmn5CYUwyARkfaYB3XFPEgUFomETDFOYUk6JbF5kGGQiEh7 zIO6Yh4kChdavEwxTmFJOiWxeZBhkIhIe8yDumIeJIqApImZuNMvxbAknSKh 7HbcMQwSEaUJ5kFdMQ8SRSCtUgymiNb+4uJiW1tbZWUl/uO1gyNnSToisXmQ YZCISHvMg7piHiSKQFqlmIqKCjT4CwoKXC5XaWkpXjc0NDg1cpakUyNPbB50 aimIiChpMQ/qinmQKALpk2JWV1cRXtDgb2tr83g89+7dKywsRKJxavwsSafG zzxIREQxxTyoK+ZBogikT4qZnp5Ga7+iokJOCMTmAqEGb50aP0vSqfEzDxIR UUwxD+qKeZAoAumTYvr7+9Ha7+3tlbdzc3N429zc7NT4WZJOjZ95kIiIYop5 UFfMg0QRSJ8Ug8CC1v78/Ly8HRwcxFtEG6fGH3FJTk9P9/X1ud3ucD84Pj5V XdWtX0kyDxIRUUwxD+qKeZAoAumTB0tLS/Py8tTdI6X9jzlxavwRl+TExERW Vhb+h/Wp2pqedzcWbnwnX7+SZB4kIqKYYh7UFfMgUQTSJA9i+4CmfnV1tepS Xl7ucrkc3GhEkwevXLliPw/2dI8eOlT57DNXX3whb8/uMv1KknmQiIhiinlQ V8yDRBFIkzw4NTWFpv6dO3fk7erqKt5WVlY6OIk45MHR0cnLl5tefun6C+vy XvnxzSuXG0dGJvQrSeZBIiKKKeZBXTEPEkUgTfJgHMQ0D05Pz5SUdLz1C9e6 5289+8zVQwcrurtHdC1J5kEiIoop5kFdMQ8SRSCsFHPq1KmnP27dunXr169H 93Avf9MvxcQuD7a2Du7cUfrsM9eefy773Y35tbW9epdkWHlwz549/2apsbGR eZCIiIyYB3XFPEgUgbBSzBNPPPFAEJ/73Oeys7PthyD9Ukws8uDAwPjp07XP P3f1hXWu9etzbuXcnpw034ZUv5IMKw8+/vjjweqkqK6uZh4kIiIj5kFdMQ8S RSCCPPi1r31t95rt27evX7/+oYcekkjY1dVlc1T6pRhn8+DUpDs3t/X19beQ BF9Yd/306TpkwzQpyQjy4COPPJIRRFNTE/MgEREZMQ/qinmQKAIR5MGnnnrK 1L21tfVTn/oUeh06dMjmqPRLMQ7mwbq63vc2Fz3/XPazz1zfnlnS0mI1Wv1K MoI8+Nhjj9kcnnmQiIiYB3XFPEgUAUfyIHz1q19Fr2eeecbmqGKdYpaWlgYH B5eXl2M3CZMo8+DU1BRed3eNHDxQ8czTV9Y9f+utn7tKSjrc09PpVpLMg0RE FFPMg7piHiSKgCN5cHx8/LOf/Sx6vfHGGzZHFYejWpgrl8tVXV3d398fhzgT cR6cnJy8efPm8PD41avNr/z45gvrXPifldU4NjaVniXJPEhERDHFPKgr5kGi CESfB6empl5//XW5d8eNGzdsjio+ZzlKkJF2fqzjDBYH458IE4pibGxs+/Yj b7x+/blnbvzoPy7u2lnc2tonBWtnDJioZiUZQR780pe+9Hwgx48fZx4kIiIT 5kFdMQ8SRSCCPPj5z3/+q2u+/OUvy5FB+OY3v2lzPCoPjsdeZ2dn7seVlZWh 8T80NOTshLA4LS0tly9fvmLP9evXs7Ozjxw++/xz2/63//7v//f/lfHEE+/t 3n3yho/NkWBymKhmJeng/UWffPJJ5kEiIjJhHtQV8yBRBBx53sQXvvCFt956 K6xHEMp0czVi//jg5OQkSqC3d+jo0Ypnn8l65ulr//LdDdk3G8bGfncH0XCP MMrxwUQvvfPCyoMPP/zwq4GcPn2aeZCIiEyYB3XFPEgUgQjy4KOPPlq2pq6u bmxszH4MNOXBhtjDHObn55uyBrpg5mtrax2ckP2SdLunXa62N16/9cK63Oef u3b8WOWZM5ciKENdSzKCPMjrB4mIyD7mQV0xDxJFwKn7i0aWYmK9dB6Pp7y8 XIWXioqKnp6excXFWEzLZknW1vRserfguWdvPvvMjR3bS27f/t1Hrl27FtbR Vb1LknmQiIhiinlQV8yDRBHQOA+qCBPT8KKELMnu7pHDhyqffebKuudz3/pF XklJhzxKYiLQ8+jTuSSZB4mIKKaYB3XFPEgUAV3zICJMS0tLHMKLYlGS42OT V640vfZq9ovrXC+9eP3SpYaRkf+Z/pI8D8a/JJkHiYgoppgHdcU8SBQBXfNg /LcGAUtyenqmtLTjFz/PXff8rWeevnpwf0Vn57BpmCTPg/EvyQjy4IMPPviZ IDZt2sQ8SERERsyDumIeJIpAWHnwySefRNv76aefjiy5xDPFxJ9/Sba1De7a WfrM09eefy5707sFNTU9AYsiyfNg/IWVB3/wgx9YPG8C3n77beZBIiIyYh7U FfMgUQTCyoMO0i/FGEtycHD83Nn6F1+88cI6109ey87Jvj0xMRWsKJgHTcLK gw5iHiQiShPMg7piHiSKAPOgU2SJpqbcLlfb6+tzkATXPX/9xInqvr5R66Jg HjRhHiQiophiHtQV8yBRBJgHnYLFKSxseW9zoe9ZEtd37Chtbu63UxTMgybM g0REFFPMg7piHiSKAPOgU7A4T/3o8nPPZv/8Z7lFRR1u97TNokASzMrKYh5U mAeJiCimmAd1xTxIFAHmQadgcX53B9GDFcPD4SW76enpvr4+t9vNkhTMg0RE FFPMg7piHiSKAPOgU1iSTmEeJCKimGIe1BXzIFEEmGKcwpJ0CvMgERHFFPOg rpgHiSLAFOMUlqRTmAeJiCimmAd1xTxIFAGmGKewJJ3CPEhERDHFPKgr5kGi CDDFOIUl6RTmQSIiiinmQV0xDxJFgCnGKSxJpzAPEhFRTDEP6op5kCgCAwmV 6KV3EkvSKbkJleilJyKimGMe1BXzIFEEmGKcwpJ0CvMgERHFFPOgrpgHiSLA FOMUlqRTmAeJiCimmAd1xTxIRERERETWmAd1xTxIRERERETWmAd1xTxIRERE RETWmAd1xTxIRERERETWmAd1xTxIRERERETWmAd1xTxIRERERETWmAd1xTxI RERERETWmAd1xTxIRERERETWmAd1xTxIlDwmJyfv3bun3mLdbGlpycnJOX/+ /M2bN+vr61dXVxM4e/E0Pj5+69atY8eO7dy588CBAxcvXmxvbzcWTm5ublZW 1uzsLF6PjIzgdVlZWeLmN96GhoawyOXl5c6O1u12Y7T5+fk2BygoKMDbqakp Z2eDiIiSEPOgrpgHiZIBkk52dnZGRsby8rJ0WVpaOnjwYMbH7dmzRxKQxpB5 kfUyAjl+/LjH45HBtm/fji5Ignjd2toqfRM643HV0tKCRT516pSzox0YGJBq ZnOAvXv34m1fX5+zs0FEREmIeVBXzINEyQAhSCKPyoOXL1/G2y1btuTn5/f3 95eUlGRmZqLLiRMnEjursXb9+nWV/ioqKpD4bt++LaUBZ8+elcGMeXBwcBDd CwsLEzrjcZUkeTAnJwclPzEx4exsEBFREmIe1BXzIFEyMOXBlZWVjRs34m1V VZUapqmpSYZZWlpK3JzGVk9PjywjAqCpV3V1tfQaGhryfjwPpqEkyYNERJQ+ mAd1xTxIFK7s7OyzZ8+Ojo4is8jVbQUFBaZDJGNjY9euXTt69OjWrVszMzNP njzZ3t6u+t68efPcuXO9vb2HDh3avXt3VlYWGvYSdvDi1q1bU1NTe/fu3bJl y8LCgvqUx+ORYTDp+C1tfF28eBELiDJR54Uq9+7d2+2DXOy1PD5o5wuKPywR ZuPYsWOoEvje8S3fvXtX9cU8o0q43W7UDXz1WMzr16/PzMxgE52Xl3fw4EHU ItQNycJeQx6sra1FNdu2bRvqWE1NjfESS0Adu3DhAsa2Y8cOlImxEorJyUnM FYpo165dV65ckdEa4571AHJ8EMN4fVERrxsbG2/fvo0XmGF8CoupDnl7fV8i vr5Lly6h75EjR8rKymTgjo4ORwubiIicxzyoK+ZBonChMYwm8TvvvIP/aNa+ ++67eIFGvrqyr62tbcOGDZLdpK9oaGiQAdC0xttNmzZJd+Q+41VyaPwHnG53 dzf6YsxYbeO0qPGFsCDFFezOMMbb6VhcPxjyC4o/TBqhzPR1Y94kSal5NtWE /fv3Iz8au/x/7Z17dFTV/faDoIgo3pEqUsUqS6xt1Yq3pf6q9UpV8EJbbSli l9e+tYqXemUkSogKRUEgICDhYjQkElDCLaAQSMTIxUAIBMFKBEICNHkzf7Wr 633WPG/2b3POzJmZzCSZnDyfP7Iy++yzzzl777O/3+fsG+pMTU1NsEkP2uH8 B1LLXBTZyMD09HQTYdGiRSbC/v37ISTdFdXIvagR7PmDUHaOaASq0KjUxYsX O26YfP755y1fCEIIIRJCetCvSA8KES9GbsAnx89Dhw6xd2/y5MmMAAcYP2fP ng0VAE94586dWVlZCJk6dSojUA+ClStXrl27Fum45w86OHjwIH1vM4HOf9TW 1jITNm3aFDVyLHowUgG1PpBpgVC/J7spoeko9MyAT94zKC0txQ2bscFg6dKl 0GWoRRkZGfi5bt26oKUHkQ5SQzXDifwKUV5eHgwt0MqfOB2NPCNQI2/fvp0X nTJlCvPtwIED7LljBCP3okZw60HeEu7h8OHDZohvZWUlIqBYWS7r169Haiju WbNmSQ8KIUR7QXrQr0gPChEvdN3z8/NNCHQf/eS9e/fW1dXNmzcPjjT+NxHo 3kMO8Cf14CeffGIieOtBuNaUD6NHj4Y0aLEna2P27NnDTIhlQGxUPRipgFro 5j3ARflcHOlKoNcYSIXIe16wYIGJMG7cOISgLpmQnJwchHBYLPVgenq63emZ m5uLwDlz5uD/+fPns8rZI0g5HJdp4sRAqLvZzhPUSSP3okYIhtODuCX2YJIJ EyYgEAIw2LREUkFBgTl68OBBilzpQSGESH2kB/2K9KAQ8ULXnV0eBvYA2vOz 4EWXl5d/8cUXECYcKzh27Fgeoh4sLS01kT30ILxrbjwBT3vr1q0t9VQpAJoj dml98803USNH1YNRC6jV4EBf8Nlnny2xYCBvifdsRhSD2bNnswfZhHAbDu79 Rz04Y8YM+0LsgKNY4/Pif/uK/Krw3nvvIQIyOeAanAxlZ1KIGiEYTg86lpph D+DatWtN5IqKCjsCJar0oBBCpD7Sg35FelCIeKHr7tiD2/bed+3aRW/cgUMP 2oIlkh7EVdhPlJGR4RA4vmTixIkOEWRTXV0NAcgmK6oe9CigVgZqyF0ZDFxC lvds6/05c+YYJUUgJx160J4tCHbu3BkITc3D/2ben5vRo0cjwqpVq9yKkl20 1HRRIwTD6UHHfihz585FYHFxcWNjI6/u6PwtKCiQHhRCiHaB9KBfkR4UIl7o ujvcWva8bNiwoa6uLjMzE//DIc/NzYUnvGPHjo0bN7r1IMLN6WH14O7du+nV 48Q2Xx6zdeAoR454dMPOJs65i6oHIxVQCz9BGCiUxowZUxEODvh092nGogft 0aSgvLzciDUmuGjRIvcVbfmWlZVlp4BMMylEjRCMRw8GmySqY/N6zqyUHhRC iNRHetCvSA8KES/0tG1HHSJu9OjRCPzhhx+gONhHY28Vwa6Wt956iz9j0YPf ffcdp1ZNnjzZnpDlb0pKSpgPXDXFpqqqiqNJeSiqHoxUQK31KP8LNRRu3i7H +vr6BQsWQK9B9QebqwfHjh1rTw8sLCxEIFRzsKk/1J6QGAwNAYXipviqrq5m RbW3s0TeGrkXNUIwTj2IAgqEBs2aozBAeCmkB4UQol0gPehXpAeFiBe67uPG jaN7D4eci2yw+4+KJj09vba2lvHh8FPZjRkzhiFuPWhG07EfED+nTp2KnxMm TKirq2s4ktZ+4FYET2dWB/3yyy+NEikvL6dwgAzk1oRR9WCkAmqTh+IEUmg0 s1cId17IzMxkPWmeHgQrVqzgUVQzbuLAVWtYDyFCTX8ccoP1cOnSpcFQtjBL 8/PzKSoRgYOTKfeiRgjGqQfNhhRcARWKmAvgGD2I9Dds2GCvLrt161aEQJkm qyyEEEI0G+lBvyI9KES8mK0B4Mzn5ORwsRe44hygCN+VS1lCfeTl5c2aNcve kY3df249CDjKFOlMmzaN40sj4TjRZ0DXUIZQziB7kZP8icxhb1owBj0YqYDa BK7NQpX60UcfmTvkwpvBBPRgILS/w7x58ygGs7KyqN3w10xiRSXMzs5mPZww YYL5pFBVVZWenh4ILRqDjGINtOVe1Ahx6UHcktlgYsyYMfYWhNSDTAGFbk5n thjNK4QQog2RHvQr0oNCxAt9VLjHxqGF480uDwK31njOPArZQg3I+WvsLdq5 c6edbGlpqVGOZvHJsMBLb+1nbl327t2bn59v62hoBMgoe7QnO6oo8ThG16x8 ErWA2gRoPW6+QEaPHr18+XJzlIdspc+FN+0VRzkcFH+DTdLpww8/5CBMZtHH H3/MzlOC5n3hwoV2NkI5fvfdd/Zdbd68mZsMBkKSGcqLmjHGCFz/hxsashTM joqE0wPN6F+Ym2XLlkFdjho1CsWEZ6FCZATqeihQczr1ZlFRUbOzXQghRLKQ HvQr0oNCxIvpysG7s3v3bjMu1KahoWHPnj2QgbZ/HpX6+vqampq4TvExyEMI Qyg+5GRco2RjKaC24uDBgzt37qyurk7iuF88IGqaGYnqoLGxkeuyeuQDah0y ysMWRI0QC9u2bYOMtSdR4t4ohNtknR8hhBBxIT3oV6QHhYiXsNvbidRBBZSa cP/E6dOnmzVwiouL2fPYcVZMEkKI9ov0oF+RHhQiXiQ3UhwVUGrCvewDoXmd 2dnZHDUN7HGzQgghUhbpQb8iPShEvHD2k+8n8bVfVEApS1lZGSfSkokTJ65a taqtb0oIIURMSA/6FelBIYQQrUZjY+OBAwd27dplttoUQgjRLpAe9CvSg0II IYQQQghvpAf9ivSgEEIIIYQQwhvpQb8iPSiEEEIIIYTwRnrQr0gPCiGEEEII IbyRHvQr0oNCCCGEEEIIb6QH/Yr0oBBCCCGEEMIb6UG/Ij0ohBBCCCGE8EZ6 0K9IDwohhBBCCCG8kR70K9KDQgghhBBCCG+kB/2K9KAQQgghhBDCG+lBvyI9 KIQQQgghhPBGetCvSA8KIYQQQgghvJEe9CvSg0IIIYQQQghvpAf9ivSgEEII IYQQwhvpQb8iPSiEEEIIIYTwRnrQr0gPpjLbhRBCCCGEaAHi9UulB/2K9GAq 09bthBBCCCGE8Cfx+qXSg35FelAIIYQQQgjhjfSgX5EeFEIIIYQQQngjPehX pAeFEEIIIYQQ3kgP+hXpQSGEEEIIIYQ30oN+RXpQCCGEEEII4Y30oF+RHhRC CCGEEEJ4Iz3oV6QHhRBCCCGEEN5ID/oV6UEhhBBCCCGEN9KDfkV6UAghhBBC COGN9KBfkR4UQgghhBBCeCM96FekB4UQQgghhBDeSA/6FenBFKeiomJeiKVL l7b1vQghhBAtCHySuXPn/u1vfxsyZMgDDzzw97//vaioqK1vSgjx/5Ee9CvS gynOO++8kxZiwIABbXsnI0eO/D8hdu7c2bZ3IoQQwn+sX7++V69eaS6uuuqq /fv3NyPBNWvW0GxNmTIl6XdrI/soOgjSg35FejDFSR092Lt3b97J2rVr2/ZO hBBC+IyKioqwYpBcf/31hw4dijfNyZMn8/Tbb7+9Je7ZIPsoOgjSg35FejDF iUUPVlVVbdiwob6+3iOdhoaG7du342+z7ySqvausrMQh79sQQgghHHz//fd9 +/alienevftLL730xRdfLFy4cNiwYUYSDho0KN5kY9SDso9CxIj0oF+RHkxx 3HqwtLS0cwiEFBYWXnDBBYzQrVu3t956y5x41113MVpubu5TTz3VtWtXxDnx xBNhUjdu3GiiFRQUMNqVV14ZKfDll18+66yzjFE+7bTT8DMrK4uR9+7de//9 9/fo0YNHTzjhhDvvvBMXbY3cEUII0f7JyMgwJiY7O9s+NHz4cHOopKQkGJvZ qqmpgZ2CyeOJxxxzDH7+9Kc/ZeQY7SOSYjQkHilQ9lF0KKQH/Yr0YIrj1oPr 1q1jSK9evU455RTHoJpFixYx2sCBAxkCC+iIc/755+/bt4/R8vPzGXjZZZeZ izoCn3jiiTQX1J6NjY1IzX20U6dOCxcubNWcEkII0T654ooraDtguRyHDh06 dPLJJ/PoCy+8EIzNbEGIuQ3TSSedxMgx2kckxUAkbi7kCJR9FB0K6UG/Ij2Y 4njowbTQp8gRI0YMHjz46KOPZsi9997LaMbegf79+7/44ov33HOPCfnzn//M aLEY1oKCApx+/PHHM3DYsGH4uXr1ahwqKipiYN++fYuLizdv3jx06FDHnQgh hBAenH766TQcgUDAffS6667j0SFDhgRjM1uHDx+GnbrtttsYeN555+Fneno6 I8doH2PRg7KPokMhPehXpAdTHG89uGLFCgY++uijDLn00ksZYuxd7969q6ur GfjII48w0AybicWwkrDzIz744AOTYFVVVTBkhUtCbNmypaUyRQghhF9oaGgw Ru3jjz92RzAG7vrrrw/GY7YizR+M0T7GogeJ7KPoIEgP+hXpwRTHQw/26NGj sbGRgbm5uQzs168fQ4y9++tf/2pSKy0tZWCXLl1Q9MGE9WBNTU337t0ZftRR R0GNwnDDoJsbE0IIITyAvTBDXKZPn+6O8Ic//IFHb7nllmBS9aC3fUxQD8o+ Cv8hPehXpAdTHA89eNZZZ5loy5cvj6QHx40bZ6Lt27cvrYmtW7cGE9aDYNKk ST179kw7kosuuogT/4UQQghvjH15+umn3UcvueQSHh06dGgwqXrQ2z4mqAeD so/Cd0gP+hXpwRQncT345JNPmmge/YPnnXeeiWZGucSiB0FdXd277747ePBg e/eoe+65J9mZIYQQwofccccdRisdPnzYPvTNN98cc8wxPDp+/PhgPGYrqh70 to9G+s2YMcNEM0vExKIHg7KPwl9ID/oV6cEUJ3E9ePbZZ//www8MfOyxxxjY v39/hpgJ72lNX0SD1uCcsHowLy/PBFZUVBSH2LNnTzA07GfKlCmM1rVr10S2 cxJCCNFByM7ONpbo4YcfNuEQUxdffDHDO3fuvHv37mA8ZsvoQTOznsRoH2+4 4QaG3H///QzZsmVLp06dPPSg7KPwN9KDfkV6MMVJXA+mhSazv/jii0OGDDEh w4cPZzRYWxhZYxlHjBhx7733mmi2YTVGGam9+eabZWVlCHz++efN7X311Vcw 1i+//DJDLr/88tbJIiGEEO0aaKXf/va3ts2CMYKdsvvUJk2axMixm62cnBwT /vjjj5vJiTHax2effZYhuBxOee655/r06WOi2XpQ9lF0EKQH/Yr0YIqTuB48 88wz047k3HPPNV9EweDBgx0RYGHdhtWsvUa4v1JFRcWxxx6bFo4pU6a0RgYJ IYRo/8Abufnmm8NaEzBq1Cg7coxma8eOHbaFcu8/6G0fi4uLuVW9oVOnTuYU Ww/KPooOgvSgX5EeTHEmTJhA83HllVcypKSkxJg/E23FihUMvPDCCxli7B0U JUxVly5d8H/37t0Rvn79evsSBw8eHDRokJmgcdVVV82fP5//298w9+/f/8QT T5hdopAswzdt2nTttdear7VHHXUU7qGgoKBFs0UIIYTPqK2tHT58+HHHHWdL J+ivzMxMR8wYzVYwNNnw6quvpgXs2bMnA2O3j3l5eWYsaLdu3RD/9ttv589P PvnERJN9FB0E6UG/Ij3oV4y94xibw4cPb9682TFP3+bQoUPFxcXcI8mb+vp6 zrW3QUhZWdm6detgphO8cyGEEB0WaKvPP/88JycnNze3pKTEY55d7GYrGJKQ ZqOHeO1jZWUl7iQW6yb7KPyN9KBfkR70Kw57J4QQQoig7KMQzUV60K9ID/oV 2TshhBDCjeyjEM1DetCvSA/6FZi5x0OsXr26re9FCCGESBVkH4VoHtKDfkV6 UAghhBBCCOGN9KBfkR4UQgghhBBCeCM96FekB4UQQgghhBDeSA/6FelBIYQQ QgghhDfSg35FelAIIYQQQgjhjfSgX5EeFEIIIYQQQngjPehXpAeFEEIIIYQQ 3kgP+hXpQSGEEEIIIYQ30oN+RXpQCDfff/99spLC+1VfXx9j5H379tXU1CTr 0qkPcqahoaGFEj98+LB3hCSWsk1tbe3BgwdbIuXmEVcNFO2LFn2DhAiL7GN7 JFlthfSgX5EeTHGqq6vXrFlTXl7e1jfSgXjiiSfS0tJeeeWVxJOC/Zo2bdrn n38eS+Rdu3YFAoFRo0bt3bs38UunDj/88APqcE5OzuzZs4uKinbv3m0OlZaW Tp48GZU8iZeDDHzjjTcuvfRSFOIjjzwSKVqySnn9+vU9evQYNGgQf1ZWVp5w wgmnnnoqnjrBlJNCXDWwnVJVVYUK9tVXXzXvdHiY69aty8/Pnzlz5qeffrph w4awNhH1tqCg4IMPPpg7d+7KlSs9vjZs3bp1TThwn3Y0XGXz5s2LFy/GdRcs WID7dzvGUeO0xBvUthw8eHD16tW5ubnMapTIzp072/qmxP8i+5gUamtrw7YS 5LvvvrMjHzhwYNmyZXgdZs2aVVhYGMkh3LZtG6JlZ2d/+OGHq1at+uc//2kf TVZbIT3oV6QHU5wvv/wSbSAsY5tcHYaYTkibXD2JbNy48dVXX4U7F0vkm266 Cfbud7/7XYIXhdnKzMxEyxzjK1ZRUREIAcOX4KVTB1goqLOABQx6WVkZjzY2 Nubl5Y0ePTqJH5whPFF8Xbp0+fWvfz116tRI0ZJVypASSAfX4k/4Nmkhvvnm mwRTTpx4a2C7A88F/+e1115DvfrHP/7RjBTwrr311luBI5kzZ45Ddm3atMkR Z/z48ZEkIfyxQDggbUwcpA/vzhHh/ffft3uWY4nTEm9QG7J9+3ZHc0Hmz5/f jNRa2n6h7iH9lBoM0DxkH1ufqqqqsK0E+eKLL0zMrVu3ul+KTz75xO7sQzuw ePFiR5z09HRjaoPJayukB/2K9GCK07Z6sLS0FFfPzc1tk6snETh4MGEPP/xw LJG//fbbrKysBNtMvFmTQsT+fqG53rx5Mxr/RK6bUsDiwyShCs2bNw/PtWXL FuMqm6+UeGq4uO+8886hQ4eSctF7770XZT1z5kzvaEkp5aBLD4L8/HyY5gST TZxm1MD2BWrXtGnTjPPTDD0IwZWRkYFzp0+fXl5eDg9txYoVVJdLliwx0Wpr a9988022wxs2bIDkHzt2LH5CC4RNFnmOo4WFhSuOpLKy0sSBO4c4uDrqD65b XFxMl+/jjz+OK06wBd6gtmLPnj1jxozBAyJ7kbfr169fuXIliobla5dIjLS0 /WI1SJGRAIkg+9j6HDhwoMgFWolRo0ahUqHqMhpauddffx0hkydPXr16NdoQ NCxso9AsmNTQLvE1wVEUEKJBaDNk//79JlpS2grpQb8iPZjieOtBvN1Riy+R 8u2YejApwGdD1m3bti1ShKTP+kligpGS2rRpU0FBAWpd2KPLly9fu3atHQKD RWfbnIKUJ0yYgECYNhMN9gshs2fPjpRyXFx44YUo61YbY+bWgylC1BoYDHll jpBYmpRgDJUtltqYYI3NzMwMhPrpUJeapwfZ6wfvyH5kVGN6XyaEOTlx4kQT bceOHfS1tm/f7k6WHY7e2cg4dumUlZUhBA9l3oJY4pDkvkFtAhxUlCAL1NHj BieZuQ2FGFeasdivROxjx9SDSUH2MSwQvMiWN99807TM8+fPR8i4cePs0QhL ly5FINS0CUElR4jdw4vbQCOGQMdFE28rpAf9ivRgiuPQgzCaMJezZs2qqqrC 3zfeeGPUqFHvvfeePTgNzQ7ilJeXFxcX45/XXnsNdhYNhT1uHP/jUHZ2tn0t NBQI/Oijj8yF+LX29ddfx/9wh+zIu3btcgxxTwoQC1dfffWqVasyMjKuvPLK nj173nTTTfxWNm3atF/+8pennXYawnHUcWJeXt5jjz3Wr1+/Pn363HHHHXPn zmU4avg111xz/vnnw96dccYZSPy2226zr1VUVDRy5MgBAwaceOKJyDGET506 FeHMB4TccMMNt956q230p0yZgsBnnnkm0lNwpoPdXBugU1CayFgUHITSypUr Dxw4wEP19fXIZ5Qmf1ZUVIyPgP1xDzUBzw57gYLGncNSOJz8WEpq2bJlSBbS BvUEtQVJ4Rk3btzoiDZjxgz6V25TsmTJEvpydmBhYSFcJoc9ys/PR0z8tQPZ C4BH9r5PgNz74x//eM455/Tv3//RRx+1x+DBf0bBde3aFWWNqoL/PcaL2qUc tCoe1AEKF9XsF7/4xauvvurITPxEBFSh008//ZJLLnn77bfXrFnj0IPXXnvt wIED7bPwXEjquuuuQ33+n//5nxdeeKGurs79XA888EDv3r0vvfTSESNGuAe5 4Q7xLvTq1euKK6546aWX0DJ45FKkGsjGASWL/MfNIw6/t8PDQTHhrRk9enR6 ejpKH1nhPhHtzOLFi9999106Lah4Dtcdz4UIqMOIgOJYtGjR7t27cWJOTo6J g2uh2qMu4UKIgyIIW+5R6y1Oh1cJN4nqrBl6cMuWLbjDDRs22IHIHDpgJoS6 zDFnhyM5kS2ONGFPmTke162trcV7gWy036OamhqqHq6YEUscm9jfoNQE2Yv7 h0VzdzwhB5jbxg4mYr+SYh8hUfE/ywLvEf4vKSlhtBayjET20U/20QHO4tAC e0IlvzmgpbJj4hHQru7Zs8eEwBt0Sz+KbtRVx4USbCukB/2K9GCK49CDcBJo NGnpONiJmMUKFixYgJ/02ew49ndXNCYIQYNvX6uyshKBMCvBkD1FEwrnMBAa hY7/kaCJSdttXzRZPP300zBM119/Pf6efPLJnIp10kknPfnkk/ine/fuxx57 LAPRJpuzYDsYCPuIOPyfHh3cRVhAJnXcccedffbZP//5z+1rwatnfKQM3x7h cLbx07iXnC4xbNgw/oTzefzxx3fp0sUereEATiZHbjjC0QJz/CQaeXaTsd+B 3xtxq8xtRoajPu5IOG4E2OXIIWTIAVaJQGj2k7FHMZZUVlaWqUimwiBNR/cH agXsYCA0ncc2efxciQpmD02JxPvvv+8weQB2PxDDLKEXX3yRhdWjRw/WhE6d OhnfBkdRvkcddRTCoZvwP1y1SEk5SpmV4cYbb8RflK+pZsOHD7fPwk+Go5od c8wx+AeukUMP4ie8MvMTHg5UHs869dRT+Q9qnT3ZnzdDl4yXRkxbgPC6OIQT mcjRRx/trmCGSDWQjYMpYlQ2DmLk0ET2PZlq9tlnnzlORCVk+8NBTQB+iIkD U2JG95k49GdsYc5P2bwNvg74i9fKvs9Y6q0ZOdZsPegGtZrjrIyA5VsJHG4k OyUdQzeDTZINXiva6rKyMrwaUAqxrICxbds2nIgm1+O7vUecGN+glIXdsuZd dkDbhCaOPxOxX0mxjxAIfBcCIe2P/+mKt5xlJLKPPraP/BJldw7ykXEhNK24 Q7zjXPAKbYvjXFS/QGhysTl33759/JDlnm6ZYFshPehXpAdTnLB6kKYK73sw 9PmLLadxYGjvAqFJW/y29u2333L+C1wdxolq70ik8TZfffUVL5H0RTNogzp3 7oybD4b8H2OPXnjhhfr6erhbzzzzDH7279+fp2zZsgUSoGvXrrDRwVATClOF COecc45JNux4GF4LjBw50v7U5rB38OdpLjmBZeDAgTzF4ynobLhnOtDVRAHR WFRXV9O48Oufw945wFEWdFFREUPw8rKLBw07rQBulQma2eixlBSsLa0wDBY/ zqN6MGX3CgMwtRyFYkxeXGKQJg+G2+Eh484DoR4lj3Ozs7PpluCucM+4E+Qz QqDLzGyLYNN4UcfSam7C6kFA9wPPgpyk3jRfjyFqEIKalp+fjzioiq+88grP iqQHkUUXX3wxQn7zm9/gNQyG3mh4XAh5/vnnGQdVnafw8zuciscffxwh/fr1 YwQUB2o4IvCLLtLEuYhw8803R3q6SDXQNA6ozMapMAv3sR7iLVu1ahU9H+PV mBOhg2AykPlmFJ95d/ACsgLDy0Ic1FjWjYClBymjkDIFIOptXl5eIKRD7VFY cbUwietB5AC8LNwbv/CjtWTrajIHt+c4hTUZ7p8jnMOx4HyyyTXAB/a4ATw7 XysI8+bFieUNSmVmz54dCNffSlDfmI1sNxKxX0m0j+7xoi1nGYnso1/tIyLD 9ASO/FLKxuedd97hqFEb0x9N6urqOJkaeYsMhIUyU57d10qwrZAe9CvSgylO JD1oT49CAxiwJrzQ3qG5tgeco0WlG8afCepBtJA4hHtL+nQV2qChQ4eakPff fx8hvXv3NiHwzzuFYCOPm0GzaRtlhPTq1Qtnma0NPOzdfffd57gHh70LNimR n/zkJ3AX8c+AAQO8JyPwm6R7FArKEeEwECYE5gNmgosZets72kr72ybdYBga uxQ4ZQ/OlcmKqCVFDzZgrfECPvvss0jWxJg8ePJxiUHztdY9ggXuN+/Bozni QjEwjnbgQw895PgYnogevPvuu00E3AkkGAKNj3rHHXfgJ5xAO50777zTQw9C NbD22vmD4rjtttvMXhh8LuOLBkO+wQUXXIBACkCUIP7/8Y9/bBa9xL2h9D2G OUWqgWwcHAPhkBocY8fKlnScjB/IE+3eQFQtXsW4UhB9gSNnhgab/HyjB/kK 2INdcXX61fZ37LhamMT1IAdWETy4nW9cqME9to0+lemxMnAGEN0t1HNofLju DHFXewP7TOHCuQcSxxgnljcolWF3m3sMnoEeOLI3mAw9mBT76NaDLWcZieyj X+0jpwaj0bDHIRi5Ggh9elq1ahVaVzMG4+uvv7ZTKC8vtweMURuG3V0iwbZC etCvSA+mOGH1IJpEu72lZ2KGptPeOeZnoZT5jSvx76stCm2QPTaVC/ibSQ2E A/Acc5fQoubk5KDZzMzMPPPMM407HfS0dw59EQxn78CQIUP4sbRbt270SSJx 6NAhNrbuOT4lJSV2246Cs82Qh71DJgRC3wntmRHsakHgSgvOzvMYJ+kGLitO cSzIyd6fSENKkPOc6RC7GETFo0cHw+02vmiImJqZLeKmb9++yH/HEgTw5BF4 1113mZBE9KCjGwjyEIGm/kOR2ZWKZGVleehBemsPPPCAx53wuZCN0yy4fyJ7 AUC/fv3ws0+fPq+88gpK2b0OjI1HDWTj4JgbSNBEwCtes2ZNUVERvB0OrzI+ G090dN9w0gqnOqJM6ebZXQnBpoFMRg8yWaRvV1r61Y4v3rGTuB7EGw3PDQXN NWrwDponpVJwtxJsLe1phgS5gdMR31ZtnD3k+JLgOPr66697DFqLGieWNyiV YXtid/TbmNrFEXqJ68Gk2MfWX09G9tGX9hHVj32m9jYTwaYKTA/Q/hhI8WtX Repc3ueyZcvMJG5kF8el2CTYVkgP+hXpwRQnrB50fJT+/vvv3XrQ7fKhYQw0 Db1IcT1oj8IKa+/OPvts295NnjyZo/IcxGLv3CO+wto7eLldunRB+JNPPun9 CMbeuRtbNOloqzmrhaDRNtYzkr2DAnrttdfgDjk0jhn4FBb33taRYL+Do/OC 3TocwRiWhQsX8kJh9Z0DZAWrH/RR2BtDYCQngaCs00KT5hzhX3/9NcLPP/98 E5KIHnRUhj/96U9GD5obcDwsfAwPPThixAjv0VNMNhLIEEaDqvrVr35lwnv2 7JmRkRGpL8mjBrLOOLrwgqFBTZRCDhx60HEi3RLqQVyLpzi0KmfoUA+aOGFp 9iYdyZ0/aHww9p5AIARCQswRc/369QGXlxgJUyLuiYTwABGOF9xjxFoscaK+ QSkOhw2jVQl7tLq6mk/HTr3E9WBS7GNb6UHZRxsf2Ec2Jo7OwWDTfOGAayFW jiNFVvApkKWcrG3LyYaGho8++igQbmxDgm2F9KBfkR5McZqtBx0j2x1f72nv HKtd0XNrd3qwpKSkU6dO3bp1Gzp06KxZs5YvXw4HlaIgifYObTVt6BlnnBF1 6yU2zpGiofXetGnTokWLzAwjTlYKa+/279/PaI7BIcGmWeHwSHe4iGs1g7Cz OTgV3f1pkXCmGLKIUx7y8/M9TB6elx9L4ZZE2vbILJzoYaaR826h9+mnnzrk WAvpQcDho46UWbUi6UG4lGmuRWkcnH766YiD+rDShWPXDLgNb7/99i233MKq +Oyzz0ZKM1INDCvr+OKj4uFJ8e5s374dMpCdubHrQcDRSmYQGqGWoR40rdDG jRvdlTaWXuawJFEPEn5a54IYuGf2HDncV3YQuLcghBuGF9k9Xo4do44FKOgH Bjx3UoglTjC2NyiVYS9MVlZW2KPMBNMvloj9SqJ9bBd6UPYxxe0jPHCuVev+ TMePUe465vi+hHujW+i4Cj3DQNNXFEOCbYX0oF+RHkxxmq0H7d2ygk1j4GHy 2GKgeWGDYDs59HDc9tSe2dTSNMPewczh/7/85S/mKFo/rqLmsHcPPvig97WI 297BfT322GPht3OZR3uWWVjoS7v3Jvu/IcxPtMZcuIyD09z2Dl4lI4RdjAJP FwiNh7GtAP6HsfAeT2hjLIs9uIjzCxzDkg3G2MEWm7kSkUwenpELisLHdoxf soHIYv30uNVbb70Vme8YojNy5EiHMmo5PXjDDTfYPwkqnoceXLt2LX5edNFF drnj7YOPYbqWbr75ZsSxN86IyuLFi+l6RWq9I9XAsLKOi/nbGYuS4pbHcelB LsZid/OZCmzGi3LRg7KyMjsdVFfUw2ZbombrQTw4Krl7UzD6+ea5uE4Fl+Mg eC5ulud24fD4AddAMvp1cIPtF3Pz5s3MZI+BsrHEIbG8QamMmaXlWHk4GGqO 6POb1VwTsV9JtI+sJ1GbmiQi++gz+xhsGiULgelWZ8gNCk/HtyB+mjALwlAP ovY67ofdiGhAHJ9hE2wrpAf9ivRgitNsPRgIjb1h4eIsDoYx6aBp4qgM2Di2 IXD8+AXbtndmjrOjpaqrq4Nphs/j+O6UOM2wdzzlnnvuMUefeuopx3gY3Cp+ Qil4X4s47B3y57LLLkMI3F3kJCeR2QtruOG8eEcbjjzkgthm0Jexdxx24rZ3 3AUAeirsS4rH5zdtsy8AipUrY+BWGRK1pGhKHJNEuIZG2G/1xtiZBRjt6fMO k4fb5rgaeNTuJbJtuApHpN4BgtxLC62cYBwwWNJu3boh0P7a33J6kKecddZZ RiUVFhZ27tzZQw+igWWv4nPPPWcicK7N/fffz59cDb5fv35m7n9lZWXfvn2P O+44rrsC8YhEEN9k77p163DKueeeG+npwtbAYARZx0C8+Ewf1ZJCL149yKmC gdBgPCgg+NgmHaMHOQ8O7r15WERjW2T7h3G1MGH1oDsFd8iaNWt4oj3nkd0K cKKMB0iHDe8a7xm5xCW8MjMz3bUaTxEIjS81Q7yQn1TK9qYbKGJ2O3IN/7DE EscQyxuU4rCOIedR7U0Z7d69m80LzJBpcxKxX0m0jzzLfilazjIS2Uc/2cdg KPe4kpL7yxLhLEg0NcaioWnlhwizea5Z2gu62FwCGnDmzJkInD59uiPNBNsK 6UG/Ij2Y4jRbD3JIRkZGxsSJE/mF2V5EPWjtAobUOKqBp9j2jt+RAqEdr+wW nmN7Aq7v/InTDHsH36xTp05poZ2Sfv/73/fv3x//n3LKKba9wz+Mc9555z30 0EORrkUc9o57Ctx44438CRvE9CONFQk25Y97nC0nswdC6zSiMachQP7zK7TD 3pm5AxBTM47EeKownSxc+DwI5xQwhJjd3KKWFD1bxzZqXDzNPZeHe4TZxo7Y i2zb4TSOrFqZR+LYKZ6zLcxS4WGB1bv99tvTQvtkoTiuueYazlhxbHzccnoQ BTRgwADewK233or/UanOOeccDz0YDFVgilZIvEGDBv3oRz9KC21xaPbngtvD LsJTTz317rvvRlU/6aST7M/su3bt4rZil19+Oe5z2LBh/LyPXI30dJFqYFhZ R5+HTQRef9YitgZx6cFg0ywwG0f/IAqR/cWopTg0adIkVmCHAxlXCxNWD7pT cIegOeVILXhTuAFkF0eKBlyTGblyDptHsxG5YyN7An+MCwDiufAPsohf+NGE mtG/cNXMJKlMF5s2bYoxjk0sb1CKA1fE5DOaQWQ1O5JYQI5t1Jptv5JoH83n DkTgeMWWs4xE9tFP9tHcAOp5pKGbCDcrw6DlRJNC1Yyr2JOReZ+B0JjqnJwc 0+wgsr2PLUmwrZAe9CvSgykOFxw2tqyuri7g2gyLc+0d++3iZUfjw8YQLQlS cIyZZy+A2cIVzT531HJ8TYIHyI9RtvPJj6W4jaRPneD6G/aXK9wAQgYOHGhH 69OnDwLNKIh58+bRzU4LLbWBZhP2Me3ItShhNfjpEqYq0rXIyy+/nBZa7zEY yn+IDnjjXGeAcIMAe01LBzU1NXBg0CA7vgfCGV6yZInZGBeZj3IxtoOrfplR HHD5AhGwp4HDD8Hz0lehj20b4qglBdsRcHVA2OtG2sBeu40doclzrL9dUFAQ 6REcX1xRe1FLo1Yn1P+HH36YC3KCn/3sZygsRxw6PFHnsNilHIxQGR588EHH CFXk7X333Wd2gh48eDC76m666SYTB54VKqGdDuwvBCNVYdeuXa+66irHbBcU 6PDhw1mrKSfhZdkeAtwVnEX9mxbae8LsLxOWSDWQnoB7GQSUNf0HOmB4KIoa I6vDnsj1CuxnQQ3HT+QYbg+1CLdNt82W/6gqiMBWhZcrLCx0jHSKq4XhAsuO GuVOIWyauJk5c+bw9SHIB/dW2shPjnQlqKv2qvgO8CzQxezaI9ALdvNrtg0K C13TWOLYxPgGpTjw+dHIGKtE8FZWVlY6YjbbfiXRPiK3Z8+ezUToXbecZSSy j36yj4DDzr0HAOClwG2wmrEmoxK6r4InNd9PCISkezP6YMJthfSgX5Ee9B/2 l3yULxxjj9HyKH1E8Nj0Khj64o3WzJGIY6h/KoB2Hm6Dx13hQfbu3ev9sA6Q mnu6QdhABxxjFmnWDwwWVHwS96iCyUA5hp3OkMSS4uSLSEeRId7bTkWCjpZ7 3pAHMHOOfQ1aEzxmeXl5vIt1o4zgw3hP4YeH5liPxQY5jLxyr1EZFu8aGBZU S7gZza4tyBDoR0e2LFq0CLcxd+5cd3w4JB4L3CVeb90pREoThYLWAz5tWF/O gFcMcVBGsbQhqCRIs6KiIq4Gp3k04w1KZdDIIOvwiiGrvWt7M+xX0u0jKg8i mMY8BS0jkX1sp/aR4BG2b9++Y8cO74xF84u3BjEjjYBNvK2QHvQr0oP+I9KS 8qKVmTdvXmZmZrNXTewg1NbWjh07dubMmS20g3NHppVrIIcHv/322/Th4fxA PXHQo3vNFpEs9AbFhexjiiD72Pokpa2QHvQr0oP+Q/YuRcCbtXDhQjS/rbkA Xfuiurp6/Pjx8+fPb6eL5Kc4rVwDDxw4YDYxhCrkEgeB0GBRj6VlRSLoDYoX 2ccUQfaxlUlWWyE96FekB/1HeXn5kiVLPGZzi9bkyy+/bKG1BXzA119/7bGl r0gKrVkDa2tr8/LyOD8FehBKsLCwUFKl5dAbFC+yjymF7GOrkay2QnrQr0gP CiGESC6NIdr6LoQQQiQT6UG/Ij0ohBBCCCGE8EZ60K9IDwohhBBCCCG8kR70 K9KDQgghhBBCCG+kB/2K9KAQQgghhBDCG+lBvyI9KIQQQgghhPBGetCvSA8K IYQQQgghvJEe9CvSg0IIIYQQQghvpAf9ivSgEEIIIYQQwhvpQb8iPSiEEEII IYTwRnrQr0gPCiGEEEIIIbyRHvQr0oNCCCGEEEIIb6QH/Yr0oBBCCCGEEMIb 6UG/0tDQID0ohBBCCCGE8ACqob6+vq21i0g+jY2NKFn8besqJoQQQgghhEhF jGRoa+0iks+///1vDRkVQgghhBBCRIKDRSEc2lq7iBaBel+SUAghhBBCCOEA MkGdg/7mv//9L0sZfzVwVAghhBBCCBEMDROlTGhoaIBkaGvVIloQlC9K/F// +heXlxFCCCGEEEJ0ZLiADAQCVKHEYAfhP//5D1Qhy10IIYQQQgjRYYEehDSA QGhrjSKEEEIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBBCCCGEEEKI1OL/AUki phI= "], {{0, 209.}, {600., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{600., 209.}, PlotRange->{{0, 600.}, {0, 209.}}]], "Output", CellChangeTimes->{3.7633835069212637`*^9, 3.763417325790848*^9, 3.76374399996024*^9, 3.7637446347976637`*^9, 3.7637449117733173`*^9, 3.763810069155517*^9, 3.763815816565401*^9, 3.7639148429281282`*^9, 3.763989552804554*^9, 3.764005775858334*^9, 3.764081111474266*^9, 3.764589520596929*^9, 3.798186204889843*^9}, CellLabel->"Out[25]="] }, Open ]], Cell["\<\ The transformer architecture then processes the vectors using 12 structurally \ identical self-attention blocks stacked in a chain:\ \>", "Text", CellChangeTimes->{{3.763326192005732*^9, 3.763326223239588*^9}, 3.763326264822359*^9, {3.763386690233284*^9, 3.763386690704877*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetExtract", "[", RowBox[{"net", ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{{3.763326234575265*^9, 3.763326236830769*^9}, { 3.763417586477314*^9, 3.7634175874749527`*^9}}, CellLabel->"In[26]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzt3Ql0XGd9x32HpCQhlO1ASCFJWyjlkMLpAVoK9LS8L30pDRwKISFtKScH UnYSO4tDyAKR90VeZMuLHNvybsvyvlurbUmWZEnWYsmyLNmSbcnaF488IpA4 zvuv/se3N7NpNPNId+7193Nkn5lnnrlz5z930f3pznP/8vFJ3/3pOyZMmPDs HfLfd3/0/P/7zDM/euHh98md70189hc/m/iTHz84cfJPfvaTZ774+K3SeEr+ /fPtEyb87+23AAAAAAAAAAAAAAAAAAAAAMBDXn/99f7+/o6OjsuXL7cBAAAA AAAASDCXL1/u7Ozs7+9//fXXnU4TMR6uX78uH3d7e3tPT8+VK1euAgAAAAAA AEhIV65c6enpaW9v7+vru379utPJIsaQfL7d3d29vb3yufv9/iEAAAAAAAAA Cczv91+9erW3t7e7u5vw1sP6+vr6+/vls3Z6iQMAAAAAAAAQratXr/YPczpf xJh4/fXXe3p6iG0BAAAAAAAA17l69WpPTw9j3nrSwMCAz+djkAQAAAAAAADA dfx+v8/nGxgYcDplhHk6vK3TixgAAAAAAACAWOiAt06njDCPEW4BAAAAAAAA 99LRbp1OGWHelStXSG4BAAAAAAAAl7p69eqVK1ecThlhHsktAAAAAAAA4F4k t15FcgsAAAAAAAC4F8mtV5HcAgAAAAAAAO5FcutVJLcAAAAAAACAe5HcehXJ LQAAAAAAAOBeJLdeRXILAAAAAAAAuBfJrVeR3AIAAAAAAADuRXLrVSS3AAAA AAAAgHuR3HoVyS0AAAAAAADgXiS3rtbX19dj8+abb1oPkdwCAAAAAAAA7hWQ 3EqLPQmUu07kkYjKH//4xyferrm52XrUFclt7cWS3eXLovwZvOob/zm8Ouir XD07yp9LZUfHfw4BAAAAAADgSQHJ7YYNG+xJ4Nq1a52IJBGVP/zhD/IZHTx4 sPWGa9euWY+6IrndXb7s52s/F+XPlcH+8Z/DwYH+TV+9O8qfytWzx38OAQAA AAAA4EkBya3P57NiwJkzZ65Zs8aBRBLR0eS2vLw85KMuSm6nHHhkwdGfRPh5 avM/OZvcbvv6hwt/cH+En6yHP0pyCwAAAAAAAIMijHObkpJCcpvIPJPcLjj6 k3XVr0T4mZz5VWeT213fuKf2Fx+L8FP4g/tJbgEAAAAAAGAQya17kdyOA5Jb AAAAAAAAOILk1r1IbscByS0AAAAAAAAcQXLrXiS344DkFgAAAAAAAI4guXUv kttxQHILAAAAAAAAR5DcuhfJ7TgguQUAAAAAAIAjSG7di+R2HJDcAgAAAAAA wBEkt45bvnz55MmTlyxZMtonktyOA5JbAAAAAAAAOCLm5PbatWu5ubmHDh2q qKgIfrSmpkYeqqqqGm0aGb/GxkaZsfT09OTk5OXLl+/fv7+np2f8ZyN6c+bM eeKJJxYsWDDaJ45FcrvPkChfjuQWiexsQnK6KrFwumahOV2VWDhds9Ccrkos nK5ZaE5XJRZO1yw0p6sSC6drFprTVYmF0zULzemqxMLpmoXmdFVi4XTNQnO6 KrFwumahOV2VWDhds9CcrkosnK5ZaE5XBWMi5uT297///RPDJk2a1NfXF/Do jBkz5KH09PTRppHBqqurN2/eXFlZOWJPv9+flpb2RBCZw6ysrPjnZIyQ3JLc IjE5vdcNzemqxMLpmoXmdFVi4XTNQnO6KrFwumahOV2VWDhds9CcrkosnK5Z aE5XJRZO1yw0p6sSC6drFprTVYmF0zULzemqxMLpmoXmdFVi4XTNQnO6KrFw umahOV0VjIn4k1uxadOmgEcNJrdbt26VSS1fvjxyt/b29pdeeknnZ+LEicnJ yWvXrp09e7bc1sa6urr4Z2YskNw+tfmfJmd+NcLPL9f/vbPJ7ZZ/+d/wNsLP tq9/mOTWe5ze64bmdFVi4XTNQnO6KrFwumahOV2VWDhds9CcrkosnK5ZaE5X JRZO1yw0p6sSC6drFprTVYmF0zULzemqxMLpmoXmdFVi4XTNQnO6KrFwumah OV2VWDhds9CcrgrGhJHkduLEid3d3fZHDSa3q1evjia5lbnVmZk7d679HbW1 tSUlJclErl+/Hv/MjAWS2yh/HExuo/whufUY3fe1Jwz37ouppClU0hQqaQqV NIVKmkIlTaGSplBJU6ikKVTSFCqJcWMkuRXr1q2zPxo5ub127Vpra2tZWVl9 fX24VxdvvvnmH//4x4ULF8qklixZEjzzVs+amhqdjXnz5r3++usB05EWecWA xuvXr8sUXnvtNaulv7//9OnTpaWlp06d8vl84eb84sWLVVVVMvPNzc32ebD7 /TDr1ZuamqS/vOXg2XgrKLmVPtK/vLxcXkjefrjivDWWye1+m8gJ7f4go0pu B6/6rgz2B/z09vfkHck5lHXgRHnpgK/Pah/tezFlcKBff3z9feWlJVkH9x/J ze7v6bbarZ+rgz6nZhJjgX2xKVTSFCppCpU0hUqaQiVNoZKmUElTqKQpVNIU KmkKlcS4iT+5TU9Pl/+ffPLJjo4O69Fwye3rr7++Z8+ep556yp76zpw589Kl S/ZuRUVFAX2CPf/881b/efPmaWNDQ0OEqDPgJaT/rFmzNORMS0uTt2BNXG4H XNdMVoRNmzY999xz9nmQbvJ2ArLic+fOyUNPP/203+9ft27dpEmTrP7PPPNM cNCqyW1qaqp8FqtXr5Y+Vv8XXnihvr4+3FvwQHIbTmdnZ05OTnZ2dk1NjZQx zqkZIbMhMyOzJDMms+f07CAEn89XFopsFrSDrBTV1dUFBQWy+jc2NspdbZcP V+5KY35+fklJiWyOtP2m3RdTSSoZJSppCpU0hUqaQiVNoZKmUElTqKQpVNIU KmmKeyuJEcWf3JaXl7/yyityY/Xq1dajIZNb+XynTJmiz5o0adK0adNefPFF a7wF+zi0R48ejRzbit/85jdWf00758yZE25ugx07dkye8tvf/nZgYMCaK3ss bB9dobCw0J7rPvvss/Z8NWCY39raWm23Yt6nnnrK/nR5d/b+mtzKnGgZtThW 56effjrcCcBjndxGMyrC3mHBz4p/ybx48WL2sJMnT8bwRsySGZDZ0PmRGXN2 ZhCOLPMhl9ITJ07oozk5OXJX/j906JDcOH78uD7x1KlTugDn5eXpU1paWoZu 4n0xlaSSUaKSplBJU6ikKVTSFCppCpU0hUqaQiVNoZKmuLeSGFH8yW1JSUlx cbHebm1t1UdDJrdpaWnaLSMjY+jGOAOyaL388svS+Morr1iDA/zhD38YuOF3 v/udPLp48eKBt7PyTLkdMkGNTJNbkZSUpP/Lu5BSvP766/Iuamtr7Z0vXLgg fV566aWjR492dXVdH3bmzBlNnp988klZTazOVnKrYfK5c+euXbv22muvHT58 WBt//etfyxu0+mtyq1auXCnTf2t46AZ5y9p44MCBkG/B28mtuHTpkp55Kx9N b29vNE/p7Ow/frwpyh/pHM005aVlBvRsW/4qlOAGBwd9N/T09Mi6I8tkR0eH PFRXVydLZkVFhXYrKiqSu7Lm+v3+gwcPyu3u7m55SPYdcrugoGDoJt4XU0kq GSUqaQqVNIVKmkIlTaGSplBJU6ikKVTSFCppinsriRHFn9weO3bs2rVretpq WlqaPhqc3FZVVYXLV8+cOaMPyYcb/EKarEa4QpksnPr0w4cPW42yVJx5O2mx P8tKbjUWtoalDaexsTF4BN3KykqdgixvVqOV3Mo7feONN+z9t27dqg/J0ms1 Wsmtff7fGg5vJ06cKO2rVq0KOUseuELZiGQFz8/Pz87Ozs3NPX/+/IgjJ1RV XfzpTzKi/JHOkacmLycvKi8tMyCzITNj6n0hHFmdT548KRt/WbmOHj0q/8vt 2CZVUlIii+Lp06f1blZWlqxisu/Qu11dXQ0NDTJxaZG9jOxEdOkaGBiQZ8mH PuTyfTGVjIBKmkIlTaGSplBJU6ikKVTSFCppCpU0hUqaQiVNcW8lMaL4k1v9 7n9ZWZnevXDhwluhktvMzEw9PTUgQVV68qp8jsEPjZjcWsGvfRQCq9Gyfft2 +7Ps59zaz4CNxvXr1/v6+urr648cOaIT2bFjh/WoldzKPAQ8saOjI7i/JrfJ ycnBL6R5eMiH3hrj5FY+i5ycnNyoSWf7aboGF1FZo2Xp0pEKioqK2traInTW 5PapX+393eQjEX6kw4jJrbyQvJy+rsyAzIbBN4VwZJ2S5efQoUOyOMlyJbdl R6AP9fb2yqP6h7yh4cGQZZUPt39pbm7WP+HpfkF2E3K3sLDw3LlzJ06cOH78 eF1dnbU3kXZ5tKqqqrW1tbS0VG7LXmzI5ftiKhkBlTSFSppCJU2hkqZQSVOo pClU0hQqaQqVNIVKmuLeSmJEppLbN998c/r06XJ3yZIlVhppT27nz5+v/ReE og8FX9HsrSiSW/nQ9em7d++2GhsbG5+5QR/NzMy0P8tKboPz1XC6uroOHDgg b+Tpp58OiIXtJxJHSG6lSnoa7auvvmo1aq2kCMGvuHDhwiduXEYt2Jgmt9nZ 2bKGDkRNOstTxiK5HRo+91VWfD35VgdPkO2DtfrbaXL74tO5S6fWR/iRDuGS W5msTFyHR9BTbeWlE+QqaZ6nf4aT5aeyslIWKin74cOHZW+ij8oHIQ8dPHhQ diKyvOmYOU1NTSEnpXsiK+fv7u7eF0S2XbqOyG7I3p6Tk6ODqLt3X0wlI6OS plBJU6ikKVTSFCppCpU0hUqaQiVNoZKmUElT3FtJjMhUcvuWbegA+aA1crQn sRrARma/xlnAEyMktxpghnu6NYVwya0sWuGmbLl+/bos2NZ8PvXUU8nJyfLu 9uzZM6rk9q0bF1NbvHix1RIhuZWPwKnkNjc3V1bhP0RNOutfatRYLKs+n6++ vl7HLhB5eXk1NTWtra32txlzcisTkUnJBGWyOn15IXk5edGxeC8I6fLly7Lw yMZfo3L5UGS9k7tWB/lEdCeiI+TI3ZDTkV2MPJqfnx8w5X3Df+CTVaOrq0t3 MbIDklfREdSlv2zE9A8QsloNuXlfTCUjo5KmUElTqKQpVNIUKmkKlTSFSppC JU2hkqZQSVPcW0mMyGBye/369VmzZulpt8uWLQt5zu3kyZMvhtfX1xf8Qpq7 ygTDzclbNwZbeO6551577bVwU4gnud25c6d2nj179smTJ+2vouffRpncynqh D23ZssVqJLkdFZ/PJ2t9QUFB9g3yulKEhoaGtra2iorz0Se30lmeIk+Up1uB sJCJy0uQ2Y4/+Szs+4XOzk65W1ZWZu8jK6AuYzrmeUjax77b0r/uHThwwPpY Gxsb9w1/O6O1tXXf8Bc6tF3WmqysLGnp7+93776YSkZGJU2hkqZQSVOopClU 0hQqaQqVNIVKmkIlTaGSpri3ktHM803OYHJrDy1feeWVgOTWujhXV1dXuGmG pGnw3LlzI/TZsGGDTjw3Nzf40fiT25deekl6zps3L/giZaNKbq3Tko8cOWI1 JmZym1CjJQTz+/2yaaqrq7NHuGLjxgPRJ7fS2f5cmZRMsH34oodjPf8ISXYW svDIJkLv6kZeditWB+vLGvtufH0jeCJ6zUphz951vB3ZNVgtFy9e1N2QXvtS 1lbrId0BNTc3u3dfTCUjo5KmUElTqKQpVNIUKmkKlTSFSppCJU2hkqZQSVPc W8lo5vkmZza5FcnJydaoAvbktqqqShvnzZt37dq1cJMNlpaWJs965plnZG7D 9env79cEdfLkycF5aZzJrbzTJ598Unru3Lkz4CFZ4PWhkMnt8ePH7Z1ff/11 HQpYtLa2Wu2JmdwmzhXKRtTT03Pu3Lnq6urCwsLRJrfyFHmiPD3mCynCIFmQ Dhw4YCXnsnmXZUn2QXrX2n3U19fLah5uJ9Lb2ysPFRUVBbTr1zSs/rqbaGpq 0mF87F/T0LPHrX2f03vg/xP9vphKRkYlTaGSplBJU6ikKVTSFCppCpU0hUqa QiVNoZKmuLeS0czzTc54ciufbMjkVqxcudIac6ChoeGNN97Q+LGmpmbevHny aMgXys7O1melpaVZs9rT03P8+PE333zT6mYNOTtx4kS5LdPv7++XZUPmQdtj Tm6vX7/+3HPPSc/f/OY3MsG3hjPY8+fPL1++3HqnIZNbsWHDBllir127dvHi xblz52rj+vXr7dNPzOQ2TmOxrEbjZEVz9MmtdHZqPhFMlth9w1eltFry8/P3 799vLcYtLS37bN/p0J1I8F/o9I96VVVVAe369MOHD9fV1emFLLOysgYGBuR1 9WsaZWVlp06dkg3avhuDqLt0X0wlR0QlTaGSplBJU6ikKVTSFCppCpU0hUqa QiVNoZKmuLeSI84zjCe3Qi9PFpzc+v3+2bNnW6nmxIkTn3/+eevuk08+6fP5 Qr7Qr3/96yds9PRaIQuP1e3atWvWaLTBJk+eLMuGfbKxjXMb4MUXX3zhhRci JLfBpkyZEvA2Eza53T8sMJONzlgsq5HJ0tXT05OfXx3bFcrgOB33prq6Wu/q 9yxk82LvE3BqdMgzpXXP0tjYGPxQU1OT9aWPI0eOWF8PkRu641DFxcV9fX1D I40539RSv6Vw3ubCZPuPtJxtqbP6bK84nXL0ZMCPNNp2ry2bNpYE/zQ2Xoh5 X0wlqSSVpJI3QyVDviKVpJJUkkpSSSpJJakklXRXJTGimJPbP/zhDzpQQHBy Kx+TppT2PFO9+eabx44d02FjLc8888y6des6OjrCvVZ3d/e8efMCItBZs2bJ QhLQs66ubs6cOZMmTbK6Pf/88+np6cHvsaioKPrkVmZ79+7dVmIsXn755R07 dvz+97/XIXa3b99udbaS28LCQo1elcyVdJa6BUxc39qSJUuCX5fkdkSy7ssW o7S0NC8vb7Tj3JLc3oQ04R8YGAh+SK99aR+lJ/K+uOrs8d9sf3D+kR/bf17Y 8Y2KM8esPj/ZWfjQ7qpH9p2yfuSuNFodysvPPvf0/unPH7f/PPf0vqqqppj3 xeODSppCJU2hkqa4pZIhX5FKUkkqGQGVNIVKmkIlTaGSplBJU0ZVSYwo5uQ2 TvI5NjY2njlzpq+v7/r169E8RT5Z6X/hwoVLly7J0yP0fOONNy4O6+/vNzS/ /+u1115raWmpq6uLPFkruW1oaHhr+J3W19fLbOvoEAY5m9zuHeZUcitrup5g b7/EWF5e3q5dR0luYcpI++LiKXu/t676FfvPtH3/EbAv/kHO+R8dbbV+5G7A vvh3z2cHLJ+/fT4r8bOdUaGSplBJU6ikKQ5WMuQrUkkqSSVNoZKmUElTqKQp VNIUKolx41Ry62FWcht8rTSzbsLk1u/3t7S0lJSU2NPayspKaezt7ZVHq6ou ktzCFLIdU6ikKVTSFCppCscsplBJU6ikKVTSFCppCpU0hUqaQiUxbkhujfNA crs/SLiEdjyT20uXLhUWFmpgm5OTU1VVdfny5YD3qMntk7/Y9fykwxF+pAPJ LUZEtmMKlTSFSppCJU3hmMUUKmkKlTSFSppCJU2hkqZQSVOoJMYNya1xJLfG k9u+vr4TJ05oZnvkyJHGxkZ5dyF7anIb5Q/JbWS9vb31w+SG0/PiDLIdU6ik KVTSFCppCscsplBJU6ikKVTSFCppCpU0hUqaQiUTjYcTDJJb4wLGuR07Yz1a QjwMLqIXLlzQq4/l5uaeOXNmcHAwQuf+fl9jY3uUP9LZ4Hx6ycDAwI4dO6ZM mZI0TG7I3ZCjizulr6+vsbGxv79/TF/lZsh2qKQpVNIUKmmK5ysZ8hWpJJWk kqZQSVOopClU0hQqaQqVTByJn2DEieTWOKlnZWVldXX1m2++OaYv5Pnk1u/3 19XV6am2paWlib+58Aafz7dixQrd3C0dphtAaUyo6z9euHBh//79BQUFDQ0N Y7Rs3AzZzhCVNIdKmkIlTfF2JUO+IpWkklTSFCppCpU0hUqaQiVNoZIJwi0J RjxIbt1rLJLbxOH3+6urqzW2le2P3HV6jm4WeXl5spVLTk6uqKjQjb/ckLvS mJ+f7/TcvY3uRPSPBWOxK7lJsp0hKmkOlTSFSpri4UqGfEUqSSWppClU0hQq aQqVNIVKmkIlE4SLEoyYkdy6l7eTW41tc3JyZCvh9LzcXFJTU2UTd/To0bM2 clcaZZtwIcGcOnUq4Hxv2W7LhrqpqSn+iSdmthP/+wqJSppCJU2hkqZ4tZIh X5FKUkkqSSVjQCVNoZKmUElTqKQpY1rJOC1cuDBcgpGamup0vmIGya17eTi5 lRVNz7ZtbW11el5uOtOmTZNNXH19vX27J3f12wf7biYx7IuTdj+SeXBN7g0/ 2JQdvC+WRqtDZuahlyYfGlW243RVYkElTaGSplBJUxysZMhXpJJUkkqaQiVN oZKmUElTqKQpVDJB6NgIIROMadOmOZ2vmEFy614eSG5r27r21J4P+Nlafnre zsPys76o0t4+eJUBE8bD3LlzZRN38uRJ+3avoqJCGmfOnFmSYI4fP37w4MGA Tbe05OXlFRUVxTnxGPbFU/Y+mle6t/aGH2UeCd4XS6PVITe37OVfZ40q2zFS t2BU0hQqaQqVNMWrlQz5ilSSSlJJKhkDKmkKlTSFSppCJU0Z00rGaebMmeES jLlz5zqdr5hBcuteHkhuV504/c8rD0T5MzA46PT83hQyMzNlEyfrfkNDg270 5IbclUZ5yOm5e5uBgYH8/Hxrx3HkyJHTp0/39PSYmn4M+2KXjoRJJU2hkqZQ SVM8XMmQr0glqSSVNDV9Kmlq+lTS1PSppKnpU0lT06eSY/FaMXBRghEzklv3 8kxy+82M4od2VUb4+eqabJLbcdPW1jZjxgzZyi1fvvzQMLkhd6VRHnJ67v6P tfsYux3HTZLtUElTqKQpVNIUb1cy5CtSSSpJJU2hkqZQSVOopClU0hQqmSDc kmDEg+TWvTyT3H5vX619YxX887UNR0hux1N9ff3s2bOTbOSuNDo9X/9Hdh+y 5I/1juNmyHaopClU0hQqaYrnKxnyFakklaSSplBJU6ikKVTSFCppCpVMHImf YMSJ5Na9NLktLS29doP9UZJbxKOvr6+goCBzmNyQu07P0duMz7J9M2Q7VNIU KmkKlTTF85UM+YpUkkpSSVOopClU0hQqaQqVNIVKJpQETzDiFJDcvvnmm1YM uHDhQpLbRKbJrd2FCxesR0lugTjdDNnO+KCSplBJU6ikKTfDMcv4oJKmUElT qKQpVNIUKmkKlTSFSmLcBCS3mzdvtieBJLeJ7Pr162fOnDlt88c//tF6lOQW iBPZjilU0hQqaQqVNIVjFlOopClU0hQqaQqVNIVKmkIlTaGSGDcByW1nZ6c9 Cezo6HAikoQBJLdAnMh2TKGSplBJU6ikKRyzmEIlTaGSplBJU6ikKVTSFCpp CpXEuIkwzi1cjeQWiBPZjilU0hQqaQqVNIVjFlOopClU0hQqaQqVNIVKmkIl TaGSGDckt15FcgvEaaR98fHJGf8y/cB/2X+e2/r/2ffFP9tV+N3M4u9tL7V+ 5K402vfFz0za/crzufafpyfuusmyHSoZLSppCpU0xcFKhnxFKkklqaQpVNIU KmkKlTSFSppCJTFuSG69iuQWiFPkffH4c+++mEqaQiVNoZKmUElTqKQpVNIU KmkKlTSFSppCJU2hkhg3JLdeRXILxIl9sSlU0hQqaQqVNIVKmkIlTaGSplBJ U6ikKVTSFCppCpXEuCG59SqSWyBO7ItNoZKmUElTqKQpVNIUKmkKlTSFSppC JU2hkqZQSVOoJMYNya1XuSi5/dqGIw9uKYrw8/+sPkxyi/HHvtgUKmkKlTSF SppCJU2hkqZQSVOopClU0hQqaQqVNIVKYtyQ3HqVi5LbKH9IbjHO2BebQiVN oZKmUElTqKQpVNIUKmkKlTSFSppCJU2hkqZQSYwbkluvckVy29l/5WxHT5Q/ fr/f6fnFzYV9sSlU0hQqaQqVNIVKmkIlTaGSplBJU6ikKVTSFCppCpXEuCG5 9SpXJLcYay0tLRkZGbt37x7n17106ZIHknb2xaZQSVOopClU0hQqaQqVNIVK mkIlTaGSplBJU6ikKVTSiNbW1qysrI0bN6akpCxevHjdunXFxcUDAwMxTMp4 CuHz+dra2gxOMGYkt15FcgtRU1OTlJQ0e/bscXtF2VTu3r1bXrS/v3/cXnSM sC82hUqaQiVNoZKmUElTqKQpVNIUKmkKlTSFSppCJU2hknHy+/0FBQXTp09P CrJgwYKurq5RTcp4CtHY2JiSkrJv3z5TE4wHya1XkdxiyInkdnBwUDe2JLfG uW5fbKGSplBJU6ikKVTSFCppCpU0hUqaQiVNoZKmUElTqGScsrKyNDpYvnz5 yZMnL168eOrUqT179syePVsaU1NT+/r6opzUWKQQhw4dkgmS3GJMkdxiiOQ2 PuyLTaGSplBJU6ikKVTSFCppCpU0hUqaQiVNoZKmUElTqGQ82trapk2blpSU tGfPnoDk6vz583oi7quvvhplqEVyC5ciucVQUHIrW/L169eXlpZWVFTIjeTk 5KVLl+7du9favmmH4uLigoKClStXSofNmzdLf2tZqq+vlw65ubn2V8nKypJG ee65c+fWrVun28w1a9bIlMfzzRrHvtgUKmkKlTSFSppCJU2hkqZQSVOopClU 0hQqaQqVNIVKxmPLli06KsKVK1eCH62qqtJU4fTp00MxpRC7d++WR5ubm7du 3SqvsnTp0kOHDl28eNF6+o4dO6TD5cuXrZbW1lZp0YsE7du3T54lE5w/f740 ykuMTRmiRXLrVSS3GApKbisqKuRu8Egysh3Toby1g5o2bdq8efP09sGDB3UK JSUlcle2XfZXSU9Pl8by8vJTp07ZJ7ts2bJxfr9mnU1ITlclFk7XLDSnqxIL p2sWmtNViYXTNQvN6arEwumaheZ0VWLhdM1Cc7oqsXC6ZqE5XZVYOF2z0Jyu SiycrlloTlclFk7XLDSnqxILp2sWmtNViYXTNQvN6apEa+HChUlJSceOHQvX Yc6cOdJBc9QYUohFixbJ7alTp8r/ycnJmoHINDs6OvTpmnWcP3/emqBUT1pS UlLk9vLly+3TrKmpGYsiRI/k1qtIbjEUJrkVK1eurK2tHRgYKCgo0JaGhoaA Dr29vdJSXV2tmzvdWEXeZg55cbSERON0VWLhdM1Cc7oqsXC6ZqE5XZVYOF2z 0JyuSiycrlloTlclFk7XLDSnqxILp2sWmtNViYXTNQvN6arEwumaheZ0VWLh dM1Cc7oqsXC6ZqE5XZVYOF2z0JyuSlR8Pp+GBrW1teH6rF27VjqsW7duKKYU wkputYO0r1mzJml4TF3tEDm5HWK0BIwLklsMhUlup02bZr9QY2pqqjSWlZVZ HWT71t7ebnXQqzRu27ZtiOQ2AThdlVg4XbPQnK5KLJyuWWhOVyUWTtcsNKer Egunaxaa01WJhdM1C83pqsTC6ZqF5nRVYuF0zUJzuiqxcLpmoTldlVg4XbPQ nK5KLJyuWWhOVyUWTtcsNKerEpXLly9raNDS0hKuz/79+60TaGNObnfu3Gn1 7+jo0HPSNOsguUUiILnFUJjkVjZi9j46Jszx48etDqtXr7Z3qK2t1REVhm6y 5BYAAAAAABjU1tamoYE9OA2wc+dO6bBq1aqhOJJb/WaxZcWKFdJYVVU1RHKL xEByi6EwyW1AMLtp0yZpLCoqsjps2bLF3qG5uVnP1B0iuQUAAAAAALG6cuWK hgaaIYSkIUNGRsZQHMlta2ur/SkbNmyQxvz8/CGSWyQGklsMxZrcbty40d7h 9OnT0piamjp0Y5upo81YdLwFklsAAAAAABDZypUrg4MFS1tbm45sUFxcPBRT CqHJbXNzc/CLVlZWDt1Ibs+dO2c9Ku0ktxhnJLcYijW5nT9/vn3hycrKksbM zEy5LRvGpOHrl1mP+v1+vVAjyS0AAAAAAIjsxIkTmhtUV1cHPOT3+zdu3CgP TZ8+va+vbyimFEKTWx0TUslDM2fOlMbLly/L3ZSUFLl9+vRpq0NeXh7JLcYZ yS2GYk1uRW5urj566dKlWbNmSUtpaancraurk9tTpkyx/ji1Z88e+zcdZBOq dy9evDhubxMAAAAAALjF3r17k4Yvj56Xlzc4OKiNHR0da9as0UihsLBQG2NI ITS5XbBggV6cXTrs2rVLz1LTDq+++qrc3bBhg95tbW2dPXu2PbnNzs4OHknS KSS3XkVyi6E4ktuk4cHAZTM1Y8YMa3gZ4fP55syZo8PerlixQrZ7Vn9rjBrt IE+0/1EMAAAAAABgaDhNzcjI0DBh6tSpS5cuXbBggRUv2E92jSGF0ORWSLu8 yrJly/RRa/yEY8eOaYeUlJS0tDQdnMGe3BYWFmrLvHnzdIAFB5HcehXJLURt ba1saubOnat3deSWNWvW2Pts2bIl6cYAMprcylZOR4DRDd2OHTtkU2n1P3/+ /OLFi/VR2XJu375dNqpy++TJk9qhtLRUv7kgBgYGxuu9AgAAAAAAdxgcHMzP z7cnsSI9Pb22tjag52hTCE1uMzIy9FQ0sWLFCvtk5aX3799vBbYLFy7UM3vl VbRDX1+fPEUfPXDgwLjUIyySW68iuUUM7Cfldnd3Nzc3+/3+kD27urpk42lP dO2kXToQ2wIAAAAAgHD8fn9XV1dzc3NLS0vky+VEn0JoctvQ0HD16lWZbE9P T8inDA4Oyut2dHREeEV5brhUZNyQ3HoVyS1iEHI4BcSps7Nzw4YN1ig9GC3Z lJWXl+/Zs2fjxo27du0qLi62xkHCqEglS0tLd+7cuXnz5ry8PPm1x+k58oKS kpKMjIyzZ886PSOuVFNTcySI9a03jMqlS5ekerI0ZmZmHj16lD+ejlZ3d3d+ eFJep2fQfaqqqnTfvXfvXrnt9Oy4ld/vl63i7t27t2zZUlRUFCFeQDgRfhWv q6uT2spSevjw4TNnzoz/vLnLiAc1HPVEKVyhOOoxyEpunZ4RM0huvYrkFjEg uTVO1sTly5dLVbdv3+70vLhSX1+fjkpkJztijlxGa2BgQBdFu4KCAqfny93O nTs3ZcoUqWRJSYnT8+JKS5cuTQoijU7Pl/uUlZVZ3xC0xmSTA0On58tNmpqa gpdGi16nFVGyLgtuJy2On7PkOvJb0KpVqwIqSSw2KhF+Fc/JyQmorXWNZgQb 8aCGo54ohSsURz1mkdzCFUhuEYPKysqpU6euW7fO6RnxiPb29tWrV+tul99h YrN169ak4avsHTx4UPa88jt2cnJyEn9fGL2dO3dqJY8cOSKVtK7HysmiMZP9 rHXtA5LbGAwODurYYrt3795nc+zYMadnzWVaW1t1OczMzJS1++jRo7qdtC6X jGh0dnbuC6Ij702bNq2lpcXpGXSToqKipOGLgB86dEiWSflfV3b+XDhauu+W JVDW6wsXLhw/fnzGjBlS2Pr6eqdnzR0i/Cou9ZTFUoopvxc1NzdbKW7w6JoY iuKghqOeKEUoFEc9Zi1btkzW8aamJqdnxAySW68iuQWcJYcn1nDo/A4TG5/P J0crUj17knPixAktaV9fn4Pz5i5SSV0a8/Pzrca0tDRpkV8OHZwxV9u1a5e1 gpPcxkCOlJNs19BEzGT/okd21gmNekFS2X6GGwgO0ejo6NAtZ1lZmdPz4jLr 16+Xum3atMlq0evhrl271sG5cp3e3l7dxeh1hJX+FjR//nwHZ8wtIv8qricz 792712rRqx1xBkuwEQ9qOOqJUoRCcdSDyEhuvYrkFnCW7mdfffVVjXf4HSYG ra2tixcvnj17dnd3t9U4MDCgtW1ubnZw3txFinb06NEdO3bYx/xnyYxHTU2N HjuvXLmS5DY2paWlHCPHT37f07MZ29rarEb5DbC6uvrs2bN8OT1mUrr09HQp 7JYtW5yeF/eR9Tpg/3Lw4EFpkQ2mg3PlOrIKS9GmT59uX5EHBwd1lB75HcnB eXOFCL+Ky5ZTy2gf81//nigiXyDpJjTiQQ1HPVGKUCiOehAZya1XkdwCzpKd ckVFhfyynZeXx+8wBtXV1SUNfwdTtnJOz4uLyfHgrFmzpJInT550el7cR36p njNnjlTvzJkzemoZyW0M9u7dmzT8jf4dO3asWLEiPT394MGD4a78i3B0qISF Cxf6fL5Tp04dGFZZWckvgXHS7/vPnDmzq6vL6Xlxn4KCAqme7GUaGxvl7vnz 5/U7vwwiOirV1dXBp9f29/drkiPru1Mz5hYRfhU/d+6c/jIZ8Oct/UOYPDru M5vQRjyo4agnSqMtFEc9sJDcehXJLZAg+B3GoL6+vsWLF0s9169f7/S8uJUs kDpyoy6WXLI2BprW7t+/37pNchsDPV05wJw5czhkHpUzZ85I3ZYtWxZwGSMp LzF4zGTDqH+dycrKcnpeXGlgYEAH8dD0W2/IIsoyOSoXL17U0tmvNlhWVqaN DOIRveBfxU+fPp0UargeaUliqNvwRjyo4agnStEUiqMe2JHcehXJLZAg+B3G FDkS1KhHDgO5ZnrM7Fet3bVrFwNnjZaeibdkyRIdQZTkNjZ+v1/znJSUlMrK ytbW1vLycj0rT45TGJ01elaMkzR8hTIp4+HDh3WsvIyMDKfnzq10oOApU6Zw wm1spG66bbTbt28fq/aoDA4Ozps3L2l4FGsNveV/vSS9OH78uNMz6BrBv4qf OnVKv60Q0FODMtkCjO8MugbJrSkjFoqjHgQgufUqklsgQfA7jBFyGKiR47Rp 006fPu307LhYR0fH5cuX8/Pz9XhwxYoVTs+Rm1y6dGn69OlTp061LjRPchsb v98vB86FhYX28dz066uCy6ZHT5Y9LZr9aoP6XXUh67uD8+Zea9askept3rzZ 6RlxKy1gSkqK7K97e3vlf7ktLenp6U7PmstUVVXpuiy7niVLlsj/VhJeXl7u 9Ny5RvCv4vol9ODkVlqSGIkiPJJbUyIXiqMeBCO59SqSWyBB8DtM/FpbWxcs WJA0PGheQ0OD07PjEXpSmeCksui9+uqr+s30EzfIobS0bNu2TW5zUkT8UlNT pZ4FBQVOz4hr1NbW6ops/x56d3e3NvKd3xhcvnxZq3fmzBmn58WV+vr69NpP Fy5csBplP64jiNr/WINoyFqsubfufSoqKpYuXcraPSrBv4rLwqm/VQb01C+D 8KtmOCS3pkQoFEc9CInk1qtIboEEwe8wcWppadFhx+bPn3/x4kWnZ8eVZI/Q 1NQU8Ff7wcHBGTNmSGHlMNCh+XIfXRTDqaqqcnoGXcPn88mxif2i3kq/CExy Gz3NH8TAwIC9fdGiRSyTscnOzpbSpaamOj0jbqXX1UpOTg5o11PIGJ01Nn19 fb29vUPDG0/dd7e1tTk9U64R/Ku4FFO3nPYLP8lWVBupbTgkt6aEKxRHPQiH 5NarSG6BBMHvMPG4cOHCrFmzpIDLly/n1NCYNTU1SQ2nTp3a399vNcptPQOK 88qil5WVtfft9HJvq1atktv8jh29s2fP6gFyc3Oz1djX16eJRF1dnYPz5i4D AwM6qq39BDxp1G9Vs9mMgQ6BcujQIadnxK107Zb9iz0T8/v9OkRPdXW1g/Pm LnIot3HjxhUrVtiHPSkvL08KdWktRBDyV3HNx+x/vNbazp49e9xn0DVIbk0J WSiOehABya1XkdwCCYLfYWImB3r65fTU1NTe3t7Bt3N67tzEOr12165dumuQ lszMzKThCx/YD64xWoxzGxufz6fR4saNG3V1lv8zMjKkZcGCBVzGaFT27Nmj AzbqgAmy5dy9e3fIIRwRDR3oki8jxGxgYED3ONu2bZOlURtzc3M1zmW0hFHR C2bt3btX73Z0dOjyWVxc7OyMuUvIX8VzcnJ0AAq9VKv8r6eFS7tDs+kCJLem BBeKox5ERnLrVSS3QILgd5iYWRfmCKmxsdHpGXST0tJSrdvcuXO3bt2q36QW J0+edHrW3I3kNmaFhYW6EM6ePTsjI0OXyalTp9bU1Dg9ay7T0dExZ84creTm zZvT0tKShi9rwvh4MZDfn3WxtI/RitEqLi7WMsp6vW3bthUrVujd/Px8p2fN ZaxKLl26dOPGjXo+3qZNm5yeL5cJ+at4b2+vjmq7YMEC2QfpN2iWLFlCShYB ya0pwYXiqAeRkdx6FcktkCCOHDkiO9ydO3c6PSPuo6dDhNPU1OT0DLrMiRMn 7MO0pqSk8J30+MmhtBSztLTU6RlxpeLiYvsymZqaynodm87OzvT0dL0slIY8 BOCxaW1tlQJKJUlv4nTy5EkdHsH6oyHbydjk5+frOcxqy5YtOuAtohfuV3FZ 3/U8WyW3+ZNNZCMe1HDUE6XgQnHUg8hIbr2K5BYAEMDv91++fPns2bN8XxUJ QpbJ9vb25uZm/b4q4iG/+7W0tLB2I3F0dXXJ2t3Z2en0jLibz+e7cOGCrN1s J8dCT0+PLKWyJ7IG9wCARENy61UktwAAAAAAAIB7kdx6FcktAAAAAAAA4F4k t15FcgsAAAAAAAC4F8mtV5HcAgAAAAAAAO5FcutVJLcAAAAAAACAe5HcehXJ LQAAAAAAAOBeJLdeRXILAAAAAAAAuBfJrVeR3AIAAAAAAADuRXLrVSS3AAAA AAAAgHuR3HoVyS0AAAAAAADgXiS3XkVyCwAAAAAAALgXya1XkdwCAAAAAAAA 7kVy61UktwAAAAAAAIB7kdx6FcktAAAAAAAA4F4kt15FcgsAAAAAAAC4F8mt V5HcAgAAAAAAAO5FcutVJLcAAAAAAACAe5HcehXJLQAAAAAAAOBeJLdeRXIL AAAAAAAAuBfJrVeR3CayswAAYFw4vc8HAAAAYkdy61Ukt4nM6aNYAABuFk7v 8wEAAIDYkdx6FcktAAAAAAAA4F4kt15FcgsAAAAAAAC4F8mtV5HcAgAAAAAA AO5FcutVJLcAAAAAAACAe5HcehXJLQAAAAAAAOBeJLdeRXILAAAAAAAAuBfJ rVeR3AIAAAAAAADuRXLrVSS3AAAAAAAAgHuR3HoVyS0AAAAAAADgXiS3XkVy CwAAAAAAALgXya1XkdwCAAAAAAAA7kVy61Uktwmuvr5+87CsrCyn5wXAuJLt 86ZNm5566qlHH330v//7v3/zm9/k5eU5PVMAAAAAgIRDcutVJLcJbtGiRROG feELX3B2Tl555ZUnh507d87ZOQFuBmVlZffcc8+EIF/60pc6OztjmGBhYaGu wmlpacbn1o5tBQAAAACMM5JbryK5TXCJk9zee++9OifHjx93dk4Az6uvrw8Z 26qvfOUr/f39o53m8uXL9enf+MY3xmKeLWwrAAAAAGCckdx6FcltgosmuW1q aqqsrPT5fBGmMzg4ePbsWfk/5jkZMY1paGiQhyLPBoARXbp06WMf+5iubnfd dddLL7107NixvXv3/vCHP7TC2+985zujnWyUyS3bCgAAAABwHZJbryK5TXDB yW1paemtw6Tl0KFDf/3Xf60d7rzzzuTkZOuJ3/72t7Xbtm3bnn766dtvv136 vPe97/3Od75TVVVldduzZ492++IXvxiu8eWXX/7oRz9qRUYf/OAH5e6KFSu0 c3t7+/e///33vOc9+uif/umf/vu//7u86HhUB/CiWbNmWavb+vXr7Q89/vjj 1kMlJSVD0a3CXV1dss7K6q9PfOc73yl3P/3pT2vnKLcVMintJhMP18i2AgAA AAAcQXLrVSS3CS44uS0uLtaWe+655wMf+EDA16j37dun3b75zW9qy6c//emA Pp/4xCc6Ojq0286dO7Xx85//vPWiAY2/+tWvJgTRlNjv98vUgh+95ZZb9u7d O66VArziH/7hH3Q9krU44KH+/v73v//9+ugLL7wwFN0q3N7eHrySvu9979PO UW4rZFLaKBO3XiigkW0FAAAAADiC5NarSG4TXITkdsLwKW3PPvvsQw899Cd/ 8ifa8sgjj2g3K40RDzzwwIsvvvjwww9bLT/+8Y+1WzSxz549e+Tp7373u7Xx hz/8odwtKCiQh/Ly8rTxYx/7WFFRUU1NzWOPPRYwJwBG5UMf+pCuRElJScGP /vM//7M++uijjw5FtwoPDAzIOvvggw9q48c//nG5O23aNO0c5bYimuSWbQUA AAAAOILk1qtIbhNc5OQ2NzdXG3/+859ry+c+9zltsdKYe++9t62tTRt/9rOf aaP1ReloYh8VcuzKtWvXWhNsamoaGs6ISobV1dWNVVEA7xocHLRW8MzMzOAO 1sr+la98ZWg0q3C4cW6j3FZEk9wqthUAAAAAMM5Ibr2K5DbBRUhu3/Oe9/j9 fm3ctm2bNn7yk5/UFiuNmThxojW10tJSbbztttvkox+KO7nt6uq66667tP0d 73jH5z73uZ///OeZmZnWjAEYFVl3rFPoV69eHdzhBz/4gT769a9/fchocht5 WxFncsu2AgAAAADGDsmtV5HcJrgIye1HP/pRq1tOTk645HbBggVWt46Ojgk3 nD59eiju5FYsW7bs7rvvnvB2f/M3f6OXTwIwWta69swzzwQ/+tnPflYffeyx x4aMJreRtxVxJrdDbCsAAAAAYMyQ3HoVyW2Ciz+5nTRpktUtwjm3H//4x61u 1veao0luRW9v7+LFix966KF77rnHCmQefvhh08UAbgrf+ta3rFRzYGDA/tCp U6fe+c536qMpKSlDo1mFR0xuI28rrJA2PT3d6mZddCya5HaIbQUAAAAAjA2S W68iuU1w8Se399133+XLl7XxF7/4hTY+8MAD2mJdNmjCjTPrhmxfxw6Z3O7Y scNqrK+vLxrW2to6NPxF77S0NO12++23Dw4OjlFZAA9bv369tVb+9Kc/tdp7 e3s/85nPaPutt97a0tIyNJpV2EpurdGwVZTbiq9+9ava8v3vf19b6urqbrnl lgjJLdsKAAAAABgfJLdeRXKb4OJPbicMXxLoxRdffPTRR62Wxx9/XLv19vbe euutVm7z7LPPPvLII1Y3e+xjRUYytblz51ZUVEjj888/b81eeXl5S0vLyy+/ rC1///d/Pz4lAjzG7/f/x3/8h339lRVT1ln7earLli3TztGvwhkZGVb7L3/5 S2sQ3Si3Fc8995y2yMvJU37961/ff//9Vjd7csu2AgAAAADGGcmtV5HcJrj4 k9uPfOQjE97uL//yL60z68RDDz0U0OG+++4Ljn2sa82r5OTkoeHz6O64444J oaSlpY1HgQAvki3zv/7rv4Zcs8TUqVPtnaNchRsbG+1r6/ve9z5tj3JbUVRU dPvtt9sfveWWW6yn2JNbthUAAAAAMM5Ibr2K5DbBpaamarjxxS9+UVtKSkqs cMbqlpubq42f+tSntMVKYxYtWvSzn/3stttuk9t33XWXtJeVldlfoq+v7zvf +Y41eOaXvvSl7du36237uXCdnZ2/+tWvPvShD1mT1fbq6up/+qd/ss76e8c7 3iHzsGfPnjEtC+B5PT09jz/++Lve9S57yPmRj3xkzpw5AT2jXIWHhgfF/fKX v6xbg7vvvlsbo99W7NixwxoJ4c4775T+3/jGN/Turl27rG5sKwAAAABgnJHc ehXJrVdZaYx+q3pgYKCmpibgakd2/f39RUVFTU1NI07Z5/PpFYvspKWioqK4 uLivry/OOQdg6ezsPHr0aEZGxrZt20pKSiKMBxv9Kjw0HPb6/X69PdptRUND g8xJNGs62woAAAAAGB8kt15FcutVAWkMAITEtgIAAAAA3I7k1qtIbr2KNAZA NNhWAAAAAIDbkdx6FcmtVy1btuyXwwoKCpyeFwCJi20FAAAAALgdya1XkdwC AAAAAAAA7kVy61UktwAAAAAAAIB7kdx6FcktAAAAAAAA4F4kt15FcgsAAAAA AAC4F8mtV5HcAgAAAAAAAO5FcutVJLcAAAAAAACAe5HcehXJLQAAAAAAAOBe JLdeRXILAAAAAAAAuBfJrVeR3ALBLl26ZGpSsn75fL4oO3d0dHR1dZl66ZuN 1HlwcNDpucDNhW2FG7GtAAAAgPeQ3HoVyW2Ca2trKywsrK2tdXpGbiK/+tWv JkyY8Nvf/jb+SXV0dKxcufLo0aPRdG5ubk5KSpo6dWp7e3v8L+24np6ewvAu XLhg79zd3Z2dnb1p06Z169YdOnQo3AJ/5swZ6bZ+/fotW7YcOXLk4sWL9kdL S0uXL18uq8wYvqvx1dfXV1BQsG3btrVr10px9u/ff+7cOadnCv+HbYURbCsA AACA+JHcehXJbYI7ceKEHKGvXbvWkVc/d+7c7t27y8vLHXl1g6qqqn73u9/t 2bMnms5f+9rXJkyY8J//+Z9xvmh7e/ucOXO2bNkS5SpWX1+fNKy5uTnOl04E TU1NSeEdO3bM6nn69OkZM2YEdNi1a5f9pDi/33/w4MGAPtOmTauoqLD32bFj x8yZMw2eBumgs2fPBpdFbN++PYapjfW6nJ2dLdPv6+sbo+mPG7YV449tBQAA ABA/kluvIrlNcM4mt6WlpfLq27Ztc+TVDdq4ceOECRN++tOfRtP5/PnzK1as iPOIXtasZcOiX7/8fn9NTc3p06fjed3E0d3dnRckNzd36tSpslDJoqXd2tvb p0+fLi3Lly8vKChoaGg4dOjQlClTpKW4uNiaWmVlpSYw8qh8QNJty5Yt2tLZ 2Wl1kxquWrVq0aJF/f394/2GjWptbZ09e7a8u/nz5+/evbusrCw/P3/16tX6 lg8fPjzaCY71uizzKdO/fPnyGE1/3LCtGH9sKwAAAID4kdx6Fcltgouc3Mqx 54gfXzyf782Z3BqRmZkppTtz5ky4DsZHWTQ4wXCTqq6u3rNnjyx1IR/Nyck5 fvx45CnX1NRIWebOnStbHm3Zvn27tCxYsGBgYMDqlpWVJY3Lli2zWmQhlBb7 mZAyG8uXL5fGgBc9f/68NG7YsCHcfCa+/v7+hQsXyrtISUkJOIs1NzdXM6iy srJRTTOadTmebcXNmdwawbYiJLYVAAAAwKiQ3HoVyW2CC0hu+/v7U1JS1q1b 19TUJP/PmDFj6tSpS5cuPXXqlPUUOSiWPrW1tUVFRXJjypQpCxculMNY+5h+ clseWr9+vf215DBWGrdu3Wq9kJ71N336dLm9ZMkSe+fm5uaA4QeNSE1N/fKX v3zkyJFZs2Z98YtfvPvuu7/2ta/pOVcrV678u7/7uw9+8IPSLo8GPHHHjh2/ +MUvPvnJT95///3f+ta3Nm3apO2yhP/jP/7jJz7xiQkTJnz4wx+WiT/44IP2 18rLy3vllVe+8IUvvPe975WKSfurr74q7VoHafnqV7/6b//2b/ZIKi0tTRon T54c7l3oKJT2MMFy7tw5+TSlsPLBrV69Oj8/v7u7Wx/y+XxSZ/k09W59fX1K GPaTxGRJkPe+YMEC+aBlzrOysqygw5qZET+p7OxsmWxxcbEsJ7K0yKTkPVZV VQV0S09P1/QvOOg4fPiwJo0RXkWeJTWRbvbBPDXxq6urs/eUt9DS0tLa2mq1 yNIeHLxo5JWbmxvwQnpuqhQw8rtOWLLyyvzL2h18MqfUUJZt+zYhnnXZyLZC 6i+3NU+eN2+e3C4pKdFuY7SVUGwr2FYM3fTbCgAAAMBCcutVJLcJLiC57enp 0UhHc5hZs2ZZ4/jJUbn22b17t9xdvHixtlt97OfvyaGutKSmptpfq6GhQRpX rlw5NJz2yAH+zJkzdYRAuS0TtHpqsmR/UVOeeeaZCRMmfOUrX5H/3//+908Y 9r73vW/SpEly46677rrjjju0ccqUKdazpk6dqo1333239NHbMs/y0MDAwP33 36+Tete73nXffff97d/+rf21PvvZz2p/mXJhYaG0v/TSS3J34cKF2k2Hsvzh D3+odxsbG9/97nffdttt9u/nBti3b59+Vzegvb6+XoqpX4GX4msNlyxZouet yaxqtbXz6dOnF7ydflNY2D9HHfVRKqCLhNi4caOVlkT5Sa1YscJakKwFRqZ5 9uxZezdZKtLS0pKGh1q1BzJ62pssYPYvIwerqqoKOIlO37K8kGyFZA7z8/P3 799fWVkpy3nAc48fPy49V61aZT23o6MjOTk5KdRQnzKdpFjHg00EOTk5Mv9W ohhA11P5uPVuPOuykW1Fdna2TFOfKB+u3NbQbOy2EoptBdsKthUAAACAheTW q0huE1zI5FaDFDkaHRo+jUqP6zMyMrSPpjFi8+bNeo7W+fPn5RBYWrZs2aJ9 RkxjVLhvWJeXl+tL2M/1NUITkltvvVVmfmj4+uBWWvLCCy/4fL6urq7JkyfL 3QceeECfUldX9453vOP222/Pzs4eGj7AX7hwoXT4i7/4C2uyIb8Bra8lXnnl FfspWwFpzLlz5zTM0cFFv/nNb+pTIrwLjcKCR6HU4RblA9Ioo62tTaMPPYss II0JII/qB52Xl6ctsvLOmzdPYwfNKGRWdYLWNX2i+aQGBwc1I0pJSdGTPGXx 0CkHX6epr69Pv3dsBTJRRjHSecmSJUlvP4lOTzhctGiRfg/azjpvU/X29sqS qTGjFHDnzp26SIccSETqIA8lJydHmJ9EtmHDBpn/nJyckI/KR6Alam9vH4pv XTa4rQgeLWHsthKKbQXbCrYVAAAAgIXk1qtIbhNcuORWjjetPnJ4njR80Ra9 q2nMwoUL7YMByvG+niKld+NMbuX4XR6SeTM+PKAmJI899pjVsmrVKmm59957 rZYLFy7cMkwjCJkZOai3R0bScs8998iz5G1qS4Q05nvf+17APASkMWL9+vXS 8ld/9Vfp6ely4wtf+ELkgSL13Lbg7x3L5yjtWVlZVktfX19nZ6fP5xsaKY3R JMd+jlxjY6PGIPZPobi4OGl45EarFCN+UjrYo7B/R/7AgQPhsg4rkNmxY0eU UYyoqKjQhMT+BW0rLBJpaWlHjhwpKCiwrsN18uRJ+xRqa2vtJ5lrMmOfZ0tH R4d2cOnGTU9hDf4GukWzspqamiETya2RbUVwcjt2WwnFtoJtBdsKAAAAwEJy 61UktwkuZHIrB+z2NEDPGrKGDdQ0ZufOnfbpyKes50rFf57emNKExD4yw9Gj R6XFGnBS3XvvvdIYcPEmOd7PyMiQg/o5c+Z85CMfkQ7W6IUR0phFixYFzENw GiMeffRRPenuzjvv1MQsnP7+fo0Curq6Ah4qKSmxJw/ywdlDkghpjBQhafh8 M/uolQUFBdqYbyOfe9LwcKMR5jBAUVGRPGXNmjX2Rr0SVrgvEUvldRTKKKMY Wfz03ELrBD+lC5gu4RpJaWeNnuyLoqZMOp/Z2dkHDx5cunSpluv8+fMBLyeb Ne1sjQvqLlpbHbI1mCwzGvfp99PjT26NbCvG/wplbCvYVrCtAAAAACwkt15F cpvgQia31hCX6tKlS8HJrRzsB0xKDtuTbnzZNsGT27S0NKslZBpz33332dOY 5cuXf+Yzn5kQJJo0xv5aKmQa09raetttt0n7pEmTIr8FK40JjgJ8Pl92draO OKqWLl1qZTvh0pgzZ85MmTJlxowZFy9etLdbX3UPyQo3RqSXYg+4do9+YV8v wxTS3r179YW2bNky4kmVZWVlSUEn0elb04nIDXu7fjNaSqHvQko6derUgDBn cHBw69atSaGu7iTPCheIucKOHTtk5qXCIR9ta2vTd6cnysaf3BrZVjiV3LKt sGNbcbNtKwAAAAALya1XkdwmuJiT24BRB63z9HSQRk1jAq7urVeocV1yW1JS csstt9x5552PPfbYunXrcnJyysvLP/WpT5lNY5KTkzXh+fCHP6wDPEag0UG4 bv39/dXV1fv27dPRF4UOWRwyjens7NRuAV8HHrpxbZ01a9Y0BhnVNaFCjrSp F/QJPkVNZWdnJw1/y16Ho9y5c2eEQEa2MLKkSbeCgoKAh9rb27UCAYmflWjp aZ8yb7rYB7yKLvlWhmnp6uoabSSVUPTMxhUrVoR8VKMt61zTeNZlg9sKVyS3 bCvYVnhsWwEAAABYSG69iuQ2wcWc3Mphsv2T1fEJZ8yYocezcvCrh6v2c730 S6/BaY91raJxEEMa89hjj8ntJ554wnpUjs31qvEBacyPfvSjyK+lgtOYqqqq O+6440Mf+tDjjz8uD333u9+N/C50GNKAi60PDW9I7R+Kz+fTC7XrtaiC05jB wUHtsG/fvuBXkXeXNPwNaHtGIbf7+/sDTleLwMo97F8n17EfAwblsFhRTGdn pzWOZYRARr/3nZycHJyNSDU09ikrK7O3azBoXTZI0xhZegPmR0+3mzJliv2L 4eLixYvaP8oiJBprNFH7FZqUfDSazmVmZmpLPOuywW2FJrcBp3qOKbYVbCuG bvptBQAAAGAhufUqktsEF3NymzT8bWv9cOVZ+vVnazpy4Kzfw83NzdUj3AsX LshkA9IY60oxAcfRvb29mZmZ27dvDzh/KX4xpDH6lIcffth69Omnnw74BrTM qtz91Kc+Ffm1VEAaI/X5/Oc/Ly3p6elSyT//8z+X2+vWrYvwLvTqQgEJg9Rw +vTpSbZLt1tpjH7RODiN2bVrl7SsWrUq5Eoqb1/PjTxw4IC2yMcqn4u0yKxq y4iflAYdAQN4VlZWhjvn04pi9Ny/obdfhCg4kJHq6fW2gk+iUzoC55w5c6zQ r729XWPArVu3WhPRd7pv3z7rJfr7+9esWSONq1evDphmTU1NuPl3C12Lp0yZ IlsA67NraWnRUssqadU/nnXZ4LZCn2X/lMduK6HYVrCtYFsBAAAAWEhuvYrk NsHFnNzql3BnzZq1ZMmSKVOmyO25c+daR9BDN0Ys1Knp91j1KfY0Rs9HEjNn zrTnD/ptblFRUWH2/caQxuTl5d1yyy1y97Of/ex//dd/PfDAA3L7Ax/4gD2N kRva5+Mf//j//M//hHstFZDG/Pa3v5W7//Iv/6J3d+3apdMP9+3goRv1CR5l Qi8JJKSYW7du1ZhC6q9nMwakMda4jqmpqelvZ53zdurUKf1wk5OTpX3OnDka 9zU2NtrnJMInJdVLsp3AqfRi8cHjrObk5AREMcoKZIKvUqQzMG/evHBfRpZ2 6/pBq1atWr16tWZW8ir69Wf7fCYNjxKQkZGxZcsWPQFPOtfW1gZMU8fVlKeE fEVXkM2yLCT6lqUyslbqyZl6fmBzc7O9c8zrssFthV4oKmn4DF79tv7YbSUU 2wq2FWwrAAAAAAvJrVeR3Ca48vJyPXjXu729vXrSkb2PXrHIGvdS0xg5FJVD Yz1Ul+NcmULAeIZyFCyHtHrkmzR8BfOqqqrgs5IKCgr0pKapU6dajXrSncyG 8WEtn3322QkTJtjPgJIZkJZvfvOb9m7333+/NFrfe928efOf/dmf6blzd999 txzUP/jgg3Lbfi2b1NRUPQXuAx/4QLjXUi+//LK0axIu9b/tttvuuOMOvVqT euSRR6TDt7/97XDvoqura8aMGcnJyQHnlQ0ODh4+fFjPV9QkQT4XK9nQq5xb 39utrq5OCsN+MZ2TJ0/K+9UkTT5ueTv2mGjETyojI0OeWFhYaG/UzFDee0Dn nTt3BkcxSgMZ608MFukfPP0AAwMDMhu6rOqSJgth8KvIO7XSS7V06dKADFPJ uiDL/HiOuToWpCxScGsNVfL5NjQ0BPSMeV02uK2Qam/YsEEnojnY2G0lFNsK thVsKwAAAAALya1Xkdx6j6Yx+oVT+XwvXboUYSRD+fSlQ29vb4QJ+v1+OdYO mEjAMIyJ4Pz58w0NDRHmSt5Ie3t75DcbQKYWPBRkyMYAOmCj/B/y0a6urra2 thEvsx69gYEB+RxDDjVp8JPSgTHDPSoFCTkDUZK3cPbs2cbGxsiF7e7uPnPm jPQM951uDRWDR4h1KSm4LNi1tbXyru0nFgaLYV02vq3w+XzSwVqwE3ArodhW sK0Y8ty2AgAAADc5kluvIrn1HnsaAwdt3rx5zpw5nZ2dTs/ITaSnp2f+/Plr 1qwxmHR5GNuKBMG2YvyxrQAAAIDHkNx6Fcmt95DGJAhZs/bu3Tt//nzrejoY U21tbSkpKdu3bw83TiYCsK1IEGwrxhnbCgAAAHgPya1Xkdx6T21t7eHDhyNc Ewfj6cSJE2N0hSYEOHnyZFFRkdNz4SZsKxIK24pxw7YCAAAA3kNy61UktwAA AAAAAIB7kdx6FcktAAAAAAAA4F4kt15FcgsAAAAAAAC4F8mtV5HcAgAAAAAA AO5FcutVJLcAAAAAAACAe5HcehXJLQAAAAAAAOBeJLdeRXILAAAAAAAAuBfJ rVeR3AIAAAAAAADuRXLrVSS3AAAAAAAAgHuR3HoVyS0AAAAAAADgXiS3XkVy CwAAAAAAALgXya1XkdwCAAAAAAAA7kVy61UktwAAAAAAAIB7kdx61eDgIMkt AAAAAAAA4FKDg4M+n8/plBHm+f1++WTlf6cXMQAAAAAAAACjY4V7TqeMMO+N N95gwAQAAAAAAADAjXSohDfeeMPplBFjQpN5wlsAAAAAAADARa5evcoJt952 /fp1/ZTlf4ZNAAAAAAAAABKc3+/XQG9wcPD69etO54sYQ/L5yid+5coVvWAZ AAAAAAAAgMSklyS7cuWK3+8ntr1JXLt2bWhoSD93AAAAAAAAAAlocHBwaGjo 2rVrTqeJAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAICb1/8PHe8h GA== "], {{0, 105.}, {927., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{927., 105.}, PlotRange->{{0, 927.}, {0, 105.}}]], "Output", CellChangeTimes->{ 3.763326237590453*^9, 3.76336262207662*^9, 3.763362750427683*^9, { 3.763417583274341*^9, 3.763417590024332*^9}, 3.763744000667651*^9, 3.7637446349843473`*^9, 3.7637449119618998`*^9, 3.763810069604693*^9, 3.763815816839136*^9, 3.763914843236204*^9, 3.763989558072072*^9, 3.7640811118507757`*^9, 3.764589533256353*^9, 3.798186208647283*^9}, CellLabel->"Out[26]="] }, Open ]], Cell["\<\ The key part of these blocks is the attention module comprising of 12 \ parallel self-attention transformations, also called \ \[OpenCurlyDoubleQuote]attention heads\[CloseCurlyDoubleQuote]:\ \>", "Text", CellChangeTimes->{ 3.763326267345145*^9, {3.763384112938608*^9, 3.7633841250828342`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetExtract", "[", RowBox[{"net", ",", RowBox[{"{", RowBox[{"\"\\"", ",", "1", ",", "1"}], "}"}]}], "]"}]], "Input",\ CellChangeTimes->{{3.76332625997388*^9, 3.7633262727326736`*^9}, { 3.763417593620451*^9, 3.7634175940828657`*^9}, {3.764595686313715*^9, 3.764595694121537*^9}}, CellLabel->"In[27]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsvQeUHFl9/8t6YQ0G/70G1saATbANhsM5gI8f8J7/hIcfxoADxtgYTLL9 WNj15gi7GEbSSitplFZppFEc5dVKWoVJPTnnnDXTE3tyas8wbcyy4X3p35n7 7laH6VDdVV3z/ZyRTtWtW7eqbldXfer2r+59z7/d/5U7f+01r3nNw6/Hf1/5 18f/74ce+tcf/cPtmPnH+x6+6/v3fe///cJ9j3zv+9976BP/disS/wz/deDf r6ZfIYQQQgghhBBCCCGEEEIISSIvvPDC4uLi1NTUxMTEOCGEEEIISTCQrunp aQgYNMxqEyTJ4OWXX8bHPTk5OTc391//9V8/I4QQQgghSQHqBQGDhi0sLEDJ rLZCkkDw+c7Ozs7Pz+NzX1lZ8RFCCCGEkCQCAYOGQcagZBRvB4MHq8XFRXzW Vp9xhBBCCCHrF8jYoh+r3ZAkhBdeeGFubo7KTQghhBBiOVAyiBljvB2J1+td WlpiYAkhhBBCiOVAySBm0DOrDZGYj4RzW32KEUIIIYSQXyEB3lYbIjEfRnQT QgghhNgHie622hCJ+Ug/gVafX4QQQggh5FdIX4JWGyIxH1o3IYQQQoh9oHU7 FVo3IYQQQoh9oHU7FVo3IYQQQoh9oHU7FVo3IYQQQoh9oHU7FVo3IYQQQoh9 oHU7FVo3IYQQQoh9oHU7FVo3IYQQQoh9oHU7FVo3IYQQQoh9oHU7FVo3IYQQ Qoh9oHWnNAsLC3MaL730klpE6yaEEEIIsQ8G60aKbnGYtcIlSUT84he/uOfV DA0NqaUpYd2do7VXGw9G+Lf8s6Xk7+HPlpdajm2N8G+soSz5e0gIIYSQlMBg 3adPn9Yt7uTJk1boJImI//mf/8FnlJub61nlxRdfVEtTwrrh0j84+acR/v3X 8mLy93DZu3j2s78T4R/EO/l7SAghhJCUwGDdS0tLSuG2bNly4sQJC2ySRIZY d2NjY9ClKWTdG3K+uqvse2H+Hjj3SWut+7nP/27lN/8gzJ/rH95B6yaEEEJI GMLEde/Zs4fWbWccY93w6qy2n4b5e+TiZ6217ue/+LbOu94b5g/iTesmhBBC SBho3akLrTsJ0LoJIYQQYgq07tSF1p0EaN2EEEIIMQVad+pC604CtG5CCCGE mAKtO3WhdScBWjchhBBCTIHWnbrQupMArZsQQgghpkDrTl1o3UmA1k0IIYQQ U6B1W05GRsYjjzyyf//+aFekdScBWjchhBBCTCFm637xxReLiory8vKampoC l7a3t2NRa2trtCYZP/39/dix48ePp6enQ2izs7Pn5uaSvxuRs23bNsjzrl27 ol0xEdZ9wyQi3BytmxBCCCHrhJit+7//+79l1Pj7779/YWHBsHTz5s1YBPWN 1iQDaWtrO3fuXEtLy5o5V1ZWDh06dE8A2EOXyxX/niQIWjetmxBCCCGOJ37r BmfPnjUsNdG6n332WRSVkZERPtvk5OSTTz4p+3Pfffelp6efPHly69atmJbE rq6u+HcmEdC6Hzj3SXh1mL+7T/0f1lr3+b/4lXiH+Xvu879L6yaEEEJIGEyx bpjt7OysvtRE6z527Fgk1o29lZ3Zvn27fkTj4+NpaWko5OWXX45/ZxIBrTvC PwutO8I/WjchhBBCQmGKdYOsrCx9aXjrfvHFFz0eT0NDQ09PT6itg5deeukX v/jF7t27UdT+/fsDd17lbG9vl93YsWPHCy+8YCgHKdiiIRESjhJ+/vOfq5TF xcXu7u66urqOjo6lpaVQez46Otra2oqdHxoa0vdB57/9qK0PDAwgPw45cDde CbBu5EF+uDQ2hMMPVTmvJNK6szXC23V2AFFZ9/LPluDShr/5xbni0sI8V059 Y513aUGlR3ssZgHxlr+lxYXGulpXbnZpUcHi3KxKV38/W16yaicJIeuNm5Zi 9dGbiVnNTbFh9dGTpBK/dUOt8f+99947NTWlloaybijotWvXHnjgAd3Yt2zZ MjY2pmerqqoy5Ank8ccfV/kh25LY19cXRlMNm0D+p59+WgT10KFDOARVOKYN 72BOTk6ePXv20Ucf1fcB2XA4Bs93u91Y9OCDD66srOBh5P7771f5H3rooUBJ Fuvet28fPotjx44hj8r/ox/9CA8moQ7BAdYdiunp6cLCwoKCAjxPoRrjLM0U sBvYGewSdgy7Z/XuEGIlXV1dDcFQ39b+/v6amprS0tKOjo6ZmRm1Iq5LbW1t 5eXlZWVlzc3NCwsLFh2BE1hX1r20tBT0lMNNXzLIqVVRUYGbO04/zEo6zknM IrGkpKS2thayEVh4Yq16LZJXiX4SWpNkTeK3bljfT3/6U0zAGNXSoNaNz3rD hg2yFlx006ZNTzzxhIpR0eOucUEOr9zghz/8ocovpgp9DbW3geCyj1X+8z// 0+v1qr3SlV6PSKmsrNSd/OGHH9bd2BDW3tnZKelK0fEEoa+Oo9Pzi3VjT6Qa pXJUZth7qIb3RFt3JJeL634C14r/zBwdHS3wg1uz5V0gYgewG7I/2DFrd4YQ y6murg56QRDrbmpqwnRubm5RUREmcnJycJn1+b9HuF8jJS8vz+VyYQL/yyIS A+vKunFHC3rK1dfXy9LCwkLM4n+cXZjAKSor4rlPbk/FxcWyyvDwsKHwSG52 icNJNUnWJH7rxiNPTU2NTHs8Hlka1LpVByMXLlzwrcZm4Mv74x//GIlwThVQ AZ/0rvKTn/wES/fu3et9NcpFMR3UfsMj1g3S0tLkfxwFquKFF17AUcCc9cwj IyPI8+STT0KYZ2ZmXvbT29srTw0w6snJSZVZWbc8CLjd7hdffPHnP/95fn6+ JD722GM4QJVfrFs4cuQIyn/FH+6CQ5ZE3LOCHoKzrRvgUVpavPHRzM/PR7LK 9PRidfVAhH/IHEmZ2DR2QFq5+XRPiM9/41jSEAnHpQ+LcAWTW7Y0kfX19WG2 ra3N5/9GYxrXXqwOP8ddHrOqhY1Ei9jvZNKxxLrB8vKyOuXm5uZwZ8QdZ2pq yuf/8QXnEh73JFtVVRVmsas4zfD0h+nZ2VksgiViuqKiwlCy3Laako4l1p3Q miRrEr914xIKq5TmYni1LA207tbW1lBuDH2VRUG7/hYrDvM2Jb7+sjq0ViXi DOl9NUjR11LWLUqvwrBD0d/fHxgx3tLSIiXg3FOJyrpxpL/85S/1/NIfC8CZ rBKVdev7/4pfvKUDlqNHjwbdJQe8Tbkm+LKXlJTAeIuKigYHB9eMNmltHb3z exci/EPm8KVhc9goNo0dwG5gZ8w6LkKsZXp6uqGhAffZ7u7uyspKPFfG/CMO viP4ypeVlckFB5d6zKrnUyTi8iuzkhOXLFkkQt7R0WHGAa1HUs66sVZzczM0 D7dOnDD4H9OxHXttbS1OHpy9MutyuXADhSXKLB79cHahcKTAJ6GLcu/wer1Y C5d0Q2kpZ922rUmyJvFbt8RL4AIusyMjI68Es+6LFy9Ks7DBfgVpNMZnGrho TetW0q5HbqhExaVLl/S19LZuveU5El5++eWFhYWenp7S0lIp5PLly2qpsm7s g2FF3OMC84t1p6enB25InmWCLnolwdaNz6KwsLAoYpBZbx438RTFtxtnl0R3 4GllfHw8TGax7gf+4/pPHikN84cMa1o3NoTNyXaxA/wdnDgJ3GTVtxXfX3x5 c3JylpZ+9Towrhv19fXqLoy7LZzc7XYHLQffi7y8PKyrbvrS7g3NhgnA58UN VOYb/vCSgYGB4eFhiT/xeDyJPlinknLWjTumnAA43+TTh/LJovn5eSyV5laf /6kQN/RQJjk0NCQNrWKAOEUxi5MNZylOXZyBXV1dyhuRjqV4GMSZVldXh2mc k4YCU866bVuTZE3Msu6XXnrpqaeeusff2YgySd26d+7cKfl3BUMWBe3zZE3r xgkgq1+9elUl9vf3P7SKLIX262sp6w5041DgoQ83FxzIgw8+aFB6vQE/jHWj lqT5OjMzUyWG6TlQ+m+RVz4DSah1QzXxbfVGDDJjlURYt8/f5oyLgDR6S8AJ rhXqUqAj1v3Eg0UHNvaE+UOGUNaNYlG4hJRIEzc2bZM3OgkxC2ngwuktrwbD kJUALywsSDwnxFv9xKxiOw2IvWN1lSKR27hUysO7TKsXKnFl1n8Uwx088cfq WFLLuqWxVM4W3DJwUc3Pz8eZJktxmb3hfxcAuohzUs5APJ0FLUqcU7XAzM7O Bv7eqn58gXDq6Sr2SSe1rNvONUnWxCzrfkULt8CHLrqoW7TIc3j09zENK4ax bpHPUKurEkJZt6Gn8aC8/PLLOMnVfj7wwAPp6ek4umvXrkVl3a+svvi5d+9e lRLGuqUTckusG4/P+Dr/T8QgszxxJ8K6haWlpZ6eHon3AMXFxe3t7fAE/TBj tm4UgqJQIIqV8rEhbE5a/whxGPIyo4RoAgmxVg1i6mYt2SorK4M+5OJ2Ly9e 6b2UyCryCgyuQuLzIu0TExNiC3D+hoaGHD98IStmUsu68enjo4fmSSMGLrk4 GTCrMuB6K7oo8cOYDVoOZFIeGA0ly0McTjmceCKTUE1sRU5R5MepKE1DKsZJ kVrWbeeaJGtionVDTeGH0tx98ODBoG3djzzyyGhoAkeWf2XVmVFgqD15ZTVA 5dFHH9X73zaUEI91X7lyRTJv3bq1ublZ34q0e0do3fiOyKLz58+rRFp3VECD cQWoqKgoWAXbRSX09fXhgb2paTBy60ZmrIIVsbqSeYDCsQn6NnEqi4uLIsYq Bec/BFj/TUfdf2HRQZUb4EFVnFxPlNv04OCgzKqoEkzj4olpJWz49t3gC1lx kFrWLWH8ygDxZIdZPHzpeeQMubH6Nl9QDGeRb7UNVoVI+VZ/UkFOwymKe6I8 FeIroJeZWtZt55oka2KidevCKZ3g6datXiSUPjoiR0x++/btYfKcPn1aCse9 I3Bp/NYtY80HHYInKutWPwfgxqQS7WndtoowCQR6gIt/V1eXrt/gzJmcyK0b mfV1URQKnPS/rJ3o/SfEQkZGRm5orzFKT2K6/arAEkHFeBsQbTA0VtfU1CBR hZXiKiRtbr7V4BP9V2lpUbe8a9AUJbWsG1qo/ywiOqf3YKN+YbmxGiARWIj0 pAH0VhGJRoYEqpTR0VERTumRA/ditUhUUwa5U6SWddu5JsmamGvdID09XUVi BO3DBO4adIjGUEh/gw899JAvxECQr/i7+xD7feSRRwJdN07rxpFKb9tXrlwx LMLJL4uCWnd1dbWeGcYuoe/3aF0svmJX67bP25Rrgsdzt9vd1taGx/BorRur YEWsHvML4ISkHNLvLtxbZqVDP+ncz+e/88prU/h/YmJCxXgHloOLv8GifauR 3u3t7TIrTWQSYSJCrjpLkTY6PPAm5iidT2pZt+H3FHHOydWOoZQo9vT04CYe Shfn5+exCI+EhnT5hUXlFyEcGBiQIGc9EEJ+k518dYdUqWXddq5JsiamW7fe eYjh7cgjR46oOA08l0mvelBHXJ+h4lgadEO4Jsta0G+1q3AkXMZfeukllU2F WN93332YRvlQcZwnMnRmPNb98ssvy3g3P/zhD1HgK35/HhwczMjIUEca1LrB 6dOncfbKOPLbt2+XxFOnTunl29O64yQR52okNDcNRW7dyGzVfhJiIdKOrX4a llYs1WaFdJfLpWK55SYuI2jo4KaPW7/eLCYsLS3JTR/3a1zbZVraw6XdDA6A iyQWyS/UoRrSyZqkkHUH/p5SUlKSnZ2tblLS/7OKmhBdDGxHlVMo8CVcWT0/ Px8ns3SvIQMwYbtymjU0NOBhU54T1euBihSybpvXJFkT0637ldWeNwKtG1dp +LYyUujx448/rmbvvffeoIMwYkOPPfbYPRqqCxGcSCobzFZFXwfyyCOP4DzR i40trtvAE0888aMf/SiMdQeyYcMGw2Ha1rojHyvHDtaNswvPYiUlbXH2YUKI 44EJ6y3M0p+JPhAVxFuP5V5YWAi8mCDxRrC2Mp+/j19pMZNbtm5oAwMD6rdv SDuulgzoipkUsm75yUP/PeXGq98s8Pl/tQwzK4hD9vf3By7STy2cfioAAxOi iEJNTQ1OXcO6KWTdNq9JsiYxWzeUT4IrAq0bH5kYZuCAOC+99BJ0V8KkFQ89 9FBWVtbU1FSobUGMd+zYYdBXuChOGENOPJ1BYvUR1SH2kP/AY8TNInLrxm5f vXpV7zDwxz/+8eXLl/FEICHlemfgyrorKytFmwXsFTIH9g0uhyY9Lhqgda8J rgO4euCRXPoeiSrChNZNSOLwer1Bb/fQbBg+FrGVLE5SyLqTg7S9BB1bQXrk CPWOfApZd3KIuSbJmsRs3XGCzxTODF/Cs9LLL78cySr4lJF/ZGRkbGws6FA7 il/+8pfSKcri4qJJ+/srfv7znw8PD0PswxerrLuvr+8V/5H29PRgtw3jVMaP tdatjwWffOvGt15+5NJfh4R4P/98Ga2bELIeoHWbRVTWnZ6e/s1vfvOxxx4L XFRaWopFd955Z6pbN0kcVlm3gwnfX7eJrEPrxiMMHnxqa2t1025paUHi/Pw8 lsY/Sg4hhKQEtG6ziMq6/+Ef/uE1fp555hnDomvXriH99ttvp3WTUNC6TccB 1p0dQCi7TqZ1j42NVVZWimwXFha2trZOTEwYjlGs+967nn/8/vwwf8hA6yaE pDQxWzeupXJhb29vp3X7YrXuO+64o6ysjNZNooLWbTq0btMvIwsLC/X19eLb paWl/f39oQaiFeuO8I/WTQhJXWK27traWvHGxx57jNbti9W6wV//9V/Tus1i fn6+x4/+ZrfzoHWbjiGuO3Gsk54DR0ZG5E3JoqIiPMiEGiZPWFxc6u+fjPAP mU3cT0Kcx8rKysDAwMTEBHsasSFOte6FhYX+/v5kDnoYg3V/7GMfkzqEKdnW upNfk7Hh9XovX768YcOGND+YwGzQdzkdAK3bdFCfLS0tbW1tenfiicDx1o0b fVdXlzRx19XV2f/SQYjzwL2vpKQE30GJ6aJ+2wenWrfP39iSnZ1dUVEhQ28k dFu+mKz7/vvv//KXv4yJt771raWlpfa0bl/SazIGlpaWDh8+LLJ9wI/oNxId 2VMKrTt1SYR12wfc3PHkIsqNKzzv9YRYhYi3GAL12z442Lp9q7ooZ12ipTE2 6y4rK3vb296G6S996Uu2tW5fcmsyBoqLi+HY6enpqBA5tZr8HcUgEZcdq/fO fGjdqYuzrVuUu7CwUI1bTQiJDRgyrgbLy8tLfnBxgEgv+llYWJDes2dnZ2f8 TE9PT01NQa4m/IyPj3s8nqGhIXwZ9d+z8vLyampquru7h4eHR4gVWGvdSTjA jo4Ow6+oMDQo2cDAgLkbis26Mb1//36pyd27d8ds3eYeS1CSVpMxgKqDYOMR 5qYGZpG4b98+q6+d5kPrTl0cbN340kkrN273Vu8LITYF5gw9hvT29PS0tLTA gUtKSiDDuX5ycnJiHuWKpASRWzcE5pMaH/3oR8UV3/GOd+jpn/vc5yK3bquP 3nxisG7w93//9xJngm9fbNa9zpF4ElzEdOvGLBI3bdpk9VXWfGjdqYsDrLtz fOZa56Dh79nG7h1X8vF3qqpFT1/+GX/RJuuX5eXl8fFx3IxkJFZIdVS3Nhg4 VhEhh5nn5+e7/MgvSkVFRSizxE9paWlZWVl5eXmFn8rKyqqqqurqakNbt5SJ 1ZETwl9Lkk7k1v21r33tNRFw2223RW7dSThAnHU4XQ1nHVJwruKcNHFDUnJs 1o3zX+JMvvjFL8Zm3SYeSCiSVpMxsGXLFgh2c3Ozbt2oHCRu377d6uuu+dC6 UxcHWPfR+u5PHcmJ8M8btvcSQhzGysrKzMzMwMAA7kcqrFoH8gxDxm2rtbW1 r69vZGRkenra6/VKGAksHVcAs6Kv9V+osV3cEz0ej/2vMM4mcuu+++67f0/j zW9+s2j2G97wBj39ve99b+TWneij098mADjVu7u75+bmErGteKwbHDhwQOrz 4YcftmFcdzJrMgYuXrwIwYZt4iImpxYmMItELLJ678yH1p26OMa6v3Sh5u+f bwnz99kTBbRusk5YXFzs7+8PbJvKzs4uKytraWlxu92w8WS+3S/KnZ+fD/+n bNsHB79NqUQxOYoYp3WDr3zlK+onA1tZd5JrMgbGx8c3b94Mx87IyMjzgwnM IhGLrN4786F1py6Ose5/vNH5r2WeMH+fO11K6ybOZnZ2FvfE8vJy3bQLCwsb Ghr6+vqgOuF7qk8cvb29lG174lTrhijivpZMRYzfulWcia2sO/k1GRs9PT1b t25N08AsEq3er4RA605dxLrr6upeXEVfSusmxOasrKxMTEy0t7cXFRUp087J yYEUud1uOwwSIZ2fWL0XJDhOte7kn3LxWzc4ePCg3aw7hb68CwsLFRUVF/1g ArNW71GiMFj3Sy+9pBRu9+7dtG47I9atMzIyopbSugmxLfh69vb26rKdn5+P uzC+wla1aZOUw6nWnXyisu6vfvWrqLoHHngglJD/9m//tk2sm9gQg3WfO3dO tzhat515+eWXcePu1vjFL36hltK6CbEhU1NTuNuqHkgKCwvb29s56AyJAVq3 WURl3SZC616HGKx7enpatzjcIKzQSWICtG5C7MPy8rLb7S4rK1ON29XV1aOj o5RtEjMxW/fg4OAzfnBC0rp9tG6SRMLEdZOUhtZNiB3AN7GjoyMvL091u9fe 3j4/P2/1fpGUJ2brjhNaN62bxAyt26nQugmxluXl5e7ubuXb5eXlbrebYdvE LGjdZkHrJkmD1u1UaN2EWAW+en19ffn5+XJXraqqmpqasnqniNOgdZsFrZsk DVq3U6F1E5J8VlZW3G63Gjy9vLzckQM9EDtA6zYLWjdJGrRup0LrJiTJjI6O FhcXy520pKQEs1bvEXEytG6zoHWTpEHrdiopZN2Q6i+crwrz95lj+bRuYme8 Xm99fb3cQ4uKigYHB9k5CUk0tG6zoHWTpEHrdiopZN0R/tG6iT1xu93yymRu bm5fX5/9v3fEGdC6zYLWTZIGrduppIR1Ty/+182puQj/2HhI7Mb8/Hx1dbXc Omtqahw8ijGxIbRus6B1k6RB63YqKWHdhKQoeAbs6+uTISbz8/MHBwet3iOy 7qB1mwWtO2Y8Ho/L5Tpz5syePXv27t2blZVVU1Pj9XpjKGpsbMzctrWlpSUb vsxO63YqtG5CEsTc3Fx5ebncMRsaGmK7xRASJzctxeqjN5MblmL10ccIDLmi ouKpp55KC2DXrl0zMzNRFXX16lWsuLi4aNbu9ff340HAhtVL63YqtG5CEsHw 8HBubi4u5gUFBeylhEQIrsY4W6JSkTWhdZsFrTsGXC6XOHZGRkZzczNO746O jmvXrm3duhWJ+/btizzibnl5WYoy0brz8vJQoA2rl9btVGjdhJjLyspKe3u7 3CgbGhqWlpas3iOSGoyMjKgu3HHmmFUsrdssLBNuP1YffSyMj49v2rQJWgvN NpjG4OCgNIBnZmZGKCG0buIAaN2EmIjX662srMQ1PDs7u6+vz+rdISnDxMSE nDMwiu7ubpxCpvxEMjY2Jq8VtLS0xF8aIVFx/vx5iSSBaQQubW1tFYvGCY/Z np6eU6dOFRUV6XlcLhcS8fjmdruzsrIk/4kTJ65fv46lV69exdKhoaFnn30W Wzlw4AAsWv/iXL58GRnw5VIpHo8HKVjR53+MwloocOfOnUjEJhJTDbFA63Yq tG5CzALXdtwjcCXH//p1npA1aWhoqKqqUrPSnXucF2cqN7GW3bt3w2nLy8tD Zdi2bRsyiAPX1tZiGvarZzh+/DgSGxsbOzo69JjwgwcPYukzzzyD6Y0bN+L/ 9PR0aTxHmVNTU7L6jh07kKK/xg6BR8qePXswnZGRoZfZ3t6eiEqIDVq3U6F1 E2IKfX192dnZMJzKykq+OEmipaamprS0VKZx/sCWS0pK8D8SPR5PDAVSuYm1 LC0tic12dnaGynPy5ElkyMrK8q1l3b5gESbKuiUD0k+cOCEx5JIhvHX7GGFC kg6tm5A4wTeooaFBYi/b29v5hSIx4Ha7cf5APPD4Bt92uVxdXV2whbKyMqRH +9MJlZtYDk5akeTh4eFQebKzs1XDdczWfeXKFZV/ampKmr4nJyd9tG5iP2jd hMQDbgQyAk5ubm6Ymwsha9Lf319cXFxRUaFrNq7P+fn5dXV1kZdD5SZ2YHx8 XCQ5zDgFEGZkOHr0qC8O6za8QXP48GEktra2+mjdxH7QugmJGXx9pEdueNH0 9LTVu0NSDI/HA2GYn5/XE0dGRgoKCvSU5uZmqHiEZVK5iU3A5VEkWZw5KCLV Fy5c8MVh3YYQrNOnTyOxpKTER+sm9oPWTUhs4MqPCzsu14WFhXNzc1bvDkkl VlZWJCoJN338r17+8vlbvJGiHuJwfYaEt7W1RVIslZvYiiNHjqiw7UDGx8cl GqSmpsa3at2GzPv27VvTuoeGhgI3Kl8BsW69cxKk07qJhdC6CYkBaLZ0rQzx NrHzWLJO6OnpgUtDraHfs7Oz+iKoBRaVlpbCSWZmZurr610uV9CO1wxQuYnd wNkrnhz42Igz/8yZM1j01FNPyUA5UGvMwpn1PNItSXjrrq6uVqtg0ZYtW5Ao MVqwa9UzoVBcXEzrJhZC6yYkWiBL+fn5uFBXVFREokOEGIBUKxOAS/T390NL 3G43ppEyOTmphsspLy+P5JeUhCo3/P/ChQu5ubkxZyDrluvXr0s3I9BdOb19 /ncepbMRUFlZKYldXV2Y3bBhg2qavnbtmh6jAgmXWdUjt1i3GlkeGZ5//vk0 f//bkiEzMxOzp0+fllmPxyNjYirrxhMuZs+fP5+MuogGWrdToXUTEhUQDBnq vbq6Wt1ECIkKSLV0Dry0tAQbkTgl+elE2v2kDdwQ8h2KRLdySygsDCfmDGTd gjMZT2Riy3DvAwcOyMA0gt7IjO+CdN+9adOmw4cPw5xVNhUZLhk2b94sTeJi 3Wn+PrqxlYMHD8pSFXOCh1bJAM0+dOiQBLTo1g3nl5QdO3bY6kciWrdToXUT Ejkej0f0pqGhgV8cEhWQio6ODjlt2traysrKICRwb5i2BJlMT08XFBRIjGvk JCGwhNZN4mF5eRknuW7R4Pjx44H9eA8ODu7du1cywL0vXbqEExvTzc3NkqGu rk5iToDX6xXrhm/DtCURuq4Xi01nZ2cr2d69e7e0qGMrkgEPudLnCcD3KCn1 ERG0bqdC6yYkQpRy4xYAX7J6d0gqAeWWLgFHRkYwC82GDPT09FRXV+uvek1M TBhergxPcmK5ad0kfnDNnJmZGRoaGh4eDv8uDLJBv/GVCboU6cggI5GpngOh MSg2VCwW3BvbDfO1QoFY11ZXdVq3U6F1ExIJSrlt9SskSQlEufP9dHV1SaL0 VZKXl6f3wID7Pk4zFbYanpiVG35y6dKlzMzMbdu2paenHz9+XDo31kvGjh04 cGD37t0XL16U19x0qV4zg/O4evXq6dOnJycnMYEDR71lZWV1dHToeWBu169f z8jIQMUeOnTo2rVr+quyzz//PErAoxYy7NmzBx8cpk+dOlVXV1dfX3/y5Mnt 27fjs5AfO7q7u8+ePYut7Nu3D1UdSkHXOUH763YGtG6nQusmZE2o3CRmlHJL hyRALcLphJOqtLRUXYTHx8eRIq+GhSdm5Ub+DRs2yE/q6sd6UFtbKxmmp6eh f4EZlFSvmcGRiOCpA1eo+sdV4umnn5ZE6UZD4o1VZ9F4QpGoYxVILE8rKv5B gWciQ6Ktgh/sA62bpBy0bkLCww7ZSMzoyu3zN2AWFRWppSsrK0iRlyjdbvfA wEBhYaEKYQ1DPOfkgQMH0vy9OkxNTWEHsF2Ja83MzJQMhw4dSvMPFzg7O4sM eEwQA1RSvWYGRyKCJ31x4LkDz0f79+9HyrFjx3z+j1IqFv/j0/H5H6AyMjLS /MOdy01WrDvNP4BLdXU1lLupqUlSjhw5AmnHKihNUrBWR0fH4uKijN6Ynp5u 8fHbEtQSPhF8cazeEfOhdTsVWjchYaByk3hob29Xyu3zn076CDgCdKu2tjY7 OxvK3dnZueYFOZ5zcn5+/ty5c9DmyclJlSidKkuvDlDxNH/vbXoG6Y1NpHrN DE5FrPv69esqRcZbgUv7/COKii3r0UETExPyPNLT0+NbtW7UlcqgrFudIaqD a4n/9/lPD0lhj0nrClq3U6F1ExIKKjeJE5iS/oYXZvPy8uBagTkjfJPLrHMS zgzDLy8vv3LliowkIl0cd3R0SEOrnrm3t1dJ9ZoZnIpYt/7Zud1uiTnBdFVV VWC1gKNHj6atdkkt1l1XV6eWinXv2LFDpQwNDUncjkpZXFwU6+bQAOsKWrdT oXUTEhQqN0kEsFacV7G9HGfKOQmvU12l6Yh1l5aWpvl7ddNXkeZWkeo1MziV wBBiad8W6y4pKcH0yZMnDWtJV9UyfpBYt16CWLc+GuPw8LAhngT3aFr3OoTW 7VRo3YQEQuUmCWJhYSE7O7u3tzfaFU05J+fn52WcEbjic889V1VV1d/f39ra qqxbPBBarq8lDbAi1WtmcCrhrbu6utrgzwI8HOnFxcW+VetGhaulUpkSGS4E WrfX66V1r0No3U6F1k2IASo3SSgNDQ25ubkRjjspmHVOSijy5s2b9R7tpPla TG98fFwyyBCZQk1NjZLqNTM4lfDW3dPTk+bvukSvFgizjD8uQyvSuknk0Lqd Cq2bEB0qN0k08CiXy1VRUZHkWG5QW1ub5h/1T0WbQ/Okvzv4oc8fXg7lw+yV K1dk92ZmZmQIb5HqNTM4lfDWjc9UquXSpUtyS0XlXL16VWpbnrBo3SRyaN1O hdZNiILKTUzE7XaHatD2eDy5ubmhhtLTMfecHB8fl141du7cefny5aysLL3D bRkxcGBgAKIo7wZeuHBBIlJ0qV4zgyMJb90+rUOSvXv3Xrx4UfoVRG2rkXRo 3SRyaN1OhdZNiEDlJuaC00kf7d1A+EGxhUSckzA95ckSoT04OChCqLbS3t4u nXJLMElRUREm9u3bpwpZM4PzwNEZnHl0dNRgyJ2dnapT7g0bNmRmZqKi1FLp K0Y/JSTg58SJEypFTF7v1US9TcnhKdcVtG6nQusmxEflJgkgvHWvSeLOyeXl ZY/HA9n2er1hss3MzAwPD4e5QayZYX0yNzeHzz183RISHlq3U6F1E0LlJokg HuvGOZmdnc1zkpD1Ca3bqdC6yTqHyk1MpL+/P28VnFQ4tWS6oqIi8kKo3ISs c2jdToXWTdYzVG5iLtPT0zdv3uzzA3NubGy86WdwcDDCEmTUeJ6ThKxnaN1O hdZN1i1UbmIKuIQGDeKFdUcu28LQ0BCVmxBC63YqtG6yPqFyE1OYmZkpKiqC YDc1NRn6347KunEprq+vF+VGUQnYU0JIykDrdiq0brIOoXITs6ioqIAkDwwM wLE7Ozv1RYWFhcPDw2uWgItwR0eHnJCgtLQ0YTtLCEkNaN1OhdZN1htUbmIK Xq93cXExLy9veXkZs93d3TippqamIi9BfDs3N/fGKlB3DoZCCKF1OxVaN1lX ULmJWVT7UdEguJCWlpaWlJREckUN9G1BjWNICFnP0LqdCq2brB+o3MRERkdH cS5BvFXK9PR0dnZ2V1dXmLVC+TZAIhu6CSE+WrdzoXWTdQKVm5hOQ0MDNHtu bk6ldHZ2ImV2djbUKl6vd34Vt9vNhm5CSCC0bqdC6ybrAQ47QhIBrp/5+fl1 dXUqBZfTpqampaWlNddV52RFRQUbugkhOrRup0LrJo6Hyk0Sx/DwME6tkZGR qNZS52Rra2tPTw8bugkhOrRup0LrJs6Gyk0STW1trcvlirylWlduzMK62dBN CNGhdTsVWjdxMAa9ISQRSP+Bhs66QxF4TsK62dBNCNGhdTsVWjdxKlRukjQ8 Hk9Usdz6Odnf3y89fhNCiEDrdiq0buJIqNzEboQ6J6nchBADtG6nQusmzoPK TewGz0lCSOTQup0KrZs4DOoNsRs8JwkhUUHrdiq0buIkqDfEbvCcJIREC63b qdC6iWOg3hC7wXOSEBIDtG6nQusmzoB6Q+wGz0lCSGzQup0KrZs4AOoNsRs8 JwkhMUPrdiq0bpLqUG+I3eA5SQiJB1q3U6F1k5SGekPsBs9JQkic0LqdCq2b pC7UG2I3eE4SQuKH1u1UaN0kRaHeELvBc5IQYgq0bqdC6yapCPWG2A2ek4QQ s6B1OxVaN0k5qDfEbvCcJISYCK3bqdC6SWpBvSF2Y3R0lOckIcREaN1OhdZN UggqN7EbVG5CiOnQup0KrZukClRuYjeo3ISQREDrdiq0bpISULmJ3aByE0IS BK3bqdC6if2xj3I3aIRKT5WlJB6o3ISQxEHrdiq0bmJz7KPcIE0jVHqqLCUx Q+UmhCQUWrdToXUTO2OW3iwvLwc2/DY2Nvb29kqGpaWl9vb2ysrKqqqq/v5+ fC/UutPT0/X19aWlpVhqN3OmdScfKjchJNHQup0KrZvYFhP1BlJ9Ixi1tbWy tKioCLMFBQV5eXmYgHvLilBu2YfCwsKcnBy7mTOtO8lQuQkhSYDW7VRo3cSe mK43Sxrz8/O5ubkofHJyEou6u7sx3djY6PO3ildXV2N2YmICs83NzZju7OzE NL4soQz22QDsv5REC5WbEJIcaN1OhdZNbEhQvenv74cYz87OIrG8vLylpWVu bi628uvr65VLA2nihm/L7MzMTG9vLzaE6bq6OuQcGxuTRWw3XrdQuQkhSYPW 7VRo3cRuhNKbsrIyJEKPsVQCQiQ+BMzPz2OpNFwDCDO0PJSTDw8PY13klzMf so3ZioqKwcFB2Hh1dXVXV5cycKg+lqI0iHdfX9/evXv37dvHduP1BpWbEJJM aN1OhdZNbEUovcFZKunNzc2Li4tIcblcubm5shTCjEWYnZqampmZyc/PxyyE OegmxN49Ho/MQs4D472Vk6+srEDI9UVqRcVNW2Lu57KeoXITQpIMrdup0LqJ fQijN5OTk0gvKSmBBvv8MpyTk4NZlaG3t1fEW16H7OrqCrqJ6elpLC0uLpZy VMmgpaUFXwdIe2lpKWYHBgawtL29XRrYm5qaysvLMV1UVCTar7Dar4Ojdo/9 dccDlZsQknxo3U6F1k1sQni96evrw6Lu7m6ZFXk2aCS0WfxZ3osMiuRBaSpl fn4eKXD4paUlSZGokubmZph5fn4+9kppduDqPttbN2PRY4bKTQixBFq3U6F1 Ezuwpt5AsLEUsi2zIsa6/arAEmmanpqaCiwEFo1FumD7VuO6XS6XShkbGxN1 n52dlWgTtai0tPTkyZOXL1/WhV8sd9I20LpNgcpNCLEKWrdToXUTy4lEb4qK imDL6lxtampSnfv5/E3fEljS3d3d09MTSrylWbuystKQLpHeKr80aEPslZCr QXOCGiytOxEk/9cBHSo3IcRCaN1OhdZNrCUSvcFZKt2MqJSSkhKspXoakW5J VPyJiPfg4KChHGnEDtzQyMgI0vPz87u6uqRTQZi2RJU0NjZitrS0tLe3t6Gh gdats7S0FDjiJ1C/QeCDa2trwwdnGPFzZWUFs0jE51hbW6s6ZtSxzrh/BZWb EGIhtG6nQusmFhJhi6LH40Ee+JvMShM0TFjPY+gnUHrbNiBvXAbt22RgYEDG zZF3NlUoC8RSBsoRaN068jQUCJ5cZGlhYeEN/7Ce8ktEdXW1rNjR0YFZfPTF xcWyCp6bDIVbZ9y/gspNCLEQWrdToXUTq4jnR/wVPzFsNMxaWAR1N/RPIsDz Z2ZmFhYWUtG6Ezo2JWpGjfiJ2svJycFnKrE6XV1d+HCbmpokW1VV1Q3/YKCo Z3nAkScj+Z1C/yHD2lqV7VK5CSEWQut2KrRuYgkpGjebitYdHuRsbm6GMLe0 tJSVlcUz4mdtba0e5+NyufLz8/URP/v6+lA4UvDRQ7zlCcjr9Up/jIE7ZqF1 x1YDhBBiCrRup0LrJsknRZXbF6LdOKWtWzonDzXiJ5aql0ynp6eh5aGcfGho SJqsxaUlCqiystLtdgeO+Il0+fQ9Hk9dXZ3002gokNZNCFm30LqdCq2bJJnU Ve5QpK51S7OzjBDk9Xqlf3IYuCyFMN9YHfFT9RIjgwcFIvY+Pj4us9LpogE1 4qdhPNDCwkL1oqXltUrrJoRYDq3bqdC6STJJgnLDHicmJmKL+o6NNf2wsOVC ftPZ8H8lbZcT5I1hxqZERcl7qVJduBTg09FfU5XeYHL9YAKzQWsAWi5voRpK lg/aMOIntiJvWSI/bL+goOBGsIGNaN2EkHULrdup0LpJ0khaK3dfX5/L5cJW kqPfa/rhvaf/rx+c/NPwf49d+MsEeWOYPkxkxE/l0kFH/FRduMh7kUGRPPpG pTU76Iif0iON6jUd1x98WEgxvMdK6yaErFto3U6F1k2SQ5IDS8QnpedtE/U7 aLtxBNb9fz5++fMHax4I9ffIxb+wxLplxM+ZmRmZDRzxUwWWqFCTwDqRPklA +BE/cQKIukvfJr29vWqRSPvQ0JBeLK2bELJuoXU7FVo3SQJKuVWf2xECo4PL eb3ehYWF+SiRTqEVMEAo3/Dw8NzcXLRFKWLrwwTW/eTVv85q+2moPzi5JdYt I36q5xEZ8RMlyKxS7p6eHunqPKh4y4ifVVVVhnQJKVH5Ra0HBgYkXFwPKZG3 ONV2I6zVBEHrJoRYDq3bqdC6SaKJULnhfrOzs0NDQ52dnXV1dWVlZfBkWdE+ 6AaL3cNOYldNse77sz594sSJs2fPXr16taKiApZrljeGsu5QI36qC4L0pK3i T0S8DS3SvtVG7MCfMGR1GfFTOipBjeEBCtuVkJKGhgY8GaEOb2gvWipo3YSQ dQut26nQuklCWVO5YdrwupqaGhXJYCA3NxfmVlBQUBglWCUnJydomdl+sBQb VfmxIUnE5vRyoIgy+Av+1w1WlWaKdT9y7v+BgkJE4Z9XrlzZv3//oUOHoMEe jydObwxl3UFH/MTW9TyGfgKDdhsYfsRP9bGWlpaqUBZMiGwL+PQXFhYM69K6 CSHrFlq3U6F1k8QRRrmnp6fb29sltEABT66tre3s7BwaGoL/wMRiPjlXVlYa Gxt1zQ6q3xKe7fV6pScN7I9YIjxQylF9cRQXFxusG7uHncSuJijCBBJ+4cKF jIwMPJXE440JHZsykg8Cro4aDlwkfZvo0eA6tG5CyLqF1u1UaN0kQQRVbihW X1+fRPwK+fn5TU1Ng4ODga2dMQPTk/cEsQPw54GBAVjfsoYau1xaXyUCvKWl xaeNXS4ByVKOvPqHL0vy47pLSkqwLdTPevPGqKz7xIkT39e46667fvjDH27f vh1nWuT15qTaI4SkNLRup0LrJokgULkh1e3t7dLt8w3/YIjRdi0CXW9ubp6d ncX/cFEZxzwwGwqExkO23W534NgrApbe0LrRwM64XC71RYCKw7vm5+dVTjX4 S0ZGxt69e8+ePRvV2JRxvk2JSsvMzMQ+rCtvjMq6/+3f/u01IfjCF76AR6cI y3FM7RFCUhpat1OhdRPTMSg33BgarGI8Kisrh4eHYzjrYNoS5p2Tk1NcXIzp uro6WYRNqLHLcUpDUDEbauxy6UMDuyHCv7S0dMPfBcfAwEBtbS0m9LHLpQfC iooKHFR3d/cN/yuBhieFBFl3W9uA+jt4MCs/v1JPwZ+zvTEG6/7Upz61x8/O nTs3bNjwta997ZZbbkH6Rz7ykbGxsQiLckbtEUJSGlq3U6F1kziZnp5u84MJ 36uVe3Z2Vg+ubmhokDwxIG58Y3WsQ3hvXl5efn6+LIUwS/v5lB+JzR4cHAxa lNj7xMSEzOpjl6tHA9WlBv7X3/vTV1Qkwro9nok7v3ch/F+E3hhmbEo7E4N1 33nnnYb0I0eO3HbbbVj08MMPR1gUrZsQYjm0bqdC6yYxs7CwcP78eT3UOSsr 6+rVqyLY9fX1SmWbmppCtTxHyPj4uMiwPna53uGGYexyfagXHWiV9KehUtT7 knhMgNtDwkXLYfJYCsmX4POWlpaKigp54zLaURRh3Q+e/9SmvK+F+rv/3P8O at2P3+/a+kR90L/H78+P0BvD9NcdOfise3t74/wQo8IU6wZ79uzBIrh3hM3d tG5CiOXQup0KrZvEBs6cAwcOQOQ2bdp02A8mMLtt2zbpjVl8u7m5WQKk40S6 p1OR2FNTUzdePdKKTxu7HBOhypGBYPRu7qStWx+7HNIlheCrkZeXh6NQXXBc u3bt5MmTkPAox6aMekR4se6fPFx6YGNP0L//fLgkQm80xboBrBW1hAeW7u7u JOi3WdY9NDT0pje9CUvLy8sjKYrWTQixHFq3U6F1k9goLCyExe3cubOlpUVE BROYRWJmZib0rLW11dAmHA/Scq46fBYx1u3IMHb55KsHOhTU2OUqbNu32k91 QUGBSpFhX2Dd2NyNV48jE1sfJmOekbX/xkf1VWxo3b5V8ZZKTrR+m2Xd4EMf +hCWHjlyJJKiaN2EEMuhdTsVWjeJjb1798LiYKQ3NTCLxB07dpjo24Jh7HIJ F1ejjSvl7vUTSrwjHLtc2sPdbrcEk7tcLtUXSmzWHQPxW/fgKvo+D8ZNe3u7 ofNzPLDU1dX19PTEX7iOidb9yU9+EkufeOKJSIqidRNCLIfW7VRo3SQ2Nm7c CIuDa+nWjVkkYpG521pz7PKhoSE9/kQ6HhkeHjaUE8nY5bW1tWKSElUieo/N 4dDglilk3cqK9X2+kTqYaN3vf//7sXT79u2RFEXrJoRYDq3bqdC6SWykp6fD 4iClunVjVkK7zd2WRHGrrr+lA5Py8nI9jyHUIWgwuXR1Iq9JGnC73frY5bOz s5K+tLTU3NysWndTyLpVvyWZGg1xg6cS1em6Ag8sZWVlWBR/+QoTrfsNb3gD ll64ELLXl8Dai/OMJYSQeKB1OxVaN4mNS5cuwTyPHj3a19cnooIJzCJx3759 UOKxsTGr9zFq4OpBxy7HdwQejkUpZN2JqJ/FxcWioiIl2xUVFfjQTY8mEsyy buynDJeDR8JIiqJ1E0Ish9btVGjdJDbgJ1u2bBHHvu4HE5jdvHmzdB54w9/R 38jISOSjT9qfZzVU4pp++HRhQyR/4xMTahUbWrdS7oTKtsIU656YmPjTP/1T LPrMZz4TYVG0bkKI5dC6nQqtm8RMf3+/xJkoMIvE5eXl3t5e1X9gaWnp8PCw k9zbwJp++KkjOfj7zNHcUH+SwTNutO4nHyra/ZO2oH9PPFiUTG+EYzc1NSVB thVxWjfWzc3N/exnPysN3VeuXImwKFo3IcRyaN1OhdZN4sHr9dbV1T3vBxN6 eAbcG5JWUFAg7l1cXOx2ux15skVi3V+6UPOvZZ5Qf196tiaoddtnbMrkPzTF YN233Xbbm/1IB93CLbfcsmvXrgjLCaw9QghJPrRup0LrJgkFZ1d/f39hYaG4 t8vl6unpUR3xJQIb+mEM1j0xMVFc3CF/LlfL44/vuHy5QqXIX4TeaGJ/3ckk Kuv+93//99e8mttvv/1P/uRPvvGNb0gfkpFD6yaEWA6t26nQukkSwDnmdrtl pPUb/jEr6+rqRkdHE3Hueb3e5uZmiFMiYiGCthsnwroVfX19O3fuLC4ujtkb 14N1mwitmxBiObRup0LrJslkbGyspqZGdYKRl5cHfR0aGjK39Ru+DU1NxHt/ sfVhElOEiaexsfHixYt79+4tLS2Nxxtp3fHUHiGEJB9at1OhdZPkAw3u6elR Td8q8Lu5ubm/v39iYiJ+Cfd6vSLeeh93CwsLcRYbaLDYVbOs+0Z2zvXr18+d O5eZmblr165Tp06VlZWNjo6GKTkSb6R1x1N7hBCSfGjdToXWTSxkbm6ut7e3 qqoqJyfHMPBKXl4etBmL6urqGhsbIeQtUYK11MA3CpfLBQNvaGiItjRBN1js npRvlnXnu1zQ7Pr6ejySjI+Pm+WNtO54ao8QQpIPrdup0LqJHcBJCOGBgUOV YZ6BEm4TAkdXx64mNK47fm8M2se4/aF1E0LWLbRup0LrJvYEZ+bMzMzY2Njw 8LDb7e7v778ZJW1tbQZ7d7lc1dXVHR0d0Ral0K0buyeRMDa37hSF1k0IWbfQ up0KrZs4EuhTbm6uyHZxcXFnZyckOf5iYxubktYdA7RuQsi6hdbtVGjdxHlM TU1BuYuKijo6OkyR7fBEYt2fOZr7FycKQv3J8JS0bh1aNyFk3ULrdiq0buIw vF5vT0/P9PR00rZolR+GwuCNpoxNmXxo3YSQdQut26nQugmJE5tbN/swiaf2 CCEk+dC6nQqtm4Dh4eELFy5cvXo1ydsdGxtL/gDu8RDb2JRJhtZtYu0RQkjy oXU7FVo3Ae3t7VCyrVu3Jm2LkG1IPjaaiHHbE0dsY1MmGSdZt1VYffSEkHUN rdup0LqJzwrrXl5eFg+kdZuOwRtp3TFg9dETQtY1tG6nQusmPlp3xNC6k4Z1 xv2q2iOEkORD63YqtG7iC7BuWMepU6fq6uqampowkZ6efuDAgevXrytDlgw1 NTUVFRVHjhxBhnPnziG/Opd6enqQoaioSN+Ky+VCItZ1u91ZWVnigSdOnEDJ yTzYeEhF607RsSkJIWTdQut2KrRu4guwbsg2Zp966qm0VwP3lpcfJYOwadOm HTt2yHRubq6UUFtbi1k4tr6V48ePI7GxsbGjo0Mv9uDBg0k+3pgJY912w8Ja IoQQEg+0bqdC6ya+ENYNjhw50tnZ6fV6KyoqJKWvr8+QYX5+HiltbW0bN25E CoryrWXdvpSNMAkzNqXdsLCWCCGExAOt26nQuokvhHVv2rRJH9hx3759SJQe 8yQDNHtyclJlkD5JnnvuOZ9zrTsoVvt1cKyuFUIIITFC63YqtG7iC2Hdzzzz jJ5HIrGrq6tVhmPHjukZOjs7JQrFt86s2+ak6NiUhBCybqF1OxVaN/GFsG6D VJ89exaJVVVVKsP58+f1DENDQ9JC7qN124kU7cOEEELWLbRup0LrJr5YrfvM mTN6hu7ubiTu27fPt2rdWVlZegaJUUlp624IwP5Lad2EEJJa0LqdCq2b+GK1 7p07d+onj8vlQuLFixcxDbWWdy3V0pWVFekUJaWtOy2A1FpqTi0QQghJJLRu p0LrJr5YrRuoHrnHxsaefvpppNTV1WG2q6sL0xs2bHC73ZLh2rVrsopYNyRc ZkdHR5N2mPFjB3N2pHWPj49fuHBB9TxpZ5aWlrC3Vu8FIcTJ0LqdCq2b+OKw bnD06NHz589v3rwZ0xAnyQwz2bZtW5o/zPvw4cM7d+5U+cW6gWTAinqTuM2x gzk70rpv3ryZFvACrw3p7+/fs2fPjRs3rN4RQoiToXU7FVo38a12P7J9+3aZ bWlpSfOPGqnngVojsaamxqd1cgJhFp2DQl++fBmyrfIPDg7u3btXlsK9L126 BFfBdHNzs2Soq6tTA/F4vd5kHWtcPBtAai01pxYSQKpYd15eHvaT1k0ISSi0 bqdC6yYxoDeGz87ODg0NyZiVgczMzEC/dRvXQToypIpyk8RB6yaEEAWt26nQ ukkMBA1BWbcEBvqOjY2FegwxZRNwvwsXLng8HhM3kWRQRTiKAwcO7N69++LF i/L6rbLu559//vTp0z09PRkZGRLRgSuVzz8EalZW1q5du3bu3Hnq1Cn1u4nP 7+1IqampqaioOHLkSHp6+rlz5+rq6gzXt7m5uevXr6PYbdu2HTp06Nq1a3hs VEsvX76MQiYmJlQKKhkpV69exTR2A5tO879HjET1zgIhhJgLrdup0LpJDNC6 FYZAX8i2jNFpYt8sgbHEErrT29tr1iaSzPT09Pbt2yW4SEUZ6dYNFU/zx/xL +o4dO1TFSsASkOlLly7JKvq7BliKVWRaf0MTCi3v/IItW7ao4KjBwUHJIGup Wd9qIzzqH9NwdT1Ivr29PUn1RQhZZ9C6nQqtm8RAS0vLxo0bDd1xr08MIQeJ 6BExMKrh2rVrp06dSq3uX3QOHTqU5n8Pd3Z2FjpdX1+P0ynQukFJSUl1dXVj Y2Nra6tKQSXjqlVRUbFhwwakYJFPs+4jR47Mz8/7/A3jUqzoMTZ04MCBNP/w qWNjYz5/xyki0gcPHpTLYHjr9jHChBCSFGjdToXWTUg8WGLdKc3U1FSav1fJ yclJlfj8888HWjcSVQbpQscwHOrly5fTVruFF+uGZuvFSvP4c889h+mRkRH5 aPSnlYmJCTHznp4eH62bEGIPaN1OhdZNSHiGh4cvXbqUmZm5bdu29PT048eP S+OqLyDQt66uLisrS9TuxIkT169fl2xerxe2BjlECRkZGUiXxlgBZnj69Gm4 IiYOHDiATaCQjo6OoJuQWGJp65YGW99ascoS8Ix9g5diAuVjK8hv1fhEODRp XtYTe3t7A61b+n4XJCLFENQha+GofSGinqRzHhwvpquqqgK3C44ePYr0yspK H62bEGIPaN1OhdZNSBhaWlokjMEQgVxbW+sLCPTNz8/XZ0Xwpqam4GySsnXr VpmA+ipnhmqm+bttTHs12HTgJkQ79bjuNWOVRUf1nRfgoua+8hkhpaWl2Doe XvREHEWgdff19akMEsg9MDAQuBZYWlqSwzQ0hg8NDaX5w7wxXVJSgumTJ08a 9ufChQtpq+HftG5CiB2gdTsVWjchYZBI4NOnT0Oe4ahut/vw4cNIyczMlAxr RphIP+fQNglsmJmZkU7OVXfoYt0bN24sLi6enp4eHx/fv3+/3mwbaHrKuiOJ VdYDnjs7O71eb0VFhaToWps0ZH9QjXqi6LHBuvv7+1UGPKekrcaBKJBBqk4V e+bMGT1Dd3c3Evft24fp6urqtNVwFB14ONJR+b5V69Y7J5G+62ndhJBkQut2 KrRuQkIxPz9/7ty5Q4cO6aHC9fX1YTTMYN1YUWaxlirBEGAs1q3CUXyrpgfz DLoJn2bdkcQqi45u2rQJwq/yQESR2NDQYFplRQyeC9L8/ZMsLCyoxJqamvDW LY8qLpdLLwqqrD4LOcydO3fqFzTkR+LFixcxjdqQnwP07eIZRH6AkCFT5VcJ uHrQTfho3YSQpEDrdiq0bkLWBPLc2dlZXl5+5coVETPYnSwKb91dXV0ym5OT k68hiRIfLtYNaVSbc7vdadpQoWGsO5JYZTWQqJ5B4s+rq6vNrahIWFlZkYZr VKaEuOBxQGLXw1h3WVmZuPrQ0JCk4EFDhBl169Oa9IuKiiTD2NiYxN5IfDgE W7Z76dIlueip3gjxSCKR9pmZmfLThpTg8XhkE8q6CwoKAuNYCCHEXGjdToXW TUgY4HgSUmIgQuuWqIZQQON9q9atB3tIC3Yk1h1JrHLQ1wylSxBIu7nVFSED AwMSp43nBezqtm3bpELCWDcqVgJvsOKZM2ew/xKpfvz4cRlAR++vGw8dEGPp 7hvlq0JUHlTgxYsXpcCNGzeqd1fxiUgGaPahQ4fkJwPduvEgIyk7duyQwHtC CDEdWrdToXUTEor5+XkRQgjwc889B0eFB0rH0RFat2je1q1be4IxNTXli8+6 I4lVtqF1g/b2dum1W1qwi4qK0lYDsH2rkR6GwR9RpZcuXVImnJ6ejllRbp/W pC+xKGn+V0ovX768tLSkF9LZ2ak6A9+wYUNmZqbeLwo+vuzsbLUJ5JRfK1Dh kmFhYUE9hUkbOyGEmA6t26nQugkJhcRXwwn1jvikCw4on8yGt255SRB2p8dU wwOvXr2KVYaHh33xWXckscr2tG4B1YJKiOoShMyjo6Mej8eQrh8mPi/UfJge Wubm5qD0qKigS/EhYnV5Jgq12yjBkh5gCCHrAVq3U6F1ExKK2tpaCWmAYkkK FFFChaG1kmII9IWJiXXL643wN2m2PX36tGqVzc3NlZZYKXZN6w6MJVbWHUms sp2t20SCHiYhhKQitG6nQusmJBTj4+MSbLBz587Lly9nZWXpvV5La3ZgoK8E pWzevFkCP2RQGLBr165nn31WHDtN6z9kTesO3ITeX/eascq0bkIISS1o3U6F 1k1IGOBy6l2/NH8v04ODgxIYLAIcGOhbV1en5FxiGGDU0lOfsGXLlsLCQrUJ WaS/OTg6OqoHsQRuQtT65s2bkiF8rLLEyajuwQXpRbympiYhtWYFOEw8a+DJ yOodIYSQeKF1OxVaNyHhWV5e9ng8kO1QYcC+gEDfpaUlpBjyQ57dbvf4+DgK jGE31owlDh+rTAghJFWgdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjch hBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsV WjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2g dTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBC iH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjch hBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsV WjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2gdTsVWjchhBBCiH2g dTsVWjchhBBCiH2gdTsVWredmSPExlj9/SCEEGdC63YqtG47Y7VVEUIIISRe or3707qdCq3bzlh9nSCEEEJIvER796d1OxVat52x+jphAjMzM88+++zu3buL ioqs3hcSKV1+ent7rd4RQghxAtHe/WndToXWbWesvk6YQH19fZqfjRs3ejwe q3cnIpqbm0/4aWpqMizq6emRRdeuXTN3oxMTE4cPH77rrru+/OUv33nnnQcO HKiurjZ3ExHS3d39Gj+/9Vu/ZckOEEKIw4j27k/rdiq0bjsT7ff6ueeeO+Nn YGBAT29paZH0/Px8864iEW0dCifWvXXr1qmpqcRt3UTS09NFO//oj/5odHRU X3T+/HlZ9LGPfSyqMgsLC+/ys3fv3sClRUVFb33rW18TwDe/+c3kP6rQugkh xFyivfvTup0KrdvORPu9fvrpp0Vx29ra9HRInaQfOXLEvKtIpFtvaGi4ceNG CoUrKOsG3/rWt/RFMVv3M888Iyt+/vOfNyxC/dxxxx2Byi18+MMfHh4ejveQ ooHWTQgh5hLt3Z/W7VRo3XbG8LWdmZm5efNmc3MztDZoE2jM1o2SOzs7sdbk 5KRh0cTExJgf5MHs6OhoY2MjFHp6ejqSrc/OzsrqgTuMRT09PS0tLePj44G7 FPl2TUe3bnD8+HG1KLx1o/bKyspGRkYCF4Wy7r6+vne9612y6Pbbb3/yySdL S0tdLtd3vvMdtQPbt28Ptautra2QZD0F9YNEfOJDQ0NrHmm/H0NioHVD+2tr a1PlpwpCCLEb0d79ad1OhdZtZ/TvbFVV1Y4dO9I09u7dCwmXpZcuXdqwYYO+ FLOQNxhgYPpTTz2likWGEydObN68WUVfZ2Rk6O3SR48elUXl5eWHDx9W5eza tUv5XqitYxFkXlKwCVUmXPrkyZNbtmxR+ffv319XV6cfbyTbFSDtEHJYujkX xwDrhnyq54hQ1n3mzJlPf/rTb3jDG7Do1ltv/ehHP4oPSxZBWd/+9rf/r//1 v2TF2267DbMf/OAHZSmOSG0Ihetl3nXXXZKuMkPpb/XzZ3/2Z8eOHXvPe96D pX/4h38oS/Fcg1Xe9KY3qQJ///d/Xw9o0VfHp/OhD31IsmECu6Gy6daNR7yP f/zjr33tazH7ute97r777qN7E0JItER796d1OxVat51RX1hYn0FrlcpKy+qz zz4buDQvLw/KF5iOoqRY+Oq+ffsCM2zfvt3tdkueI0eOqG0Zsu3evVtankNt fU6zbqX62CgcOzA/hB/yrA45ku3O+Z1206ZNks2sEGiDdYNPfOITssWg1o2H DhhpYHCIiPfg4GDgItWMDAGWlL/7u78z7MbAwMA+P6gu2XpJSYlkvuOOO37j N35DppV1f/Ob3wzcEFDPO2r1t73tbbfffnvQvZ3TrPv1r3/9+9//fkM2fEym VDIhhKwfor3707qdCq3bzqgv7N69e0U4T5w40dDQUFRUJKoJKioqkAG+mpOT AyOSxHPnzmG2o6NjcnISE6rdGD6JWfHhOb9DSjqMq7a2FoUoH8YqkkfZryRe u3Zt586dKgVSHWbrc8GsW20U8n/16lUcy549eyQFB6XMOZLtAhy+SjS0lseM sm7ItrLTxx57bC6Ydff09LzxjW+UxLvvvvvGjRs4rl/7tV/D7C233FJTUzMx MfHoo4/+5V/+peR573vfi9mf/OQnsvpb3vIWSf/pT3+65o4pbVb78N3vfhcb xSLUjyS+6U1vQkWdOXPmC1/4gqR8+MMfDlz9ta997de//vUHH3zwD/7gDyQF Dw7SpK+sG/z6r//6t7/97X/+539Wkv+e97zHlEomhJD1Q7R3f1q3U6F12xn1 he3t7e3wo9p4lbteunRJZYs2rluFrCiJVW3jmzdvnp2dndPs99ChQ5IyODio RFfv3S7o1gOtW20UZi4pbrcbSyVRddYX4XZHRkZk3S1btpje1v2Vr3zl5MmT Mg2Rzs7ODrTuU6dOScrf/M3fqBK+//3vSyIeKCQlaFw3Pk3lt/BklY48//vV yAOFrs0GSx8YGMjxoyoQjwOSE/IvkfP66ur3DjwxqaAUHOzcq637wIEDhv2/ 9dZbGWRCCCFREe3dn9btVGjddkb/zsLQIEhw7NOnTx89enT79u3in88995zK E5V1j46OKondr6ESJWhc2W9ubq5ad/fu3ZJYXl4efusG64YkG8oXVOS2UvHI tzsxMdHa2hr4HmjM6NaN2W9/+9sy+853vhOPAAbrfuCBByTlXe961+dW+eM/ /mNJ/M53viPZQr1NqUJTDh48qBL12GyhoKBgTtPm2267LegjBqoXNXnvvfdi n7/85S+r1SUMSbduvV+UL37xi5L46KOPzr3aulU2fI5B1yWEELIm0d79ad1O hdZtZ9QXtrKyUjmtgZitG3mCFqiQVlNlv/rgklDE2Kxb36j+/uOZM2ck8ezZ s5IS+XZNx2DdY2Nj73vf+yTl937v9wzWrQI5gvLJT35SsoWy7ne84x2Sfs89 96hEOPNX/ahyGhoa5jRtfvvb3x6429/97ndVHIgBg3W/+c1v1ldULfP/9E// NKdZ96233qry4KGG1k0IIbER7d2f1u1UaN12Rr6tsD4VMp2RkZGXlwcJV6Ha MVu33uxcX1/fGIC0pppr3TgWtdG+vj6V7dChQ5KoWrbtY91z/t4/brvtNt1j lXXffffdkvKNb3zjbADSRj0X2rqVtH/gAx8w9KAI05ZFb3nLWyQljHUfOHBA Ft1+++3f+c53du7ciTMklHW/JrK2blo3IYSYQrR3f1q3U6F12xn5tjY3N4tq btiwQUJqZ2dnod9hrBtarn/flXVDxvR06KWkq85DUHJNTU2lH4nZiMG69a0H xnWrlyKvXr0qKdBv9ViBg5XEyLcLCayqqgraS3ZsBFo32Lx5c1Drxn5Kyl/9 1V+pzF1dXZf8qPhzZd0f+chH9G0dPnxYlanCUeb8YTN/+7d/K+kwc0kMY91/ /ud/LotUHyO6Ywdat8qGCv/N3/xNSZSeyWndhBBiLtHe/WndToXWbWfk23rz 5k3VPgzHhtNmZWXpKep7vWvXLkmEN8Jp1TuSWEXlx7rSkzY4d+6cJG7ZsqWw sBC+ffr0aUnZvn27GH7k9ht064HWrXczeOHCBeRU71fCbFV7b4TbHRwclD4V oZGG0dtjJqh1g8997nOB1t3W1qZis++8886cnJwTJ068973vlRTItmRDtat1 kQ0PTZI+MzOjB2C/733vu9vPBz7wAZV46NAhyRzGur/+9a/Loo9+9KP4WI8d O/bud787jHW/9rWv/da3vvXggw+qbEhpaWmZo3UTQojZRHv3p3U7FVq3nVFf WPUWYSC6dUNi9UWqh8CBgQHVmJym9dc9MTERtOtsXWsjt+6gWw+0bvibiifR 2bRpkx6aEuF29QcK03sONFh3b2/v7/zO7xisGxw/fly6CjTwwQ9+UHX30d7e /vrXv14t0gdbx6fw6U9/OnB1QR/SKIx1nzlzxrAiNnfLLbcEte43vvGNatQe xdNPPy1F0boJIcRcor3707qdCq3bzqgvLKwpMzNTDZRz9uzZ69evy7Tec6DH 48Hs1q1bZZFq0wb19fUZGRlSgi5yY2NjKE2FmqT5h7xsbW1VGY4dOybpcDaV qLRZDyYJuvWuri6Z3bJli8o5Pj4ORVehJps3bz569KiKLYlqu9h/2eL27dvN Gp4SOyZ6+dWvftWwCM84sujjH/+4no5PBykq9htSfdddd2Hf9Dznz5//xCc+ IeM83nHHHfoieOy//Mu/yNCWArJ9+MMfVk3lQmlpqSx95zvfGbjbe/bs+d3f /V3J8O53v/vixYtKreVXAF3acf6ojlbe9773/ehHP1LlqC4HX/e616lEWjch hMRMtHd/WrdToXXbGcPXdmpqanBwMJLekqf9BF2E1aX7awMoGbplSlfMYbZu AP7W398/MzMT5+b6+vriLMQU8DRRVFSkAntCgceToB/B0NBQXl7eyZMnocQG Y4+cTj9BFwU2lUs/8LFtiBBCSIREe/endTsVWredsfo6QRxF+I4HCSGEJIho 7/60bqdC67YzVl8niKOgdRNCiCVEe/endTsVWredsfo6QRxFZ2fnnX4ef/xx q/eFEELWEdHe/WndToXWbWesvk4QQgghJF6ivfvTup0KrdvOWH2dIIQQQki8 RHv3p3U7FVq3nbH6OkEIIYSQeIn27k/rdiq0bjtj9XWCEEIIIfES7d2f1u1U aN12xurrBCGEEELiJdq7P63bqdC67YzV1wlCCCGExEu0d39at1OhddsZq68T hBBCCImXaO/+tG6nQuu2M1ZfJwghhBASL9He/WndToXWbWesvk4QQgghJF6i vfvTup0KrdvOWH2dIIQQQki8RHv3p3U7FVq3nbH6OkEIIYSQeIn27k/rdiq0 bjtj9XWCEEIIIfES7d2f1u1UaN12xurrBCGEEELiJdq7P63bqdC67YzV1wlC CCGExEu0d39at1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Ada t1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE 2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNC CCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1Oh dRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Adat1OhdRNCCCGE2Ada t1OhdRMSP3tsjNV1QwghJDpo3U6F1k1I/Fht1uGwum4IIYREB63bqdC6CYkf q806HFbXDSGEkOigdTsVWjch8WO1WYfD6rpJHriUzc/PT05Ojo2NDQ0Nud3u fj+YwCwSsQgZeMUjhNgcWrdToXUTEj+65d6wAevKunEFm5mZGR4evhkxyIxV eOkjhNgTWrdToXUTEj+0bktYXl6emJjo7+9XOu12u2HUo6OjHo9nfHx8wg8m MItELEIGlRkrYikKsfo4CCHkVdC6nQqtm6xPlpaWGoLR19cnGfDVaGtrq6io qKqqgp5hVtJXVlYwi8SSkpLa2tqxsTHf+rDuubk5HDKsVU+EzdbV1SEdFdXe 3q5qKSpCVbWOZFCGjE9hampK+Tb2ClINhZ6MAGRDZqXfKARFocAY9pwQQhIB rdup0LrJ+gRnflBfra+vl6WFhYWYxf95eXmYqK6ulhU7Ojowm52dXVxcLKsM Dw8HWneTRSTIur1eL9QaJff09KjEoaEh2VxRUVFubi4mUCfRNh2HqWoFalg2 hN2QVQYHB8WZMQHzj0S2A8GKejmxPTIQQojp0LqdCq2brFvgh0urzM3N5eTk wKWnpqawqKurS8xZslVVVWEWnraysiJ6OTs761u1wYqKCmdbN4xUfNhg3ZBt eeiQWqqsrMTswMBAVIWHqmqVYXFxUW0d1o2alyZu/C8vSMYJClEFysdKCCHW Qut2KrRukrrAlJqbmyHMLS0tZWVl+B/TsRVVW1sLqevu7pZZl8uVn5+vmm1n Zmb6+vpQOFJg5hBvCUiABEpLr52tO85agvRK2774sLJuPKoUFhYWFxernKgi ZOjo6Iiq5kNVtcyinrFdbL2goACFj4+PS9P00NBQhPEkkYCiUKCULI9dhBBi IbRup0LrJqlLaWkpTCwvLw9WJu2ukGdZND8/j6XKoKanpyGcoWxTIiUqKirE pWGAmK2srHS73fX19dXV1V1dXUoLpUW3tbVVQpoxDae1s3XHWUu4ROAAkXNs bMzQ1m2gpqYGGfTA7zXLD1/VvlWT7+zsxFKUL2I8MjJilm/roFgpf1JraSeE kORD63YqtG6SokizM5SspaXF6/VCmPPz8+GWshQWh0W5ublQPsiehCiECn4Q Lx0fH5fZ2dnZwHhv6KJ8UyCNenphYSG+RHa2brNqKbx1S1EFBQVLS0uGxDDl h6/qmZmZnJwcmcVjjijx6OhoIpRbQOGyFWw6qrOREEJMhNbtVGjdJEWZmJiA oUGYpYEapzH0ErMqA/xQlE8isUPpIoQQS0tKSgwlS4M2viAQMNFy6CK2Iq/+ IT88VsIeGhsb7WzdptSSL6x1o2ay/Xg8HsOi8OWHr2pUMqwbjzmLi4t9fX2J Vm6DeGOjYU4/QghJHLRup0LrJimKxB4oi5uensZsQ0ODnqe5uVk5cKhyJA8s S6VIazZ8TzXb9vf3SyQJrFIiIiQd3x2XyxVouXaz7vhryRfaujs7O8WrpRPF UDUctPwwVS1vWZaXl8PA1dA2iVZuQQbcwXbZlTchxBJo3U6F1k1SFKgjrExF Aoitqd62fX7DVH1fSJBDYCHSJwnQ4yIk2Bg6rVJGR0dFGkUFe3t71SJRSptb d5y1JARaN2qvtbVV6grlBF0rfPlhqrqxsVHWwgctPXJH+/ok5Hl8fDwG68aG pDfvwKZ7QghJArRup0LrJilKUVFRTk6OGtxEhHNy9T04JXuwREhyKKWcn5/H oqqqKkO6xDmo/KLWAwMDEqgMIdR3IxLrLigoyPCDifDaHGfOQOuOv5aEQOvu 7u6+4e+je2FhIegqkZQfqqpnZ2fhzCreIwZ/rq+vHxoaisG6geoshXEmhJDk Q+t2KrRukorIGDcVFRUqpaSkJDs7W53M0pO2UkRRPjiYoRxpWW1tbTWky+r5 +fldXV3SUYnL5fJ6vdiuhJQ0NDR0dHSUlZXd8L/9F966MzMz//Ef//GrfjCB 2VAiHX/OQOuOv5YEg3VDR+HzN/wh7jUa+suSkZQfqqplqfQrEltsCT6jmK27 t3dY4kywA6FPQ0IISQi0bqdC6yapiMRXt7W1yawEKsB+9TyGHvCCdhsoHtjf 3x+4CPaoQiNKS0tVkAYmRLYFeObCwkIY6y4oKFB6rCQ5aDt2nDlDWXf8tSSI davoFOjojWAYHmEiKT9UVUPspcE5tq65Y7Pu+vreQxnljzx8FRtlczchxBJo 3U6F1k1IKFZWVmCJqt1VRzrcUNHgYaw7IyPjqwEgMdCl48wZyrqTW2cxErSq 8WwVz0uUUVn3yIinqKg9fXvBXT947t57btz1g4uTq69VMrqbEJJkaN1OhdZN SPxEa90HDx6Mx7qxusOsOxBcl2Sg9tjeiIzcum/eHLl8ufE/f3zj7ruev+c/ rv/w8etnTle3tvZPrkZ3Yzd4kSSEJBNat1OhdRMSP1FFmIDjx4/X1tbGFmGC FbG6ocAwESZW102MzM3NSdclsSl3JNbd2jpw4kTVww9d8TduX96QlnP9eovb /aouwaUzkzCBN4QQYjq0bqdC6yYkfiJ/mxI88cQTp0+fPn/+fGVlpcGoDx06 pLs0Zg0ZsApWxOooRFfuMG9TWl03MSLtzPH00R3Kuj2e8aqqnmeeKb7nPy7d e0/23T+4uGNHQWlp59hYkEZ1CTJR45YSQkgSoHU7FVo3iR+cRfX19df9YAKz Vu/Rq1D95iWO8NbdtNrL3759+/A/nPnSpUv4/8yZM8XFxXq2/Px8ZNjqBxOY 1ZciM1ZRq6sCw/ccaOJhJqEmFYODgxJTbaJ1Dw6O5eW2bt6U/4Pv/yp4+8EH rhw9ggefm2EKkdhy7EzSDpwQQmjdToXWTeLE7Xbv3LkzTQOzSLR6v/5/FhcX m5ub+/v7g74XaQprWrfi8uXLEOaioqLs7OzTfnJzc9XSCxcuIKXCDyYwqxYh m+THilgdEygqsPyEWncSalKA3sfTe4mybtVU3tU1eP583Q8fv3bP3df+4+7n f/xk9sVn6/v61m5IVz2ZJPOJgxCyzqF1OxVaN4kHaMnTTz8N0967d+81P5jA LBInV4disQMLCwv/H3t3HlzXWZ8PPEBaSKEsHbZhaaGUMlCYDqWltEyHmXYy 0w6FsiSB0BAIpSxJIJCEAUJIbMm7YzuybMmW5EW2VmuxZFvL1b7vm6WrfV+s zZLuSJX+/z293/r9Hd9NV3c75773+Uwmc85733POe18dXT/36D3vkcfZNDY2 hiM0+p+6q6urEZjz8vKwjCbJteuioiJExLq6Oizn5uZKTSxgFYV4CRXk2jg2 wUvYHKsu18kjkLp3wt+TAp9LchtjwJEb+vr6kLo72kfPnav7+bP/O5jkxz/K PXyovLy8b2ZmD5fQ5aZOq/0Fh4g0xtStK6ZuCkZBQQEydmpq6vDwsFwSxAJW UYiXzG7dfVRcFKENjf6n7s7OzqysLGTmpqYmrNbX18tqnhMWbDab1MSCsRzV UBnl2FBWsavIp+6dMPekkJm65VbKhYWF6b2Ynf3f2yEnJ6eyskrj46//8L8y nv7p9R/9d+axoyXV1T0ox6uo43snxolT5IZKztpNRBHD1K0rpm4KxvHjxxGw Ef9GDbAql7s7Laatra20tNTlqS4It0izeCmYPfufuqGkpEQGishqS0tLbm7u 5cuXDx06lJaWhpZIORYuXLiAwkuXLqECqkm5DE3BTjzu3D11h6r3jMLXk6K3 t1el7qmpqRY/tLa2dnR0dHd3V1bWnz5d8POfXfiPr8Z/5/GzP/rvcwkJ+SUl 1Xi1yzkDjD97Mw4Il9S9trZm9m8bEcUKpm5dMXVTMOLi4hCwh4aGjKkbqyjc v3//zZixp9SNUIfYnJmZKTkQ2tvbEbDRmSdPnlTjRrCAVRQiiqOCFGITbIjN VQjfNXWb1yuBq62tlXsY9zSkpL9/Mv1yy69+VfzsMzee+WnRL3+RXlzcMTkZ +P2YQu7rZOomoohh6tYVUzcFIyEhAQG7sbHRmLqxikIkxgmLuX379q1bt4zp zmaztba2Dg4OBrnnPaXuLsM9laokJycH26ampsr9lXLvJFZRaLyn0sd9lN5S d0i6zkX4elKMj4/7P1n3/PxCY+Pg66eqn3k6739nAvxp3okTlQ31g62tbUHm bTHhvNa9vr5u9m8bEcUKpm5dMXVTMCoqKhCwT506JUMCAAtYRSFeMrt195mf ny8pKZGIWFNTY7fbV1dXQ7XzvaZu4z2VoO6jRKi+cuXKa05YwKq6p1Jq+riP 0lvqDtV7VMLak0LGdSN7+87DU1OzJSU9B+LLfvy/MwHeeu7nBakp9V2d/zcT YGdnp/9PhPdBvgJwXDcRRQxTt66YuikYDofjzJkzyNjx8fGpTljAKgrDPbnc nkhQDFNE3Nl76u68/55KmaJELhcjbB9wSk9Px6rcU4kKXbvdRxmZ1B3unhSb m5u+5zAZ/N+ZAFt/+5ubzzxdhP9e+u2Na9faBwfvy9h+PhF+VzKHCZoUpjdL ROSCqVtXTN0UpLW1tczMTON83Vi11CDY9fX1oaGh8EXEnb2n7q5791Sir86e PYtOO3fuXG1trTx3MsNJnl+JQryECqgmQd3bfZQRSN0R6Emxvb0tWddlvm6s dnSMnD9X99zPC5wzAV47fKjcVt43PTXnnpZDkrplvm40JtxvmYhIYerWFVM3 hQTySY/TopWm6Y6YAFJ3S0tLXFyc8fnv8fHxCNWFhYXtTljAKgqNdbCJt/so I3OtO2LkUexzc/8Xp2dm5muq+48fq/jxj3KRt3/2bP7ZM7XNzUM+HqMTktSN BsiD6c3uDyKKIUzdumLqJgpeAKm7oqLi0UcffeR+WVlZxjqZmZkuFbCJ8fnv GqduJF6JuyMj00XXu1595dZPf1L47DM3f/XC9fT0pr6+XYZ8hyp1S/hfjMnv kkRkFqZuXTF1EwUvgNSdnJz8iBsU7rWOrql7fX1dhnb84rn8Z5+5gci979WS ouudo6MzfgbmkKRuGejCCUyIKJKYunXF1E0UPKbukNve3pbJQ559Jv/Ea5W1 tQNzswu7p+SQpu75+XmZSgWNMbs/iCiGMHXriqmbKHghGWHiPnrEnzq6pm6Q Oxnb2uyBZebgU7c8HwfNMLsniCi2MHXriqmbKHgBpG5ISUlRoRoLWA2sjq6p W+YPhIWFvV3lFu3t7cGkbhxUjs45A4kowpi6dcXUTRS8wFJ3l/NqdrKTjyvY /tTRMnXvOKcHD+DR8GJ6ejqwuC7kQjcaYHYfEFHMYerWFVM3UfACTt0hp1nq 3tzclPsZ1RSCkSETBuLQvNBNRJHH1K0rpm6i4DF1h8/S0pIEYB9Tc4cWDiRR H4c2+90TUSxi6tYVUzdR8Ji6w2d7e1smzZ6YmIhA8MYhcCCZKpxTlxCRKZi6 dcXUTRQ8pu6w2tzclFkEg59/e1c4hMwWyLElRGQWpm5dMXUTBY+pO9w2NjZk RhGk4jBd8cZuJXKPjIzgcGa/YyKKXUzdumLqJkDYKCoqSklJOXHixJkzZ65e vdrR0cFrff5j6o6AtbU1Cd6Tk5MhD97YoUxagshdU1Nz+/Zts98uEcUupm5d MXXHuK2trcLCwn2enD17llf8/MTUHRnr6+tyn+P4+Pj8/HyoIjd2JSNYsHN8 A7116xb6kMGbiMzC1K0rpu4Yl52dLRn70qVLvb29c3NzPT09BQUFBw8eRGFq airOELPbGAXcU7e5dE3dgBNSLkrLDY9BXvTG5nKrplxCl7N9ZmaGwZuITMTU rSum7liGgC2Ru7Ky0mW6hoGBgf379+OljIwMs5oXRZi6IwkfWfL0HLk6jZAc QPbGJthQrpzL03CMn4QM3kRkIqZuXTF1x7ILFy4gVycnJ3ucIa2hoUEy+ezs LFaHhoauXLlSVVVlrGOz2VCI0KJKUC0zMxNh78SJE3ipt7dXvTQyMoKStra2 oqKi48ePp6enZ2VloaSzs9O4z7m5ORTm5uaG+N2GE1N35G1sbMjNj+ouS5w5 u8ZvVEA1lw09jqRi8CYiszB164qpO2atra1JqO7p6fFYAeeGXO6WpN3a2opl 5GFjnYsXL6JQxebq6mrZZ3x8vIxRAeQWebW9vd1l6Dj2JgPIjfssKSlBYXZ2 dsjfcvi8bmFm900Y4dsiTmM1RERMTEwgSCMzzxpgFYUyEbeCDbG5j0m5GbyJ yBRM3bpi6o5Zk5OTxkvZHp05cwYV8vPzd/xI3YgoktJtNhtOKoSZtra2uLg4 lMjFcJW6U1JSsElZWRmCkJRgW9khtjp+/DhKBgYGwvjmQ83sZO2L2X0TCfgc W7w31fauZNJvP29YYPAmoshj6tYVU3fMQqyVxOsjfuTk5KikvWvqRjjHMmKe 8eJhVlYWCvH/HUPqXllZURXS0tJQUlpaKqt2ux2rCN7RdVqanax9MbtvIgqn zfr6+vLy8sLCwvz8vFzoxgJWUYiXAjivQhi8HQ5HkHsgoljA1K0rpu6YNTo6 Khl4aWnJW53Lly+jQmFh4Y4fqfv8+fNYTkhIKDdITU1VY0gkdZ88edK4h7a2 NhSeOHFCsnpubi5WS0pKQv+Gw8nsZO2L2X2jg+CDNz5p8WuCExvJP7RtIyL9 MHXriqk7Zt25c0dS9/DwsLc6CMOo0NDQsONH6j527JjHeb/h0KFDO/dS94UL F4x7cDgcMgIczcDygQMHfA96ITJFMMF7enoa30CxbWNj48jIyNraWjhaSETa YOrWFVN3LDt16pQatq1MTU3JwsTEhGTmwcHBnXupOz093Vg5MTFRpe6EhAS5 d3LIjQR7Sd2XL192aYYMTSksLOzo6JA5VcL2jokCF0DwxhfJtrY2bFJfX49f qLm5OZzkpaWl4+PjYW0qEUU1pm5dMXXHsqqqKsnVKgPILZYIxiMjI5Ko8f+t rS28hGi9z/ncHLX59va2XJqW1H316lUsFxUVGQ/R39+PUF1XV7fjPXXLWJdj x45lZmZiobGxMazvmihgewre+G0qKysrLy9X32TF2NgYCuXXiojIHVO3rpi6 Yxl+9HK348GDB1tbW7GKnHD27Fk1MmT//v0TExNSWe5zNJYUFxdLNUndcjEc FdSQlZWVlcOHD8usJjveUzfS++uvvy67iouL49/fycr8DN7T09Oo09HR4XK3 Mn7LUMjUTUQ+MHXriqk7xuGf/kuXLkniPXDgQFJSkkzcp1I3ErXc57i5uXn0 6NF9zrm4z58/L0O+jakb1eSGShmIcuXKFbkSrq6We0vdO4ar7jLbCVHA8IHW 1tZWWFiIc6m6uhpfJEN+CD+Dt3GuHlVSV1eHbdVXVyIid0zdumLqJofDUVZW Jolahe2rV6/KtIFw7tw5qYkAc/r0aSlE9s7Pz0fwwHJ3d7dUwOl048YNCdsi NTVVzcUtY1RcRoaLpaUlqd/f3x+Zd01awsmcnJzscjOv3A4cWnsd442PWbvd jk1qamp8zBpERLTD1K0vpm4S29vbCANTU1Ozs7NqVmHEbMRmJHBjzZWVFZRv bm762NXCwgLq3L1718+jj4+P74vCabrJagoLC3EiHTlypLa2dmRkRA2Ckuc0 hZb/wRu/WQjbqDwwMGA8w6enp5ubm0PeMCKKdkzdumLqpl2F+9EeW1tb6enp iEZlZWVhPRDpDd8EZRZKRFxVeO7cuX2GxzCFlp/B226319XVuQ84wXcBnvNE 5I6pW1dM3WSijY2No0ePSlI6fPiweyyJHfg17OzsLC4uzsjIuH79ektLC++2 2yt8PUS4LSgowHmlCtGZ7tNjhpCfwdv4wFZlZGSEqZuI3DF164qpm0yEKBIX FydDAoaGhsxujmnW19eTkpJcRiMnJCRwAHCQRkdHZRYddetBOOx1jPfc3Bw2 mZ2dRatKS0tn7+FjK4lIMHXriqmbzLW2toYE4vFKYOzIzc2Vrx7IYCMjI5WV lTKTjMtzPMl/1dXVapqd/Pz8cP/hYE/B22az3fSEE5sQkWDq1hVTN5G5Njc3 4+PjEQ7r6+tVocyyCLz+GRjj3w6uX78egW70P3j/zz0ywmTLCasx/t2TiBSm bl0xdROZa35+/vTp00eOHFldXVWFDodDEqPLYw3JT0tLS3fu3KmpqXnttdfQ jefPn4/AQfc61IR3UxKRR0zdumLqJrIg9SRQlycb0l719PTI95fI3Ku7p+Dt O3VPTEw0NDT09fVhYWRkJByP+yEia2Lq1hVTN5HVrK+vy9OIrly5YnZbogk+ zcbHxwcHB42FW1tbMklOV1dXZJqxp3m8fdxEvLCwgK8MCN6yt7DeEEpElsLU rSumbiJLcTgcqampSImHDh1aXl42uznRRJ61FBcXZ5w5EMsyT87w8HDEWrLX oSY+jI2NSeTmqG+i2MHUrSumbiLrWFlZkdsA4+PjXa7Z0q7UZe3r16/LxxpK rl27Jl9hIjxWJyTBm5GbKDYxdeuKqZvIIubn50+ePCkPDBoZGTG7OVGpra1N RnEfO3YsNzc3ISFBVk0ZnhFk8GbkJopZTN26YuomsoLp6WkEReTDEydOzM7O mt2cKNbe3i49KV5//XW73W5WYwIO3t4iNz6uV1ZWNjc3Q9pMIrIWpm5dMXUT mQ7ZTB6hmJycHJmpNvSGpHrnzp3R0VHjZIxmCSB4e4vc+DpWXl6Ol7DD3t5e fnQT6YqpW1dM3UTmQrJKSUlB5E5MTFxbW9u6n9mtoxDYU/D2MbAEp0dLSwuy N+pUVlZ2dHSEp71EZDKmbl0xdROZq7e3d593yFdmN5BCwM/gjaTd0NDgz1ju hYUF7O3OnTshbSYRWQJTt66YuonMVVlZ6SN1j4+Pm91ACg0/g/fW1pY/t0/i c7ukpCSS0yESUcQwdeuKqZuIKDJCOI83vo5hP9hhSBpGRJbC1K0rpm4ioojx Frzdx/Bvb29PTU11d3e3tbV1dnb29PT09/cPDQ3Z7faGhgbsobGxkZMKEmmJ qVtXTN1ERJHkMXiPjY25zAc4OTmJOqWlpUjdra2tyNi1tbWVlZU2mw2pG/Gb H91EumLq1hVTNxFRhLkHb7mI7VJtYGAAdUZHRyPeQCIyE1O3rpi6iYgizyV4 I3WXlZW5P/6GwZsoBjF164qpm4jIFMbgjdSNBY+P0fQdvHt6erBtmFtKRBHF 1K0rpm4iIrOo4N3Y2Ij/e7zcvXMveE9OTrq/JPN7h7+lRBQ5TN26YuomIoow RGvHPRMTEzcNPF7u3nHebomPa/dypm4i/TB164qpm4gowjY2Nnp7e0tKSm66 8Xa52wif2zMzM7NO1dXVzc3Ns/fw85xIA0zdumLqJiIyhbfs7e1yt3Lnzh33 uC7c5/0moqjD1K0rpm4iIhO5Z+9bt275vty9vb39P/c0NDR0dXVtOfHDnEgP TN26YuomIjKdZG+5sxKQpf3csLGxkeO6iTTD1K0rpm4iIotA9m5paZHgjRDu zyY+Ujc+3kdGRurq6iorKzs7Ox0OR0gbS0ThwtStK6ZuIiJLGR8fl+BtfGS8 j8qzs7Pu5SsrKwjb5eXlfX19Y2NjTU1NNpsNqT4M7SWiEGPq1hVTNxGR1czM zPgfvN0hh5eUlDQ3N6vJBre3t6uqqjgWhSgqMHXriqmbiMiCXB4Z77/l5WW5 N7Ompqa1tXVlZUXKu7q66uvrw9BSIgoxpm5dMXUTEVlTAMF7c3NTBpbY7fb5 +fn29nbs4c6dO3gJIZzXuomiAlO3rpi6iYgsa6/Be3JyEvWXl5dVCcJ2U1MT FtadwtVQIgodpm5dMXUTEVnZnoJ3Z2enZGwFW1VUVIStdUQUekzdumLqJiKy OP+DNyoYUzc+3mtqalpaWjxWxuf/6upqKBtKRKHA1K0rpm4iIuvzM3ivrKyg 2sTExPb2tsPhQN7G6tLSEl7a2trCAl7CHpqbm202m0yTom63JCKLYOrWFVM3 EVFU8DN4T05OljmhZmVl5dzcnN1ur6qqwioKGxsbe3p6xsbGELxRMjg4GKnm E5G/mLp1xdRNRBQt/Aze+GCfmJhA5a2trc3NTURu5G3jLZb9/f2M3ESWxdSt K6ZuIqIoEsB0ghsbGwjedXV18tAcRm4ii2Pq1hVTNxFRdAkmePf19TFyE1kc U7eumLqJiKJOYMFbBnsPDAyEtW1EFCSmbl0xdRMRRaO9Bm8ZWFJeXl5XV7e5 uRnu5hFRwJi6dcXUTUQUpfwP3mNjYzKwZGNjA5W3t7cj00IiCgBTt66YuomI opefwXtzc3NiYiJirSKiYDB164qpm4goqgUwxpuIrIypW1dM3URE0Y7Bm0gn TN26YuomItIAgzeRNpi6dcXUTUSkh9AG762trTt37oyNjQ0ODvY5YQGrKMRL we+fiLxh6tYVUzcRkTaCDN6rq6tI1+Xl5VeuXElKSjrrBV5CBVRDZWwS6jdB FOuYunXF1E1E1vG6hZndN/4KIHg7HI6enp68vDxvMds3bIjNsZOwvi+zmH3e +WJ231C4MHXriqmbiKzD7BTji9l9swf+B+/19fXGxsaUlJTA8rYRdoJdYYeR eY8RY/Z554vZfUPhwtStK6ZuIrIOs1OML2b3zd7sGrzxyd/e3n7+/Png87YR dojd6vTPitnnnS9m9w2FC1O3rpi6icg6zE4xvpjdN3vmI3jPzs5mZmb6SM65 ubnl5eWNjY1dXV3Y3O6EBayiEC+hgo/Ejp3jEKa865Az+7zzxey+oXBh6tYV UzcRWYcxUdy0gGhPOB6Dd2dnZ3JyssewXVRU1N7ePjIyMuoHVENlbOIxfuMQ OJCJ7z1UeE5S5DF164qpm4isgwkn5IzBG5/2ZWVl7gk5NTW1oqJiaGhIJeqJ iYnp6enZ2dn5+fmFhYU7TljAKgrxEiqoytgQm2Mn7nvG4aL9nxiekxR5TN26 YuomIuuIhYRz9+7dmpoapFZjIdJsW1sbyhsaGhCP8ckcwJ6xVV9fH/bQ1NQ0 NjamdiLBu7i42H1USVJSUklJicrbaBVCNQL2oh9QDZVV/MZOsCv3+QavX78e 2NvZVWR6MhbOSW82Nzc7PBkZGXHvKOMpt729jVUU4gfR2to6NzcX7qZqhqlb V0zdRGQd7gmnyyRhSjgOhwM5BHtGRlWFU1NTcriqqqrS0lIsVFdX7/VJNPgw r6ysxLb4f1lZGRaam5vVq5OTk2lpaS55+PLly93d3ZKZUQF51Z+w7Q4bYnPZ D3aI3bocqKCgIOQP1olYT2p/TvruCo/Jv7293b2jjKdcf38/VvFdD/0vm0xP T4e7tTph6tYVUzcRWYfeCQfRVMKJS1ZERFSxBBGxsbERq+Pj43vaud1ul+6S nTQ1NWEVkXjHeeERy+5JWC5xj42Nzc3NBZa3jbAT7EouemPnLodDAEMzgu/D yPek3ufkrtADm/fcvXu3pKQEP8qlpSX3jlKnHH7Q8pVHnqCEHweWGxoaItBa bTB164qpm4isw8oJRy7kInj09PTU1dXh/1j2/61tbGzI1T8JJyorIsxUVlZW V1ermiMjI6jQ39+/p66z2Wzl5eXquu7Kygr2Iy1E4HHJwMhOcml6amrKz/Ek /sCusEPZMw7hclCVu6KrJ618TvrzdoLsbaPW1lY0YHBw0GNHqVMOJfjpIHjL 9yyHwyF/fQjsoLGJqVtXTN1EZB1WTji1tbUoKSsrQ6KQa6oIIdLstbU1vCoX AGF5eRnxxiXb4MMW4Qc15+bmXK7QumhpaUEF43DlXfePnINNGhsbsVV7e3tz c7Pdbpc4hNDlfoejBOOZmZlQ5W0j7Fb2737n5tjYGJoUXT1p5XPS2xs3CrK3 FRm9g69OkqV9nHIgf2Xo7e2VYfZYxo/Mn9aSYOrWFVM3EVmHlROOzAQijz5H 8CgvL0eSkWYjeOCl0tJSBBhEFxn84G1gg++sKLuqqKjY3Nx0KfSx/9XVVdeh tzdvIkEhVrkM5y4uLpZIPDs7G47ILbBzOQoOZzw6GoNQF109aeVz0uMbN5LL ziHpbUnvCwsLvk85SRT4KRvLKysrw3RHra6YunXF1E1E1mHlhINVBA+50IeP TYQZrKqWI/tJgJERrT4uwPrIigg8t5zm5+ddXvK9/zt37kiDe3t78am+srIi Gckl9GZmZspc3GGN3MbgjcO5TJxSUlISXT1p5XPSW8+4vJ3gexuxHK/W1NT4 6Cg55dDzOIrcZYn6SPv44oNlPSZvjximbl0xdRORdVg54RgzyfLyMlY7OjqM je/u7lZt9vEevWXFgYEByT/eplnzsX+5tIhAq67rjo2N5efnG+NuSkrK7du3 kYSnp6fDHbkFDoTD4aA4tLElaFgU9aSVz0kfnSNkZHvwvS118NP00VE45W46 R5Lgm85N5+ATKUfGsNlsKNnY2Ni1wSSYunXF1E1E1mHlhIPVlZUVaacEDDVr 8Y4zz6hZNeRP9t7eo3tW3N7e7u3tRSHCCfbjcSvf+5dBtthclczOzl64cMGY dWtra2VG7hDePukbDiSzeePQxpagYVHUk1Y+J731jIKAHfx5K3OSgHGsjsdT TjpH5jYZHh5WL0lon5qa2rXBJJi6dcXUTUTWsWvCqaioSHbCgu+IEmRN94RT UlKi5r6TCovOefl2DNEFCRBhw3dcdM+Kg4ODN50zS6+vr3vcxJ/9y9/3VaHL vCUZGRkytmRhYcHPzDw/P499njx58oUXXnj66ad/9rOfvfTSS5cvX+7v7/c/ eONwMs4EDTC2BwktWnrSyuekx/doVFVVFfx5u7a2hpeamppcyl06SqL1+Pi4 DBc3DimRuzjVcWlXTN26YuomIuvwnXBSUlIeffTRR5ywgFVvoSX4mu4Jxzjh cE1Nza1bt9SHp8xIrOKfBBhvV/ZcsuLGxoYMdcY+WwyMN7X5s3+pU15ebrfb 29raLl26ZEy5ra2t/o8tqa+vj4uL+9a3vvWIF0899dSFCxfQDH/2JuNM0ABj e9DOaOlJK5+THrtFkWfcBH/eykXs3t5el3KXjpJL3w6HA8eVISUdHR34jlZX V3fTcKMl+YOpW1dM3URkHT4STkVFhYoiKpB4vGYYZE1vCaevr08aKX9bR4ow ttxlvjUfUyJLVlR/5Z+ZmbnpiUvI8Wf/iJdy6bK4uNj4ZPaMjAyZUWTXsSVI Zc8//7xLwP7tb397+PBh5PBf/OIXjz/+uHrpscceQ7fsmr1xUDm68XL3+fPn 8a+P9Xvyps85TEw/J731jJDx1cGft5LGZdZHHx1VW1urhrJgQcK2wHcfb399 II+YunXF1E1E1uEj4SQnJ7tfdEWhe24JsmZgCccitre35XkoxgvLyD+7XuhG skKuNvYDwjYCm3uERqR/8sknVbUnnngiOzvbd56Xy91ohrFVt2/fNru3fJGe dDgcPCd9Ux3l/pLMbWIcDU5+YurWFVM3EVnHXhNOUlJSMAkHm0dXwvGT8bmQ 586dkye/+xjRXVZWJkH68ccff+yxx7Bw4cIF3xH9F7/4Bap973vfUxHdx0Vv Gd2NZqAxqmFopNn95BeekxR5TN26YuomIuvY01/z4eLFi62trS65xc+/5mND bO6yw4D/mm8pxtlL8vLyZOoSb3kYSVje+yuvvCLDS06ePOkjcqvg/YMf/ACV z5w589RTTz3iHIvS1NTkrb5MZoLGGGcyMbuf/MJzkiKPqVtXTN1EZB0+Ek7X /XeZwUsvvXT16tXs7OzGxkaX9KKSpMCqSwVsgg2xOXZijDcB37lmHXfu3DEO 5KiurvY2vGRubk5GlTz22GNZWVl1dXVY/v73v+/nTZfFxcWo/5Of/AT7l278 1re+VVJS4rGyDDJBY4xtQ1PN7q3d8ZykyGPq1hVTNxFZh++E03VvRrXExET8 H/kkPz8f/8/IyECcM1YrLy9HhSNOWMCq8VVUxiZqc7XDIGdp85+ayS0cBgcH jcm2u7sbcXd+ft49Bku6++53v1tfX48SvEespqWl+RO5F51jvOVyd2dn58LC wunTpyUlFhQUuFdGA9AMNMbYNjQ1yDcb1p4UPCetBrmlvb39hhMWtHzWPFO3 rpi6icg6dk04CqIdwklVVdWtW7euOpWWlqpXc3JyUNLghAWsqpdQTepjQ2yO BezKff9hTTgbGxvIn2NjYx7vQQtSS0uLirVJSUnDw8Pus5fMzs7+5je/kWEh eLNS+PTTT6Okra3Nz9QNJ06cwCaZmZmyeunSJblC6x68ZSYTNMY4uQqaGuSb DWtPCp6TljIxMYGzbp8BVlFodrtCjKlbV0zdRGQd/iec6upqhJO8vDwsI6jI dcKioqKOjo66ujos5+bmSk0sYBWFeAkV5DokNsFL2ByrLtckI5BwYH19vbKy 8qbzwdmhjTo2m03F2osXLyLrYv/GALywsLB//35k4//6r/+6ffu2Kv/2t7+N Qm8jwD1KT0/HJomJiaokOztbrniXlZW5VEYz0Bg0STXP+GTDgIWvJwXPSevA WXT48GEk7dOnTxc7YQGrKFzU6xE8TN26YuomIuvwP+F0dnZmZWUhnzQ1NWG1 vr5eVvOcsIBEJzWxYCxHNVRGOTaUVewq8glnxxly5Jl9IlRRB7tSsTYzM9P9 VsqEhAQE4yeffLK7u9tYLrOXzMzM+J+60XvYBDs0Fsr9gMjwra2txnK5oRJN Us1DU63ck4LnpHXid0FBATJ2amqq/AVH/nqCVRTiJbNbF0pM3bpi6iYi6/A/ 4UBJSYn8UV5WW1pacnNzL1++fOjQobS0tLa2NinHwoULF1B46dIlVEA1KZdh ANiJx527J5zOMEDbSktLb94PkQwZDC8Ftk+8KRVrc3JyXFJ3YWGhROLm5maX CC2DtI1Xv3cl09y5TzP4unOIOHZovMwuqRtNUs1DU63ck4LnZKh6MnhyoRsL owZYReHx48fN/vQKJaZuXTF1E5F17CnhIKsgomRmZnZ0dEhJe3s7wkxcXNzJ kyfV3+ixgFUUIvagghRiE2yIzVXg2TXh3IwSeJsq1l67dg2xZHJyUkXf3//+ 98jD6A33CI3ogpfy8/P9T90y02BlZaVL+fDwsFw5Ly8vV4VoBhqDJqnmoalm 99bueE5ax/79+3GWyvzzClZRiM40+9MrlJi6dcXUTUTWsaeE02W4f02V5OTk YNvU1FS5l03uU8MqCo33r/m4Z81bwpkIg9u3b9+6dcuYK2w2W2tr6+DgYMD7 NF5Mzs7OdrnW3dzc/K1vfQt5OC8vzyUqozdQ/qtf/crPyI1eQv3vfOc7s7Oz xvKZmRnsRHZlnDtlwnmtG00yXoq3ck8KnpOh6sng4asKAnZjY6MxdWMVhQkJ CWZ/eoUSU7eumLqJyDr2mnCM96+BumcNAebKlSuvOWEBq+r+Nanp4541bwkn 5G8WibSkpEQOVFNTY7fbV1dXg98tIpOKtRkZGYgl4+PjxlR8/fp1ueHRZW7t 6elped4N9rBr5F5YWJCJB3EUYzne1KuvvoryH/7whyMjI8aX0Aw0Bk1SzcOB gn+/O2HrScFzMuRHCVhFRQUC9qlTp3p7eyVyYwGrKMRLZrculJi6dcXUTUTW sdeE03n//WsyHYRcmkOwOeCUnp6OVbl/DRW6drtnLTIJR+JNOIJNZWWlcQiH +xwmcPnyZXmoDbrFWJ6fny+Xr9vb232nbhnR/b3vfc8Y6fGmDh48KOWIQy6b yBwmxgEwaGrw7zd8PSl4TlqHw+E4c+YMMnZ8fHyqExawikLr3PIZEkzdumLq JiLr2GvC6bp3/5rMjIF/f8+dO1dbWyvP+MtwkmcFohAvoQKqSSjyds9aBBLO +vr60NBQmIJNR0eH8Uk0MgjWZb5uFZsfe+yx4uJiY/nRo0cleOPte8zbExMT R44ckW3l8TpiZmbmlVdeQfkTTzzhPum3zNeNxhjbhqYG+WbD2pOC56SlrK2t oWON83VjFYVmtyvEmLp1xdRNRNYRQMJpaWmRx5or8fHxCDCFhYXtTljAKgqN dbCJt3vWIpBwwmpkZMQl2SLuzs3NuefnlJQU6Y3Lly+rQtQ8duyYlL/wwgvX rl3r6emZnJzEbhsbG5OTk7/73e/ipf/8z/80XifHqy+++KI8UN7jdXLsFs1w +UaArczurd3xnLQgnFE9Tot6TdOtMHXriqmbiKwjgIRTUVHx6KOPPnK/rKws Y53MzEyXCtjE+KxtnRLO6uqqMdmWl5cj7k5PT3u8cI2ekd5D5Jsw3HRZUFAg Ewl69PLLL0vgEfX19VL5v//7v90Hlgg0AM1AY4xts/6V1R2ek2QGpm5dMXUT kXUEkHCSk5PdYyEK91pHp4Rz5coVlWwR9jwO7VYqKyufeOKJR5z3PxovX8/M zBQVFcXHxyNLP/7446jz85//HDs0TvSNLI0SSZi/+c1vfBxFBnWjMaphaKTZ /eQXnpMUeUzdumLqJiLrYMIJCeMNldDf34/Ea5zEz8XAwIDM9QevvPLKrrdS LjpvnMzPz0dQl4u0aWlpCwsLPiqjAWiGsVUhuZUyAnhOUuQxdeuKqZuIrCMk f813/0u9P3V0SjhjY2PGfIt36vKsHHfIzNnZ2XLRWy5cX79+HVu510QmT0lJ UeNPnn/+efd7J13I83HQDGOr0Eiz+8kvPCcp8pi6dcXUTUTWEUDCAYRAFWCw gNXA6miTcPCpfvHiRZVvL1y4MDw8jNzr43K0QJ2kpKTvfOc7Kgr+5Cc/eeWV V44cOXLgwIFf//rXciulePbZZ2/duuU+O4p7nsdu0QA0QzUJzYuWf3p4TlLk MXXriqmbiKwjsITT5bxymOzk42qhP3WsnHAmJyeXlpb8rFxfX2+8sFxbW7vr 5W5lamoqLy/v97///be//W33URA//OEPT5482dDQ4M+u1IVuNMDYHjQvrH0V QjwnKfKYunXF1E1E1hFwwgk5Cyac6urqsrIyP4P33bt3keWMl7tl4m6PUwh6 Mz8/393djUBYXFxcUlKC5Dw8POz/5ov3JgzEoY0XutEwNC/c3RUqPCcp8pi6 dcXUTUTWwYTjQ1tbG5rkf/CuqqoyXl5GbJbJTHYdExIqOJBMXYJDG1uChoW7 r0KI5yRFHlO3rpi6icg6mHB8GBgYQPCuq6vzM3ivrq6eP39eZd2kpKT29nZk 4ImJiQgEbxwCB8LhcFAcWjUDTYqKaboVnpMUeUzdumLqJiLrYMLxYXp6GpF7 c3NTBW98gM/NzfnYBInXeJE5LS3NbrcjCU9NTYU7deMQOBAOh4Ma24AmRazH QoLnJEUeU7eumLqJyDqYcHxYXV0tKSnZ3t7G57YE75qamurqah+f4XgpOzvb GHozMjJkgDdScZiueGO3ErlxIBzOeHQ0Jur+xeE5SZHH1K0rpm4isg4mHGV+ fn5kZGRtbU2VIG8jdct9iCi/desWWjg7O+t7P0tLS8ZxJpCbmysTCU5OToY8 eGOHMmkJDoEDGY+LZvg/B4t18JwMFZzSNpsNX8TQ8tOnT6enp7e0tDgcjgB2 NTc3h1+HELZtc3NzYWEhhDsMElO3rpi6icg6mHB2nOm6o6NDbpzE/41Jtba2 dnp6Wq51V1VV+TnG2263n71fVlbW4OAgsvH4+LiPZ1buFXaFHWK32Lnx4e8C zQhzz4UFz8ng4ZRuaGg4cODAPjcnT55cWVnZ066Kioqw4cbGRqiaNzY2Jj/c UO0weEzdumLqJqKQwCfJ7Ozsnv4BdeeecMxlSsIZGhqqqKhYXl5GwHC58bCz s7Onp0ciN1KHjPFubm7edZ9tbW0uGfjy5cvY1agTknyQF72xOXYie8NusXOX w6EBYeuw8OI5GTybzSYZOzk5ubu7Gx8U/f39xcXFR44cQWFiYuL6+rqfu9ra 2pJdhTB146srdniTqZvCj6mbiII3MzNTWVkpqaCjoyPg/TDh7DgvaA8ODsoy MsbY2FhfX9/ExASWEcjRKoncUmHTadd93r59+8qVKy5JGBGopqZmZGREZhTE DzGA7I1NsKHMEIhdYYfGecIFvhqEr7vCjedkkBYWFuLj4xFrEbNd8sbk5KRc AE9JSfEzijB1U1Rj6iaiICF33bp1C4kL/w7a7fabfgw29oYJB/D9BSF5x5mo q6ur0Qz5RoNAu7y83N7evte8gb1hc/yMKioqzrrJyMjo7OwcvWdqampubm7X +I0KqCZ3TQrsxOXeSeHPpXgr4zkZpOzsbBlJgrzh/mpvb6+kaPmmie+V+Hro MqO7zWZDoUx6mZ6eLvUvXbp048YNvFpUVIRXcSrm5ubiKDjlkKKNH0EFBQWo gDNWlczPz6MEG2IZXYqtsMMTJ06gEIcITzfsDVO3rpi6iShIHR0dTU1NahX/ Yk5OTgb2wfK6hYWuw3bR19dXV1e3vb2NtIykLYNMkLeRmVtaWva6NxW5Z2Zm sNrT0+N+Lfqs8xZLmc1bQfxAksFWswZYRaFMxK1gQ5cbJ9W1dBwu9B0UWWaf d76Y3Td+OXXqFDJtfX29twpHjx5FBcnAra2tWEb6NVa4ePEiCvG1rr+/3zgm PCkpCa8mJCRgOS4uDv8/fvy4XDzHPtX9Dq+99hpK8KGkdoiTFiXSgThLjfuU L7ymY+rWFVM3EQUJUbC2tlaWHQ5HSUkJsqI8QHx+fn5PuzI7xfgShp7zDDEb IXloaKi5udl45e3OnTs377+5clcukVsgPKenp7uHZLh06VJ5eTnijQw78QEV UA2VsYnHXeEQAf/Jw1LMPu98Mbtvdre5uSlpdmBgwFudy5cvowJOmJ3dUveO pxEmKnVLBZTjnJQx5FLBd+re4QgTiiCmbiIKEpIh/sHCP5f4twx522az2e12 /BtXV1eHcuMfdndldorxJXwd6G5sbOymcw6TqakpVSgzB/ofZT1GboFkUllZ 6TEti/Pnz2dnZ9+4cQPV8HNsdMICVlGIl1xmI3SBaiEcdmsus887X8zum93h 119C8vT0tLc6OEXVheuAU3dhYaGqj2+mcul7cXFxh6mbrISpm4gC4DKbNFJi dXV1Q0ODMWbjs6W8vHxPk1eYnWJ8CX0n+tTT04POrK2tVR/RCwsLKPFzlhgf kVvBS9euXfMRngOAHfo4YjQy+7zzxey+2R1OWgnJxtDrAoEZFdLS0naCSN34 ODJugm+FKOzt7d1h6iYrYeomoj3xMZs0/oWtqKgwVu7u7kYU93PP/gTF2IF+ HhwcvOm8iXJiYmJ8fLyyshL96c+2e+pJ7LmgoCD4vI2dYFdBv2/SCjKGhGTJ zB5JqM7JydkJInW7DGa7evUqCvG7s8PUTVbC1E1Ee+JjNmkZaoKXZBWfLajZ 19fnz24ZuT1ClkAOQbcgcg8MDPjzcR1YTy4uLjY2Nnob7+0DNsGG8qd8Inep qalq2LY7fFWX0SByp7CkbpfKiYmJu6Zu41gsdVC5mVdSt/EWCZQzdZMpmLqJ aE98zCYNiNmogH9JV1ZW2tvbbTabx+nCXDBy7zi/sxif/27k/8Ovg+9JfGnq 7e3FDy4vLy8tLc09ZqMQL6ECqqlvWETe4HNAcrL7F3Cc2BkZGXjpwIED8qAc RGusIjMb68i0JL5Tt3GCSrx06NAhFMpoN6RrNTOhqK6uZuomUzB1E9Ge+JhN Gv9uLi4uqsfl1NfX3717d9cdMnKLkpKSIOcKDkdPIuTgx7rshAWshmrPFDtu 3Lgh04zgE0OdQktLSzLZCDQ2Nkqh3W7H6v79+9XvQnFxsXGMCkK4rKrbiiV1 qyfLo8L169f3OefflgopKSlYvXr1qqzOz8/LMzFV6q6oqMBqdnZ2JPrCP0zd umLqJqI9kSeS+5hNWkaeeLts64KRWwkydbMnybLwmZCTkyNpGdn77Nmz8mAa YbzIjO/yMn13fHz8+fPnkZxVNTUyXCocPHhQLolL6t7nnKMbR0lKSpJX1ZgT fP2XCojZ586dkwEtxtSNzC8lr732mkVmmGfq1hVTNxHtCv8U9vf3y2fFyspK WGeTjlnBpG72JFnc1tYWvqQbUzRcvHjRfR7vycnJ06dPSwVk7/z8fJzbWFa3 Ere1tcmYE3A4HJK6kbeRtKUQcd24WxwavxoqbJ86dUquqOMoUmF9fV3mPAH8 GkakP3bB1K0rpm4i8g2RW6YEVIludHQ0rLNJx46xsbGye9Ab6EBZ9n/ilx32 JEUPfErgazs+N6anp33P6I5qiN/48PH4KspRAZF7xzBzIMIMduttVBuyN47r 47IAdoht/b+HIqyYunXF1E1EPkjkLney2+2qvKurK9yzSceC5eVlec4joDc6 OzvlyY8+Jjd2wZ6kGOdxvu5ox9StK6ZuIvJGRW6ZkATUSxGbTToYra2tOTk5 CLFhPYqf8GYRDHxM+oEO8T9si7D2JHZe68bHrMvkw9zcHHoPZ+O1a9fq6urk Ii2FBFM3RRGmbiLyyBi5sYqMXVVV5VInYrNJBwDfBfbv349/jtHCsB5oV3fv 3kVP3nLCex8eHvZYba+pO9w9efbs2X1uUBiOY+mto6NDjUNWd+1x0sVQSUpK iouL0+wJTUzdumLqJiKPEOpU5N5xXqwzPgHHKJKzSfsJn2xqZgNzU/fW1ha+ rbS1ta2vr6NVCGB4+x6HnuKby/T0tJ+7DXdPotly91lRUdFNg/r6+nAcTmP4 Zirn4bVr10ZGRurq6o4fP26cyI7IHVO3rpi6icgj5C5jOMRqWVlZV1dXwDuM 5AhkmbDXCql7cnISb1ndOIbsjU4YGxsLZp8R6MmpqSl03bFjx8K0/9iRn5+P nrxw4YL6cioPRoyPj/d2qyARU7eumLqJyE/9/f0lJSWBRYVIRm4cS56RIU+F Njd12+328vJy48Nlampq3B/S57/I9GRbW9s+70/xJj/hX1j5k8HCwoIqxL+5 OAFGR0ctMl0GWRBTt66YuonIT+vr6wh73oYl+xDJyL26uioP0UA7r1y5Ynrq np+fx3s3TsRdV1cnD/cMQMR6Uh4mePXq1YKCgvPnz1+8eLG0tNSfJ42SkQwv OXXqlMx4X+LU09PDf3bJN6ZuXTF1E5H/Ojo6kL78fO6kiPDUdpK0cTi1HPnU ja8nxtWxsTHjx2x1dbWagxHl/t8FFsmelD8TuMDXmSCfWR9r8NUP/ZaUlJSW lmbsSXQvv8KQD0zdumLqJiL/ORwOm83W0NDg5x/HIxy5m5qaEGnOnDkjw2BM Sd2rq6t4vz6eFoQOlMkM8dnb3Nzs57eYSPYkfriHDh2SR2b39PTMz893dnbK PYCnT5/maGT/4VuqStrXrl1DN5aXl8fHx8uzFM1uHVkXU7eumLqJyN3ExIS3 KIgM5udggwhH7rm5uQMHDsTFxamZQExJ3RsbG/KuvQVv9N7k5KSK3IuLi7vu M8I9idTd39/f2NiIbxCqEKeEpMehoaEItEEPOPek0/CDVoX40iqFPp6TSDGO qVtXTN1E5K6kpMTHWALfj3IWkX9mYkpKivw1v/2eM2fOoCQvLw/LEZse2eFw yMODvAVvlKNvLRu5fUhMTER/7ulp9TFuYGBAArbxWyq+y0ghXjWxbWRlTN26 YuomIne+U/euTAmKx44dcx+KrPT29kamGTKxObJ3S0uLx+Bts9lKnSwbuTc3 N+fn590f2ZOcnMzUvSf4qcnp5/IwSplMPmLnJEUdpm5dMXUTkbtgUrdZ12aR Zm/c78SJE8g2aWlpWPYx0Dq0RkZGamtrd5z/bnoM3nJHqmUjN4yOjkpWnJqa UoXr6+sHDx5EoboVlHaFsC2juI2XtVEoj6pUj6AicsHUrSumbiISY2NjZfcg 7CF4y/Kerm1aZzjEjknjuvGJqiZn9hi87969a+XIveO81i2xMCMjQ2Yax/9z cnJQcvLkSd5NuSfFxcUyeaAMMtne3i4qKpISs5tG1sXUrSumbiISy8vLo6Oj I04Ie52dnaNO7iMNvLFU5N4xb+ZAI29XvH0zvScbGxvlcveRI0eQt2VERFxc XMAzjcespaUlmUAePZmVlXXu3Dl5MCV+y8xuGlkXU7eumLqJYhl+/V1GnArk Pf/DtkC4velkkcgNGRkZSDhtbW3mNkMFb+MDCn2wSE+izcah8omJif5PLU5G +D578eLF/fv3S0+ePXuWX17IN6ZuXTF1E8WslZWVqqoqpMGuri6X+bf3lLoR DsvLyyUoqln7yAgfs/39/buOzbBaT+KsWFxcnJqacnnuDwUA/9riZ2qcjJHI G6ZuXTF1E8WshoYG5O3x8XFkbJdJzCorK/1JfUiJtbW1N+/hnAwBY08SkcLU rSumbqIY5HA4NjY2ysrK5F65wcFBJL09PbPDJSWCzWaL2Q+ThYWF+fn5wLZl TxKRC6ZuXTF1E8WgZqeuri5ZxYcAgl9NTY0/nwbuKVEEM793tOvs7Gxpadnr VuxJIvKIqVtXTN1EMWh2dhbpDsFblSwvL9+6dcv3VMyrq6sdHR3V98gEg6Kq qiqWP0n2mrrZk0TkA1O3rpi6iWITUh9itvFB1QMDAyjx824vmdqOl2dFYNe6 BXuSiFwwdeuKqZsoNuF3v7y83DirHj4Kurq6/HkGippNGmkzZi/PysB4cctJ rfo/TwV7kojcMXXriqmbKGZNT08HMCm08QEuQ0NDMXt5dmtrSz1UqK6urqam ZvQefK76swf2JBF5xNStK6ZuoljW2tpqs9n8TIk7bs9MRFasqqpymetbbx5n ru7s7NzrEzDZk0TkDVO3rpi6iWKZDJNwmazbG/fHlCMrxs7lWXxUtrW1oQdq a2uN4+F3nIPk95S6Y7wnicg3pm5dMXUTxbj5+fk9jeU2jkjBcuxcnh0ZGamp qZmenq6urnaZZbGnp6ezs9PP/bAnicg3pm5dMXUT0a48BsWYsr6+3tTUJG9/ ZWUFXdHf3x/AftiTRLQrpm5dMXUTkW8MiuPj49XV1bW1tepytN1uR4csLy/v aT/sSSLyB1O3rpi6icgHBsUd5yyLNpsN/aCG4sjTPMH/z0/2JBH5ialbV0zd ROQNg6KCHkBXDA8PqxJ5mufg4KA/m7Mnich/TN26YuomIo8YFF20t7eXlZU5 HA5Vgsi9uLi464bsSSLaE6ZuXTF1E5E7BkV3yNvl5eUdHR172oo9SUR7xdSt K6ZuInLBoOjN5OTknp7myZ4kogAwdeuKqZuIjBgUfWttba2urvanJnuSiALD 1K0rpm4iUhgUd7WxscGx3EQUVkzdumLqJiLBoBgq7EkiCgZTt66Yuoloh0Ex dNiTRBQkpm5dMXUTEYNiqLAniSh4TN26YuominEMiqHCniSikGDq1hVTN1Es Y1AMFfYkEYUKU7eumLqJYhaDYqiwJ4kohJi6dcXUTRSbGBRDhT1JRKHF1K0r pm6iGMSgGCrsSSIKOaZuXTF1E8UaBsVQYU8SUTgwdeuKqZsopjAohgp7kojC hKlbV0zdRLGDQTFU2JNEFD5M3bpi6iaKEQyKocKeJKKwYurWFVM3USxgUAwV 9iQRhRtTt66Yuom0x6AYKuxJIooApm5dMXUT6S26guLrFhZdPUlE0YupW1dM 3UQai7qgaHay9iW6epKIohdTt66Yuol0FXWRe8faqTu6epKIohdTt66Yuom0 FI2Re8faqTu6epKIohdTt66Yuon0E6WRe+f+1H3TAoztMbtviChWMHXriqmb SDPRG7l3mLqJiJi69cXUTaSTyEfuzc3NDk9GRkakAj5k+vr6GhoampqaxsbG sCrl29vbWEVhTU1Na2vr3NzcDlM3ERFTt76Yuom0YcpVbnyGeMyr7e3t8mpl ZSVW8f+ysjIsNDc3y4b9/f3S2urqatlkenraPXV3mYSpm4jMwtStK6ZuIj2Y OLBka2tr8567d++WlJSgGUtLS3jJbrdLcpZqTU1NWF1cXNze3i4tLcXy6uoq XkLexnJDQwNTNxERU7eumLqJNBBk5B4dHe3u7kZg7unpqaurw/+xHFhLWltb 0ZLBwUFZtdls5eXlyNuyurKyMjIygp2jBK1F8Eb8RrnD4cBWVVVVTN1EREzd umLqJop2wV/lrq2txR7KysqwE0RfLCM8y0tra2t4VS5cw/LyMmK5t0w+NTUl l6wlSyNaY7WxsXFiYqK9vb25udlut6sEjnK82tvbOz8/39bWhmUkf6ZuIiKm bl0xdRNFteAjt1x2xk56enocDgcCc3l5ORK4vIrAjJdKS0sRvBG5ZWz2+Pi4 x11Jel9YWJDV1dVV9/HeCO3ymYPobiyvrKzExxFTNxERU7eumLqJoldIxnLf uXMHO0FglgvU+EDADrGqKgwNDUnwlpHYWPW4H8RyvFpTU+OyZ7mgjY+alZUV ieUI7TiK3GWJ+kj7FRUVWO7s7GTqJiJi6tYVUzdRlArV7ZMjIyPGLL28vIzV jo4OY53u7m6Vgb3tR+qMjo6qErmaXVJSsrm5KSVjY2MykmR+fl4Gn0g5PoVs NptLymXqJqLYxNStK6ZuomgUwhlLELCxq5WVFVmVYKxm295x5nAZWKKGmrjv ROYkARWwd+6N60acViWzs7MSpGVuk+HhYfWShHambiIipm5dMXUTRZ3QThJY VVVVUlIiw0tAAufi4qKsqsg9NDSEkOwteK+treGlpqYml3IZUqLqS7QeHx+X 4eKdnZ3GZviTuisqKpKdsOA7NgdZk6mbiMzC1K0rpm6i6BLayC3PuGloaFAl NTU12Ln6WJCZtNX4EwneU1NTLvuRi9i9vb0u5bJ5eXm53W6XiUpsNpvD4cBx ZUhJR0dHf39/XV2d3GjpO3WnpKQ8+uijjzhhAavegnTwNZm6icgsTN26Yuom iiIhfxSOjK/u6+uTVRkTgvRrrOMyT6DHaQMljY+Njbm/ND4+rgao1NbWqqEs WJCwLVpaWtbX132k7oqKChWPVUj2eB07yJpM3URkLqZuXTF1E0ULE58+GaTt 7W1kdYfD4f6SzG2iRoP7SN3JycmPuEGhe5YOsiZTNxGZi6lbV0zdRFEheiP3 nuw1dSclJQWTurE5UzcRWQ1Tt66YuomsL0Yi947P1O0+GgQuXrzY2toa2AgT bIjNXXbIESZEZDqmbl0xdRNZXOxE7h2fqbvr/jsf4aWXXrp69Wp2dnZjY6NL oj537pwxS2PVpQI2wYbYHDsxRm7eTUlEpmPq1hVTN5F14Pexvb39hhMWsGqp yK1mFwwf36m7694sf4mJifg/MnN+fj7+n5GRUV1dbaxWXl6OCkecsIBV46uo jE3U5mqHnDmQiKyAqVtXTN1EFjExMXHixIl9BkePHs3JybFI5IaNjY3u7u6x sTGP90WGxK6pWykoKEBgrqqqQv9cdSotLVWvot9Q0uCEBayql1BN6mNDbI4F 7Mp9/0zdRGQWpm5dMXUTWcHi4uLhw4eRtE+fPl3shAWsxsfH2+12s1v3/62v r8vjbBobG8MRv/1P3dXV1QjMeXl5WEaT5Np1UVFRR0dHXV0dlnNzc6UmFrCK QryECnJtHJvgJWyOVZfr5EzdRGQupm5dMXUTWUFBQQEydmpq6vDw8KgTFrCK Qrxkduvuo4K3CG389j91d3Z2ZmVlITM3NTVhtb6+XlbznLBgs9mkJhaM5aiG yijHhrKKXTF1E5F1MHXriqmbyAqOHz+OgI34N2qAVRQePny402La2tpKS0tv 3g/hFmkWLwWzZ/9TN5SUlMhAEVltaWnJzc29fPnyoUOH0tLS0BIpx8KFCxdQ eOnSJVRANSmXoSnYicedM3UTkVmYunXF1E1kBXFxcQjYQ0NDxtSNVRTu37// ZszYU+pGfkZszszM7OjokJL29nYEbHTmyZMn1bgRLGAVhYjiqCCF2AQbYnMV wpm6icgimLp1xdRNZAUJCQkI2I2NjcbUjVUUIjFOWMzt27dv3bplTMs2m621 tXVwcDDIPe8pdXcZ7qlUJTk5Odg2NTVV7q+UeyexikLjPZU+7qNk6iYiczF1 64qpm8gKKioqELBPnTrV29srkRsLWEUhXjK7dfeZn58vKSmRUFpTU2O321dX V0O1872mbuM9laDuo0SovnLlymtOWMCquqdSavq4j5Kpm4jMxdStK6ZuIitw OBxnzpyRSUtSnbCAVRSGb5q+AEjkDnnYVvaaujvvv6dSpiiRC+8I2wec0tPT sSr3VKJC1273UTJ1E5G5mLp1xdRNZBFra2uZmZnG+bqxikKz2/X/ra+vDw0N hSNsK3tN3V337qlEX509exaddu7cudraWnnuZIaTPL8ShXgJFVBNgrq3+yiZ uonIXEzdumLqJrKUxcXFHicsmN0WEwSQultaWuLi4ozPf4+Pj0eoLiwsbHfC AlZRaKyDTbzdR8nUTUTmYurWFVM3EVlHAKm7oqLi0UcffeR+WVlZxjqZmZku FbCJ8fnvTN1EZB1M3bpi6iYi6wggdScnJz/iBoV7rcPUTUQWwdStK6ZuIrIO pm4iIqZuXTF1E5F1hGSEifvoEX/qMHUTkUUwdeuKqZuIrCOA1A0pKSkqVGMB q4HVYeomIitg6tYVUzcRWUdgqbvLeTU72cnHFWx/6jB1E5HpmLp1xdRNRNYR cOoOOaZuIjILU7eumLqJyDqYuomImLp1xdRNRNbB1E1ExNStK6ZuIrIOpm4i IqZuXTF1E5F1MHUTETF164qpmyjkpqenc3JyioqKInzcubm57e3tCB80tJi6 iYiYunXF1E0Ucrdv3963b9+RI0cidkSEbYR8HHRjYyNiBw0Hpm4iIqZuXTF1 E4Vc5FP31tbWPif9Ure5mLqJKPKYunXF1E0UckzdAWPqJiJi6tYVUzdRyLmk 7tHR0StXrrS1tXV1dWHh+PHjZ8+evXHjhkrIUqGlpaWhoSE1NRUVsrKyUF/9 bg4NDaFCVVWV8Sg2mw2F2HZiYiI9PV1S96VLl7DnSL7Z0GLqJiJi6tYVUzdR yLmkboRtrB44cGDf/ZC95eZHqSDi4+Nfe+01WS4tLZU9tLa2YhUZ23iUixcv orCzs7O/v9+426SkpAi/3xB63cLM7hsiihVM3bpi6iYKOY+pG1JTUwcGBhwO R0NDg5SMjIy4VFhbW0NJX19fXFwcSrCrnd1S946mI0ysxuy+IaJYwdStK6Zu opDzmLrj4+NXVlZUncTERBR2dHSoCojZi4uLqoLMSZKXl7fD1G0NZvcNEcUK pm5dMXUThZzH1J2QkGCsIyOxm5ubVYULFy4YKwwMDMgolB2mbmswu2+IKFYw deuKqZso5DymbpdQnZmZicKmpiZVITs721hhampKrpDvxFLqJiIipm5dMXUT hVxgqTsjI8NYYXBwEIWJiYk791J3enq6sYKMUWHqJiLSDFO3rpi6iUIusNR9 4sQJ4y+jzWZD4bVr17CMaC33WqpXt7e3ZVIUpm4iIs0wdeuKqZso5AJL3aBm 5J6bmzt8+DBK2trasGq327G8f//+iYkJqVBcXCybSOpGCJfV2dnZiL1NIiIK B6ZuXTF1E4VcwKkb0tLSsrOzDx48iOWcnBypvLm5efToURnmff78+RMnTqj6 krpBKmBD4yVxIiKKOkzdumLqJgo5mX7k2LFjstrT07PP+dRIYx1EaxS2tLTs GCY5QWCWLI0IXVBQgLCt6k9OTp4+fVpeRfbOz8+/efMmlru7u6VCW1ubehCP w+GI1HslIqIQY+rWFVM3kemMF8NXV1enpqbkmZXuVlZWEL+NadwI5ajAyE1E FNWYunXF1E1kOo9DUIiIKDYxdeuKqZvIdEzd4bC8vHz16tXGxkazG0JEtDdM 3bpi6iYyXU9PT1xcnMt03BQMfLIlJyfju0x+fr7ZbSEi2humbl0xdRORZhYX Fy9cuCA3ljJ1E1HUYerWFVM3EemkoaFB5l1k6iaiKMXUrSumbiLSiYTtlJSU 69evM3UTUTRi6tYVUzcR6QR5u6ura3t7u7q6mqmbiKIRU7eumLqJSEtM3UQU pZi6dcXUTURaYuomoijF1K0rpm4i0hJTNxFFKaZuXTF1E5GWmLqJKEoxdeuK qZuItMTUTURRiqlbV0zdRKQlpm4iilJM3bpi6iYiLTF1E1GUYurWFVM3EWmJ qZuIohRTt66YuolIS7W1tUjdhYWFZjeEiGhvmLp1xdRNREREZB1M3bpi6iYi IiKyDqZuXTF1ExEREVkHU7eumLqJiIiIrIOpW1dM3URERETWwdStK6ZuIiIi Iutg6tYVUzcRERGRdTB164qpm4iIiMg6mLp1xdRNREREZB1M3bpi6iYiIiKy DqZuXTF1ExEREVkHU7eumLqJiIiIrIOpW1dM3URERETWwdStK6ZuIiIiIutg 6tYVUzcRERGRdTB164qpm4iIiMg6mLp1xdRNREREZB1M3bpi6iYiIiKyDqZu XTF1ExEREVkHU7eumLqJiIiIrIOpW1dM3VY2SkQaMfsThYiiA1O3rpi6rczs jEBEoWT2JwoRRQembl0xdRMRERFZB1O3rpi6iYiIiKyDqVtXTN1ERERE1sHU rSumbiIiIiLrYOrWFVM3ERERkXUwdeuKqZuIiIjIOpi6dcXUTURERGQdTN26 YuomIiIisg6mbl0xdRMRERFZB1O3rpi6iYiIiKyDqVtXTN1ERERE1sHUrSum bosbGhrKcrLZbGa3ZXeqtUpOTo7dbje7XURERFGDqVtXTN0Wl5CQ8IDT5z// eXNb8uqrr/7MaWJiwlsd1VoX73//+w8ePBj59hAREUUdpm5dMXVbnHVS94c+ 9CFpSXNzs7c63lK3ePnllyPcHiIioqjD1K0rpm6L8yd1j4+P9/T0bG5u+tjP 1tbW6Ogo/h9wS/aUuj/xiU+UlZXdvHnz17/+9cc+9jEVvN2vS6NJHR0dd+7c 2bUBd+/e7e/vdzgcfrZnZGQEL/nuFiIiIqth6tYVU7fFuafutra2NzmhBMn2 L//yL6XCQw89dPz4cbXhf/zHf0i1vLy8X/7yl29+85tR5x3veMfXvva13t5e Va24uFiqfeELX/BW+PLLL3/wgx9Uyfnd7343Vs+fP+9Pa3ec6Vdtm56ebjzK ww8//Md//Mcof8Mb3vDpT3/6hRdeMH4vUG8hMzPzK1/5iryFS5cu+W7P4uLi d77znbe//e3yKvb/1a9+FZ0Q5A+CiIgoMpi6dcXUbXHuObalpUUNlv6TP/kT l1EcN2/elGpf/vKXpQRp1qXOxz/+8aWlJalWWFgohZ/73OfUQV0Kn3nmGbfR Ig8YE76P1orPfvazUv7cc89JSUlJiaRoFwjM29vbLm/hIx/5iKqA1O2jPdgW 7879VaT6GzduhPDnQkREFCZM3bpi6rY4H6n7Aedl3hdeeOHrX//6H/zBH0jJ I488ItVUZIVPfepTL7300je/+U1V8sMf/lCq+ZO6i4uLsfnb3vY2Kfz+97+P 1YaGBn9aC0j4Dz30kJQnJiaiZHp6+q1vfauU/P3f/z0Kn376adW2w4cPu78F vFO8taeeeqq6utpHe/CqFP75n/95U1PT7du3n3zySZeeISIisjKmbl0xdVuc 79RdVVUlhT/5yU+k5G/+5m+kREXWD33oQwsLC1L44x//WAo//elPS4k/qVvs aVy3tHZra6u0tPQb3/iGanB/fz/Kc3NzZfUDH/iAGtH985//XAoffvhhl7fw 0Y9+dHJy0uVYHttz+fJl9QbHx8dR4nA4Wp04gSEREUUFpm5dMXVbnI/U/fa3 v10NxsjLy5PCT3ziE1KiIivSrNpbW1ubFD744IP40e+ELXU/4BzU8cD9vvrV r0q1F198UUq+973vqW0bGhqk8F3vepfLW/jtb3/rfiyP7VlZWVFX0d/4xjfi Owi+j1y7dk11FBERkcUxdeuKqdvifKTuD37wg6paZWWlt9R98uRJVW1paUll 4MHBwZ1wpm4Xjz/+uLpT8t///d+l8He/+53admpqSlWenp42voWkpCT3Y3lr Dyq/973vdTn6X/3VX7W2tu7W2UREROZj6tYVU7fFBZ+61Q2MOz6vdX/sYx9T 1dQ4jYBTN3Lvs07PP/98enr62NiYsRoKpdpTTz2lChsbG6Xwne98p8tb2FPq hrW1tdOnT3/9619///vfr4L3N7/5TW/NJiIisg6mbl0xdVtc8Kn7wx/+sBo7 /dOf/lQKP/WpT0mJuv3wgXtXv+GJJ57wkboLCgr8b61HWVlZUg37XF5elsJf /vKXUvgv//IvLm/Bd+o2tmdoaKjJaX5+fsc5pcm5c+ek2pvf/OZgpisnIiKK DKZuXTF1W1zwqfsB562FL7300mOPPaZKfvCDH0i1tbW1N73pTSqfv/DCC488 8oiqZkzdn/nMZ9Tejh071tXV5U9rPZqcnHzLW94iNb/4xS8mJib+4he/UOPA 4+PjXd6Cx9TtsT2//vWvVQM6Ozunp6dffvllKfm7v/s7v3udiIjINEzdumLq trjgU/cHPvCBB+730Y9+1PgsyK9//esuFRC/3VO3mv9E7Gm+bnfXr19Xsx0a IfOrE9J36vbYnqGhIZXnXZw7d86fDiciIjIXU7eumLotLjExUUKjenZka2ur ysaqWlVVlRR+8pOflBIVWZGEEVAffPBBLL/1rW9FeUdHh/EQ6+vrX/va1/7w D/9Q6v/DP/xDfn6+LBuvDy8vLz/zzDPvec971G79aa0PeXl5X/rSl9RU3vi+ 8LOf/UxGmwt106XHwOytPX19ff/0T/+kLuC/8Y1vRJ8UFxfv2h4iIiIrYOrW FVO3rlwuFDscjtu3b+P/3upvbGw0NTXJHNe+bW5uGrNxkLC39vb2ubm5YPbg 3h6UdHV1tbS04DtFcA0kIiKKKKZuXTF168r38AwiIiKyJqZuXTF164qpm4iI KBoxdeuKqVtXCNtPOzU0NJjdFiIiIvIXU7eumLqJiIiIrIOpW1dM3URERETW wdStK6ZuIiIiIutg6tYVUzcRERGRdTB164qpm4iIiMg6mLp1xdRNREREZB1M 3bpi6iYiIiKyDqZuXTF1ExEREVkHU7eumLqJotHc3Fwwm+O3fnNz08/KS0tL KysrwRwu1qBvt7a2zG4FEUUrpm5dMXVb3MLCQmNj48DAgNkNIQt55plnHnjg gd///veBbY4UnZqaWldX50/lqampffv2xcXFLS4uBnY40929e7fRu5mZGWPl 1dXVioqKzMzM9PT0srIyb796w8PDqHblypXs7Oza2trZ2Vnjq21tbcnJyfjl DeO7IiJ9MXXriqnb4trb25F5Ll++bMrRJyYmioqKOjs7TTl6TOnt7X3llVeK i4v9qfzwww8jdX/7298O4EAIz0ePHkVW9PMXf2hoaJ8T4ncAh7OC8fHxfd7V 19ermoODgwcPHnSpcP36deOF6+3t7dLSUpc68fHxXV1dxjoFBQWHDh0K8k8S RBSbmLp1xdRtceam7ra2Nhw9Ly/PlKPHlIyMDATpH/3oR/5UnpycPH/+fACJ Dr/vSU7+/9YjQN6+fRtxdK/Hso7V1dVqN1VVVXFxcTi9cZJLNXwfOXDgAEqS k5MbGhpGRkbKysr279+PkpaWFrW3np4eSdp4FT8IVMNXGClZXl5W1dBvaWlp CQkJGxsbkX7DRBTlmLp1xdRtcb5TN/5l3/XHF8zPl6k7YvaUugN27do1/ECH h4e9VQj5aOQQ7tB9V319fcXFxfgt8Fi/srKyubnZ297wVQJdcezYMXwGSkl+ fj5KTp486XA4VDWbzYZCfE9RJfh1QInxrxJoAII6Cl0Oh0yOwqtXr3prIRGR R0zdumLqtjiX1L2xsfH666+np6ePj4/j/wcPHoyLizt79mx/f7/aBGEDdQYG BpqamrCwf//+U6dOISQYR5liGS9duXLFeCyEBBTm5uaqAx05cgRHP3DgAJbP nDljrDw1NeUyIDZGJCYm/uM//mNtbe3hw4e/8IUvvPe973344Yflemlqaurf /u3fvvvd70Y5XnXZsKCg4Kc//eknPvGJP/3TP/3KV76SmZkp5fgd/OIXv/jx j38cqft973sfdv5v//ZvxmNVV1e/+uqrn//859/xjnfgZ4rylJQUlMtPCiX/ /M///K//+q937txRxzp37hwKX3zxRWMDZIS2MUAqExMTOMfw48bpdOHChZqa mtXVVXlpc3MTP32cY7I6NDT0uhfGi7o4P/F+kWBx+qG1yK4q3KrG7Hr+VFRU YLctLS04e3EOY1d4X729varCxYsX5Vuhe6wtLy/HS9jc455RH/2ACsbB7SdO nECJ3W431kSzp6en5+fnVQl+79wDtnydqaqqcjkQOhPl6DTf75SIyIipW1dM 3Rbnkrrv3r2LVYRtycOIdmpkKXKO1CkqKsLq6dOnpVzVQQJZX1+XOggSKEGo Mx5rZGQEhYiOO87Ujch06NAhGbOKZexQ1USkdzlo7Hj++ecRj7/0pS/h/+96 17secHrnO9/53HPPYeGtb33rW97yFilESlRbIc1KIVI66sgyehUvORwO5HDZ 1R/90R99+MMf/uu//mvjsT772c9Kfey5sbER5b/73e+wiiAq1WSY9/e//31Z HRsbe9vb3vbggw8ax0XAzZs3ZVyEyztCJsSPGC8hduKUkJ8svmTJtWU0T84B qTw4OHjyfjIqA4xnl4yOxruWExUyMjJUNvbz/Dl//rw6vdVpjH2Ojo5KBZyl yOEozM/PNwZvuUCNE9445MMI0d3lQre8Tewcn4doFb533Lp1q6enB79xLtsi b6NmWlqa2nZpaen48eP7PA19x36keT7eJhGRC6ZuXTF1W5zH1C2BGf/W7zgv KkpSysnJkTqSuiErK0uuWE5OTiJgoCQ7O1vq7Jq6hbcRJp2dnXII4zX2GCFJ +E1vehO6d8c5l4VKxb/97W83NzdXVlZefPFFrH7qU5+STex2+xvf+MY3v/nN FRUVO86Ah8CMCh/5yEfUbj2OMJFjwauvvmq83OqSuicmJiS0l5eXY/XLX/6y bOLScvki5j5CW4Yl47SR4LqwsCARV676uqRuF3hVTr/q6mopwUfKa6+9JlFT cimaJztU9y36c/4g88t3AYRnGcGOk1b2bBzdgagvoztU8N41cqMavlO4XOiW PwQkJCTIOBOj1tZW4+Zra2v4HZEvtui0wsJC+eXyOAwM7x0vIZN7e5tERO6Y unXF1G1x3lI3/jVXdRB49jlvAZNVSd2IZMbhqUhQcsFQVoNM3UhEeAlti8EB q5KEn3zySVWSlpaGkg996EOqZGZm5g1OEjvRXQh1xhEgKHn/+9+PrfCDkBIf qfvRRx91aYNL6oYrV66g5C/+4i8uXryIhc9//vPuo6Dl+rP7uA6cXShHWFUl iLKIrDKht+/ULYndeB17bGxMQq/x3GhpadnnHOGs3v6u548MigbjyKiSkhL3 fKuCd0FBwa6RG7q6uiQJGwe9qC8CcO7cudra2oaGBhkfAt3d3cY9DAwMGP/M JAnc4zyB+GosFfgxS0T+Y+rWFVO3xXlM3YhAxkwl19PUEFZJ3YWFhcb94Kcs Vw5l1uUgU3cskyRsHG9TV1eHEjUYWyCEo1ANuhDIezk5OQh1R48e/cAHPoAK asSvj9SdkJDg0gb31A2PPfaYXBh/6KGHbt++7bLJxsaGxD/35920trYa0yZO J2MY9pG68cbl+rBxRDfCqhTWGOBsROFrr73mvhNvmpqasMmlS5eMhVVVVR4H bKCfZZz2rpEbvwhyzd84YeDOvVNdftfU84NQWb5WGH8p5BuEtK2ioqK0tPTs 2bPSRfim4HI4fMBKZTVOnohoV0zdumLqtjiPqfvIkSPGOnNzc+6pG/HJZVcI QmrYAFN3wCQJI6CqEo+p+8Mf/rAxdScnJ3/mM595wI0/qdt4LOExdc/Pzz/4 4IMof+6559ybrVK3e/xDyER6lDH8AjFS5XZvqXt4eHj//v0HDx50eUCMGuDk kf8PxJSpQlzuT7x69SoK5ZZSFzdu3JBDICf7uITe0dHhfqFb3o5s7jLBi4w8 wduXlqMbZb5BY2jHV+Dc3Nx9nu5UxVbevuwQEXnD1K0rpm6LCzh1uzxvBTlE BhjI8GBJ3S4zPMg9bkzdvgWQultbW9/whjc89NBDTz75ZHp6emVlZWdn5yc/ +cnQpm4kSUny73vf+zxO5S1x0dss38iTfX19N2/elFHKIDcOeEzdy8vLUs1l 6MXOvfsHL126NOZmT/feehyFLjctul9SxreGfc5RVTJgu7Cw0GPwxmcdznlU aGhocHlpcXFR3rVxINCO4duK/JEI7ZFfQJf9y+8gGId1AcL2Xr9uEBExdeuK qdviAk7diB/Gn6yMkkXwlrSAaOF+5VP+fO+eutU9mLQTUOpG2Mbys88+q15F NpOZTFxS91NPPeX7WMI9dff29r7lLW95z3ve84Mf/AAvfeMb33BvudyKqCYA Uf7HSa0iH8rkIfh2sOMpdW9tbUkFRHT3o8izLBMSEoy5FMuIry6Xl31QWdc4 REfGSLsMr9oxRG58F1BjvD0GbxlLg+junoHRAxLpOzo6jOXyVVTdDimpG79H Lm2QS+L79+93eSbO7Oys1PfzjRMR7TB164up2+ICTt1w48YN+eFiKxleovaD QCIjCpC0JT/MzMzIJG/G1K3uO3NJKWtra9euXcvPz3e5shcLAkjdssk3v/lN 9eovf/lLlxEm6EysfvKTn/R9LOGSuvET/NznPoeSixcv4mf9Z3/2Z1hOT093 2UrumnRJlfjJytR/ajoRlbplIId76r5+/fo+59R5Hj868Jb/X3tnH1xFdf5x IqkGsaM4CEoBtcgwpHUcq6W+zJQZGVqVUqwiFduhgB1bRYsIfVFAkmiA8GZi xBAhCQkSAiTBhLcYCGBAQgghQAJ54dVwScgLSXN/3L/a6fy+c5/hzGH33ptL CGRz+H7+yNw9e/bs2d2zuZ9z7rNn5VeVrVu3Sgoam0xnjepJSrvtR+TWEtAu L4VE3fREpdwyMu+5+uFKXbxxlpYtW+ZzoFuQiPSYmBgVM1NXVyeTeMvU6FKI HB16HKpwmHZKSgoSk5KSLGXKu3gsdSaEkMDQuk2F1u1wOmzdEk6wYMGCzz// XN5qvWjRImUmnitxs1Ka/C4vm+jWLSN1AIquW5w86Qag5Tfw4B1JB6y7oKAg JCREZt6eOHFieHg4Pt977726deOD5BkyZMgbb7zhb1+Cxbrnzp2LxVGjRski rFjKt0RiyFWzxwvJo44Alxh6KWqKViG/g1isW8U/x8fHJ1+NGpeGwEuTQ38N 6fBYGQeuqanRaxKg/ci0PDBzPVEmMEFfUqXs2LHDotyCPp2g5fCXLFniL9gD 6eq5SPQpoNDSH0H5El6i1y3CO29JRkbGunXrZJAcmcvLyy1lSrS5mlaREEKC gdZtKrRuhyMTminjbW5uluE4PY/L5RIHkEWxbnzRQzlEfmAR8jpLfSs4BoRB veJE3vpnH68rLCyU4T44uUqUgXFUwxIEeyswc+ZMOK0+eolThJQxY8bo2QYP HoxEFW+Qnp7+wAMPyPh2v379IHWwdHzWn92Dx8owNYTZ376EOXPmIF36WWgh oaGhYWFh+ksVx48fjwzjxo3Tt2poaIiOjoYiWkIv3G53Xl6eep0NmgRai/JY mYVDxUgcOXIkwg/6A4OlpaU4RunHoRHiEPQuQLvtBzaLDeWVQAp5KSSOV6Wg v2BXbkHEW59jEDntZVpAFwO7lrtG2jxuB3v5ODoJ11FA1+2vyPF4o9Nx992C twkh5HqgdZsKrds8xLrlZ3Rc39ra2gDxtLj6yACZD1AgJA0OYynEEgxMggHm WVVVFeC84VTX1dUFvhwWUJo9ZNpnoudKYLPltS8KaDN6cJ04BzskFq3LPnO4 p5Paj4SL+1uLw/e563ZBtaurq2tqagIHojc2NqLThJz+4mSkG6u/i4cQQoKB 1m0qtG7z0K2bEAvp6ekxMTEBZrQmnUJTU9PSpUtTUlJuwTdJEUKuE1q3qdC6 zYPWTQKA+z03NxdCaJlnm3QiLpcrNjY2MzOTEwYSQjoArdtUaN3mUV5enpeX Z5/TmBBFcXHxLfgk7E2jtLTU56t8CCEkGGjdpkLrJoQQQghxDrRuU6F1E0II IYQ4B1q3qdC6CSGEEEKcA63bVGjdhBBCCCHOgdZtKrRuQgghhBDnQOs2FVo3 IYQQQohzoHWbCq2bEEIIIcQ50LpNhdZNCCGEEOIcaN2mQusmpDtSW1t7PZvj rg/+tYn19fUNDQ3Xs7tbDZxbt9vd1bUghHRXaN2mQuvuLly6dKmwsHDjxo3J yclr167dsmUL3z55yzJt2rQePXrMnTu3Y5vDoleuXLlnz55gMp85cyYiIiIq Kqqurq5ju+tympqa9vrn3LlzeubGxsb8/HzcYqmpqdu3by8vL/dZZmVlJbKl paWtW7du9+7d33//vb72wIEDCQkJLpfrBh4VIcRcaN2mQuvuFlRXV0dHR0fY yMzM7Oqqkc6hrKzso48+ysnJCSbz6NGjYd2vvfZaB3YEeY6JiYErBnnjnzhx Qhob9LsDu3MCJ0+etN87im+//VblPH78uP1G27Rpkz5wffny5W3btlnyfPzx x4cOHdLzZGVlzZ8//zp/kiCE3JrQuk2F1u18zp8/v3DhQnyzL1myBAJQWlq6 a9euVatWydf9N99809UVJJ3AV199BZF+8803g8l8+vTpxMTEDhgd7vcvvAR/ 10Mgjx49Ch291n05h8bGxgIbO3fujIqKwh104MAByYb+yCeffIKUhISEwsLC qqqq7du3R0ZGImX//v2qtMOHD8uth7W4EMiGLoykXLx4UWXDecNNGhcX19LS crMPmBDSzaF1mwqt2+HgK/vTTz/FF3psbOylS5f0VTt27JDven2Q7VoJJvoU jaTD5ZMguSbr7jAbNmxAg6msrPSXodOjkTuxQHtRR44cycnJgd/6zI8b5Lvv vvNXGroSOBWLFi1SzTszMxMpy5Yta21tVdnQq0Ui+ikqZePGjUjRf5VABSDq SLTsDk6OxDVr1virISGE+ITWbSq0bodTXl6OL+7o6GhL4KjH+10PVcPa1NRU SXG5XJDztLQ0PRu++pG4fv16PfHkyZNZWVkQjMjIyC+//BJqoas1dAWblJWV ZWdnL1myBLs4fvx4SUkJEteuXWupBjZHenV1tSyeOXPGEihrGPHx8c8888zu 3bsXLFjw1FNP9evXb/To0TJeunLlyieffLJv375Ix1rLhjjhb7311rBhwwYP Hjx27Fh1JnHmn3322aFDh8K6+/fvj8JfeOEFfV8FBQXz5s0bMWLE3XffvW/f Po/3nCNdrilSnnvuueeff/7ChQtqXytWrEDirFmz9ApIhLYukIpTp06tXr16 4cKFUVFRSUlJu3btamxslFVtbW24vsuXL5fFEydOxPpBH9QN3MA8wbWT/Px8 FLt//34oLvqeKArHhWapMiQnJ+OIoMF2rc3Ly5O+qs+SkR/nARn04PalS5ci paKiQs+Jap89e/b8+fMqBbebXbClO7Nz507LjnAykY6TFvhICSFEh9ZtKrRu hyMD2nbXFSorK2W8ThahB1iEqul5qqqqkAghVCnIJsGrUCyJXQEQeKUuX3/9 NVLgjSpmFYXU1dXZf0aHrSFl/vz54lTSRwCQrs49D87h/fffhx6PHDkSf/v0 6dPDyz333DN9+nR86N27d1hYmCTCEtVWONWSCEtHHvkMKcWq1tZWeLgUdeed dw4aNOixxx7T9/X4449LfpS8d+9epM+ePRuLEFHJJmHekydPlsWampq77ror NDRUj4sAmzdvlrgIyxHBCXGJsQraicYjV/Dzzz+XsWVUT9qAZEb/a9nVSFQG UL/FtNvAgmwniYmJKnBatUaUqbp48Hx4eIT3AQddvGWAGsqtt1UdqLtloFsO E4Xj/yFqhX7Hli1bDh8+3NTUZNkWvo2cq1atUtvW19cvXrw4wlfoO8qJ4PMX hJBrhNZtKrRuhyOj2XBvn2ubm5t1Ew7GunHFZfgaJiDaAHMWpVGPlYl1g7y8 PN06JJi8sLBQpWzdulWXipKSEtnw2LFjnXYKHIaYcM+ePdPT0z3ejo+y4g8+ +KCtra2hoWHWrFlYDA8Pl00qKipuu+22O+64Iz8/3+MVPAgzMjz00EOqWJ8R JrIvMG/ePH241WLduIIi7bheWBwzZoxsYqn5Z599Jj9bWNIlLBkXXcTV5XJJ e5BRX4t1W8BaEfWCggJJCaaBBdNO4PzSF4A8SwQ7mreUrEd3QPUlukOJd7vK jWzoU1gGuuWHgLi4OIkz0SkqKtI3x02HuwnpOJ84adnZ2bB3LK5evdq+L+mW wsn9HSYhhNihdZsKrdvhLFu2DN/a+q/qFmSQTewlGOuuqakRJ9HHBvfv3x/h DUCVRbFuS6SKxzsfmmXYXKqHXcgiTAl5iouLDQ5kFROeNGmSSkFnBCkDBw5U KefOnQvxItqJ0wKp0yNAkHL//fdjK1wySQlg3a+++qqlDhbrBrhYSHnkkUeS k5PxYcSIEfYoaBl/tsd1QBcjrn4sFyoLZZUJvQNbtxi7Po4dTAMLpp1IUDTQ 59+TXp7Fb5V4Z2Vltavc4NChQ2LCetCL6giAFStW7N69G71LiQ8BpaWlegnl 5eVq7F2AgfucJ7C+vl4y8N8sISR4aN2mQut2OBJ9qqZZsDN//nxkkN/cg7Fu uISM6e3SyM7OjvDOkSJ5xLohHpZ9QcAknECURsbxIN4GO7YdMWFYlkrZs2cP UlQwtgAJR6LlAVj4XkZGBqQuJiZmwIAByKAifgNYNy6WpQ526wYTJkyQgfFe vXodPXrUsklLS4von/19N0VFRbpt4rLqFzSAdePApS3pEd3BNLBg2LdvHzZJ SUnRE3fu3OkzYAPnWe6UdpUb/+5kzF+fMNBzpUcpSq/eH4TM0q3Qe5rSg5C6 5efnb9u2bfny5XKK7FPo4x+sZFZx8oQQ0i60blOhdTucrKwsfGXn5ub6XAuD ku90sbtgrFtFj/hEfEPy6JEkivXr16tVMvB4q01dKCYMQVUpPq170KBBunUn JCQ8+uijPWwEY936vgSf1n3+/PnQ0FCkT58+3V5tZd12/cNFhz1K902ARipv 92fdlZWVkZGR9ud8g2lgwSBThVieT1yzZg0S5ZFSC7hHZBfw5ADdwIMHD0bY Bro9Vx6RiLBN8CKRJzh8qTlOo8w3qEu72+2W+8L+pCq28tfZIYQQf9C6TYXW 7XBkxC8xMVGlwNNgDhLUIVMHqyFEsW7LvA3y5Jqybnm8KyUlpcaGerQtgHVX VFSo0mRKw1vtBXwdsO6ioqKQkJBevXpNmjQpNTV1x44dJSUlw4cP71zrhkmK yffv39/nVN6ii/5m+YZPHjlyZPPmzRKlDOrr6z1+rPvixYuSzRJ64QmugQWD zyh0iaeyDymj14B0nBAJ2M7OzvYp3vhfh7vDZ9tWDwvrgUAerbcir+ZEffB5 4cKFlvJxViWbPuugR+sXB9/dIIQQWrep0LodjkRx6GNr8gM3DAoCIA+XqRlO IAz28Uz5UV5Zt7xqMC4uTtcGfIZdqNG/ANaNnLJTGBr+fvnllzfiqJ1MB6wb so3P77zzjloLN5OZTCzWPWXKlMD7EuzWXVZWFhYWdt99902dOhWrXn75ZXvN 5cKpCUAU/+dFLcIPZfIQeYbXbt1ut1syQNHtewmmgbWLcl09REdipFETS8i6 Um70BVSMt0/xllgaqLvdgXEGROnRpdXTpdOqHocU646OjrbUQYbEIyMjLe/E +f777yV/kAdOCCEeWre50Lqdj8TE4gu9uLhYvEW9C0+G3WRM0uN1G4kTgGmL FZw7d06mblPWDS2Rp+q2bt2qtpLZhpOTkyUlgHUDeR22FGuZtbi5uRlFZWZm Wkb8TKID1i2bvPLKK2rtjBkzLBEmOGlYHD58eOB9CRbrxrV+4oknkIIr2NTU 9OCDD+KzmsVdIU9NWqwS/imx+mo6EWXdEshht+5NmzZFeKfO8/mvI5gG1m47 Ebm1BLTLLzv67z4eTbnVXaA/XKmLN86SPPzrr2FLRHpMTIyKmamrq5NJvNV0 9yhEjg49DlU4TDslJQWJSUlJljLlXTyWOhNCSGBo3aZC63Y+uEDyYg6RH/iz Po+xZe41iYYVK5Zf2yWuQH8cDJvIe64XL14MEYJmiNXX1NRIhsDWLcN3UhnL 04ISDxNxfa/LdDgdsO6CgoKQkBCZeXvixInh4eH4fO+99+rWjQ+SZ8iQIW+8 8Ya/fQkW6547dy4WR40aJYuwYinfEokhVwctxFKadOsivK9bgl6KmqL9yC8m FutW8c/x8fHJV6MaQ7sNrN12gjOGtTBzPVGeI9CfcZDZ7HXlFvTpBC2Hv2TJ En/BHkhXz0WiTwGFlv4IypfwEr1uEd55SzIyMtAFlkFyZC4vL7eUKdHmalpF QggJBlq3qdC6uwUwH6iReh2JTklJiZ4T5gANUDnlXX72UbjS0lJ4hQg5dAga rxuaqILPZ9YEmS9CJqzWkaF1WJYlONYkZs6cCafVRy/RPUHKmDFj9GyDBw9G ooo3wLl64IEHZHy7X79+OPmwdHzWn92Dx8owNYTZ376EOXPmIF0C+NEAQkND w8LC9Jcqjh8/HhnGjRunb9XQ0BAdHQ1FtIReuN3uvLw89TobNB7ot/JYmYVD xUhIZJFP9AcGAzewdtsJbBYbyiuBFNL31Bs8bgq7cgsi3vocg/IYgqVMC7jR sGvpMkiPFTeOvXwcnYTrKKDr9lfkeLzR6XB4g28HQsiNgNZtKrTubgSUAOpS Xl5eU1MDnZOA7fnz59u/03FNa2trm5ub2y0Q2ewTOwdGvQbF5ytOLEHCRAeX r6qqKsD5gQ/X1dW1e+F0UJo9ZNpnoudKYLPltS8KaLPL5erEeSADNLBOaScS Lu5vLQ7/Wtu2gGpXV1fjLgsciN7Y2IhOE3L6i5ORDq/+Lh5CCAkGWrep0Lq7 L1CO7du3b9q0KcA7dDqdS5cu5eTkSIDBLTVNtzGkp6fHxMQEmNGadApNTU1L ly5NSUnhbUIIuVZo3aZC6ybBo8JxIyMj7e9hId0C3O+5ubkQQss826QTcblc sbGxmZmZnDCQENIBaN2mQusmwVNaWrpy5cqMjAw9hJh0R4qLiw1+4rXLwZ0S 4LEIQggJDK3bVGjdhBBCCCHOgdZtKrRuQgghhBDnQOs2FVo3IYQQQohzoHWb Cq2bEEIIIcQ50LpNhdZNCCGEEOIcaN2mQusmhBBCCHEOtG5ToXU7mWpCiEF0 9X8UQkj3gNZtKrRuJ9PVjkAI6Uy6+j8KIaR7QOs2FVo3IYQQQohzoHWbCq2b EEIIIcQ50LpNhdZNCCGEEOIcaN2mQusmhBBCCHEOtG5ToXUTQgghhDgHWrep 0LoJIYQQQpwDrdtUaN2EEEIIIc6B1m0qtG5CCCGEEOdA6zYVWjchhBBCiHOg dZsKrbu7gCu1du3a9957b8KECX/4wx/+9a9/FRQUdHWlOocTJ06kX01GRkZF RUVX14sQQgjpAmjdpkLr7hYcPHjw/vvv72Hj6aefvnjxYgcK3Lt377teVqxY 0em11Zk3b57s6NSpU/7yxMXF2Q8N4JCjo6Nvfn0IIYSQLoTWbSq0budz4sQJ n8otjBw5sqWl5VrLTEhIkM1ffPHFG1FnxcCBA2VH3333nb88/qxbmDNnzk2u DyGEENKF0LpNhdbtcGpra3/84x+LKPbu3Xv27Nnffvttbm7u5MmTlZe+9NJL 11pskNbtdrurq6vxt8P1vybrHjZs2Pbt2zdv3vzPf/5zyJAh6gDt49Ko0sGD By9cuNBuBZqamo4dO9ba2hp8fQghhJAuhNZtKrRuh7NgwQIln2lpafqqqVOn qlVFRUVIycnJ6enlqaeeUtksiQ0NDT/60Y/uvvtu2fD222/H4k9/+lPJPG7c OMm8cePGGTNm3HHHHciDzBD7srIyVSaKkmwo3F/inDlzULKqYd++fbGYmJho P0Zl3SNGjFCJVVVVatvU1FT9cEaPHv3DH/4Q6SEhIaj5zJkz9X6BOoS1a9eO HTtWDiElJSX4+hBCCCFdCK3bVGjdDucXv/iFWOKYMWMsq1paWvr06SNrP/jg A6RkZ2fL4hNPPKGyWRLr6ursURz33HOPZMZeJAU2a8kzdOjQ+vp6yYaiJBGF qx1ZEqdNm2bf0eLFi+3H6NO6weOPPy7p06dPl5StW7eKRVt4/fXXL1++bDmE hx56SGWAdQdfH0IIIaQLoXWbCq3b4dx3333ihxEREfa1v/zlL2XthAkTPMFZ d2tr64cffvjCCy9I4pAhQ7D48ccfS2alrCA8PByrXnnlFZXy5z//WbIFY905 OTnY/K677pLEyZMnY7GwsNB+FD6tG4bfq1cvSY+Pj0fK2bNne/fuLSnojCDx 7bffVnVbsGCB/RD69u07fvz4KVOmFBQUBF8fQgghpAuhdZsKrdvJuN1uJZAb NmywZ/jrX/8qa0eOHOkJzroFf3HdSlkHDhzocrkk8S9/+YskqkCUYKxbuKa4 brFuHPW2bdtefvlldezHjh1D+vr162VxwIABKqL7b3/7mySOHj3acggPP/zw 6dOnLftiXDchhBCHQ+s2FVq3k7l8+fIPfvADscSkpCR7hj/+8Y+y9te//rWn U60bNqsSDxw4IImhoaFoMJ4bZt09vKHaPa7mt7/9rWSbNWuWpPzpT39S2xYW Fkpinz59LIcgUTcWaN2EEEIcDq3bVGjdDkdZ4vvvv29fqyKfJ02a5OlU6162 bJlKrK+vVw58/Phxz420bgsTJ05UT0r+5je/kcTZs2erbc+cOaMynz17Vj+E L774IsD5pHUTQghxJrRuU6F1O5yxY8eKJf7kJz9R098Jx44du/3222VtbGys RxPsIUOGqGyrV6/ugHWrBxg9Ace6k5OTVbahQ4dep3X369fvHS/oYqSmptbU 1OjZkCjZpkyZohL37t0rifYHQmndhBBCuiO0blOhdTuctLQ0NZb75ptvqvTm 5uZHH31U0nv27CnDvAUFBSqzDEp7tCgUn9b9s5/9TN+dUtZBgwap2Om33npL EsPDwyXlueeek5TXX39dUioqKlRwiE/rzsrK8neM/uYwsZCeni7ZUKZ6I+eM GTMkcdSoUZZDCGzdAepDCCGEdCG0blOhdTucy5cv//73v1cuLdNTT506VX9b pdJLqDgMXGkzco4fP15l0607IyNDpb/99tsqaFyfAAT7+vDDDydMmKBSsF/J 9ve//10JPzb5xz/+MXjwYJVNt27VNUBpixYtOnTokP0Yg7Tu06dPh4WFSc5n n302Pj7+vffeU6pvn4bFp3UHUx9CCCGkC6F1mwqt2/ngGv3qV7/q4YeoqCg9 8+9+9ztLBui33bpramqUwfbwFZ4xYMAASzkPP/ywGv3et2+fZd5s2K/aRLdu Nf+JcE3zddvZtGmTerxUB50L1YwDW3cw9SGEEEK6EFq3qdC6uwVNTU1Tp069 8847dWOE5cbExFhyXrp06aWXXlLx3k8//XRmZqZ8/vnPf67nhBs/88wzoaGh PbwB1ZKolBUmDEGVtb1790b6wYMH9c2zsrJUtEavXr2Q/8UXX5RFuLHKdvHi xWnTpqlZx5HNfnTx8fGyVn+lpj82btw4cuRINZX3sGHD3n33XYk2F9RDlytW rLBvHkx9CCGEkC6E1m0qtO5uBIxxz549GRkZMM+ioiL9NegWWlpa9u3bd/Lk yWCKhajbX+woA8Wtra1Hjx61PMWpU1VVhZqghHb30tbWprvxdYLSiouLa2tr r6eETqwPIYQQ0lnQuk2F1k10AodnEEIIIeRGQ+s2FVo30aF1E0IIIV0LrdtU aN1EB7L9tpfCwsKurgshhBByK0LrNhVaNyGEEEKIc6B1mwqtmxBCCCHEOdC6 TYXWTQghhBDiHGjdpkLrJoQQQghxDrRuU6F1E0IIIYQ4B1q3qdC6CSGEEEKc A63bVGjdhBBCCCHOgdZtKrRuQgghhBDnQOs2FVo3Id2R2tra69kcd31bW1uQ mevr6xsaGq5nd7caOLdut7ura0EI6a7Quk2F1t1duHTpUmFh4caNG1evXr12 7dotW7acOnWqqytFuoZp06b16NFj7ty5HdscFr1y5co9e/YEk/nMmTMRERFR UVF1dXUd212X09TUtNc/586d0zM3Njbm5+fjFktNTd2+fXt5ebnPMisrK5Et LS1t3bp1u3fv/v777/W1Bw4cSEhIcLlcN/CoCCHmQus2FVp3t6C6ujo6OjrC RmZmZgdKg65//fXXJSUlnV5PATaC8tFNuEHlG0lZWdlHH32Uk5MTTObRo0fD ul977bUO7AjyHBMTA1cM8sY/ceKENDbodwd25wROnjxpv3cU3377rcp5/Phx +422adMmfeD68uXL27Zts+T5+OOPDx06pOfJysqaP3/+df4kQQi5NaF1mwqt 2/mcP39+4cKF+GZfunQpbPbgwYO7du1KSkqSr/u8vLxrLfDAgQPYcOPGjTei tgD1RPkXLly4QeUbyVdffQWRfvPNN4PJfPr06cTExA4YHe73L7wEf9dDII8e PQodvdZ9OYfGxsYCGzt37oyKikJDxe0g2dAf+eSTT5CSkJBQWFhYVVW1ffv2 yMhIpOzfv1+VdvjwYbn1sBYXAtnQhZGUixcvqmw4b6tWrYqLi2tpabnZB0wI 6ebQuk2F1u1w8JX96aef4gs9NjbWMnoMbZDvenj4NZUZjHVfT6ugdXeAa7Lu DrNhwwZcmsrKSn8ZOj0auRMLtBd15MiRnJwc+K3P/Dt27Pjuu+/8lYauBE7F okWL8D9QUjIzM5GybNmy1tZWle2bb75BIvopKgU3DlL0XyVQAYg6Ei27g5Mj cc2aNf5qSAghPqF1mwqt2+GUl5fjizs6Oto+sImv8rVr12Lt6tWrJcXlckHO 09LS9Gz46kfi+vXrPV6Hx2cZOf/kk0/w+fPPP5dsUBQsYnf79u3Dh8jISNg+ 1EKPTW23fHQE8Fn6AkuWLMHnoqIiyXbmzBlLAG03JT4+/plnntm9e/eCBQue euqpfv36jR49WsZLV65c+eSTT/bt2xfpWGvZMCsr66233ho2bNjgwYPHjh2L ayfpuAefffbZoUOHwrr79++Pwl944QV9XwUFBfPmzRsxYsTdd9+Nq4P0L7/8 EulyzpHy3HPPPf/883o3Z8WKFUicNWuWXgGJ0NYFUnHq1Cm0IjSMqKiopKSk Xbt2NTY2yqq2tjZcx+XLl8viiRMnYv2gD+qePHkSxwuDRUNCbeGuSm5VZdpt D/n5+Sh2//79aIdojSgKx1VWVqYyJCcnS//RrrV5eXnSV/VZMvLjPCCDHtwu vcWKigo9J6p99uzZ8+fPq5TU1FS7YEt3Bu3fsiP5TQonLfCREkKIDq3bVGjd DgcyjG9tZWgWqqqqsBayJIvQAyxC1ex5IIQer3VDhObPny+RqPj82WefSbav v/4aiVgUZ4Y0ygd9jL3d8qFJKFM2XLRoET6LnEjfAUDGOvPsdAXvv/8+9Hjk yJH426dPnx5e7rnnnunTp+ND7969w8LCJBGWqLaCzUoiLB155DPOD1a1trbC w6WoO++8c9CgQY899pi+r8cff1zyo+S9e/ciffbs2ViEiEo2CfOePHmyLNbU 1Nx1112hoaF6XATYvHmzxEVYjghOiMYgIUy4uHKl0B2TsWVUT1qLZD5+/Piy q5GoDKC3E4mOxlFLFw989dVXyo2DbA+JiYkqcFo1SJRZXV0tGdCe4eER3gcc dPGWAWo0XT3kQwfqbhnolsNE4fh/iFqh37Fly5bDhw83NTVZtkWTRs5Vq1ap bevr6xcvXhzhK/Qd5UR09PkLQsgtC63bVGjdDmfNmjX41oZ7+1wLzxEVkfkl 2rViwWeEiVg3SE9Pl3HO06dPQ0uQsm7dOskTZPn2CJOSkhIp/NixYx07D85B TLhnz544UR7vXBbKij/44IO2traGhoZZs2ZhMTw8XDapqKi47bbb7rjjDvRK PF7BgzAjw0MPPaSK9RlhIvsC8+bN04dbLdZ96tQpkXYJ8h8zZoxsYqm5dKns EdoSlowGIOLqcrlEcWXU12LdFrBWRL2goEBS8C9lyZIlopripaieFKieWwym PcD5pS8AeZYfetD8pGQ9ugO3gER3KPFuV7mRDX0Ky0C3/BAQFxcncSY66vca obm5Ga1duqg4adnZ2XKbqJ+cdHDsWAUn93eYhBBih9ZtKrRuhyNDx/qv6hbE Q44ePerpDOuGyOlBrfAuGWaUxQ5bNwwKOy0uLjYgwFVMeNKkSSpl1apVSBk4 cKBKOXfuXIgX0U4cPqTOckLuv/9+bIVTKikBrPvVV1+11MFi3SAtLQ0pjzzy SHJyMj6MGDHCHgUt48/2uA7oItIhqyoFKgtllQm9A1u3GLs+jl1TUyPSq1/r /fv3R3gjnNXht9seJCga6DFOW7dutfutEu+srKx2lRscOnRITFgPelEdAbBi xYrdu3cXFhaqZ5ZLS0v1EsrLy9XYuwAD9zlPYH19vWTgv1lCSPDQuk2F1u1w JPpUTbNgAdIiKiW/uV+/dWdnZ+uJaBsy3nhNY+lmP00pJqwic8CePXuQooKx BUg4Ei0PwML3MjIyIHUxMTEDBgxABhXxG8C64+LiLHWwWzeYMGGCDIz36tVL emE6LS0ton/2990UFRXptnnq1CldhgNYNw5cxof1iG7IqiTu0kC7ivCG+tsL 8ce+ffuwSUpKip4oTxDbAzZwnuVOaVe50aRlzF+fMNBz5aYQpVfvD0Jm6Vbo zVt6EFK3/Pz8bdu2LV++XE4RegqW3eEfrGRWcfKEENIutG5ToXU7nKysLHxl 5+bm+lzrcrnkO10GqK/fuiFdll1AnyKuBBvQuj1XTBiCqlJ8WvegQYN0605I SHj00Ud72AjGuvV9CT6t+/z586GhoUifPn26vdrKuu36B8mEPUq0vwCNVN7u z7orKysjIyPR6bO8IEaFKvkk+BdiylQhlucTJeBKHim1gHtEdgFPDjCEfvDg QftAtxyObG6Z4EUiT3D4UnOcRplvUJd2t9u9fv36CF9PqmIrf50dQgjxB63b VGjdDkdG/BITE32uFYVQ465ixZZ5G+TJtSCt2/KWFjWWLkHFQZZP6/Zcbd1F RUUhISG9evWaNGlSamrqjh07SkpKhg8f3rnWDZMUk+/fv7/PqbxFF/3N8g2f PHLkyObNmyVKGdTX13v8WPfFixclmyX0wnPl+cGUlJQaG9f0LK3PKHR5aNE+ pIxeQ4Q3PkoCtrOzs32KN/7XyRw7hYWFllV1dXVy1JZ2q3or8nMP6hPhfX7Z Uj7Oqt7/VUC2r7W7QQghtG5ToXU7HBXdan9/N6RIzGfDhg2SAmGwj2fKj/J2 61bPSApi3ZAWvT3I3iHe4hhBli/WbRkCNYYOWDdkG5/feecdtRZuJjOZWKx7 ypQpgfcl2K27rKwsLCzsvvvumzp1Kla9/PLL9prLIwBqAhDF/3lRi/BDmTxE nuG1W7fb7ZYMUHT7XuRdlnFxcbqX4jP01TK8HADlunqIjsRIoyaWkHWl3OgL qBhvn+ItsTRQd7sD4wyI0ltmv5dOpXocUqwbd4SlDjIkHhkZaXknDu4CyR/k gRNCiIfWbS60bucjPowv9OLiYjWSdvbsWbEL+ICMSXq8biNxAjBhsYJz587J 1G26FaunyXT3UIEBubm50iSampokvEQ9vBZk+bKVPpzY3NyMrkFmZqZlJLA7 0gHrlk1eeeUVtXbGjBmWCBOcHCwOHz488L4Ei3XjWjzxxBNISU5OxlV78MEH 8Tk1NdWylTw1abFKtAGZ+k9NJ6KsWwI57Na9adOmCO/UeT7/deCQ5feRrVu3 SgqajUxnjepJSrvtQeTWEtAuL4W0/O6jlFvdBfrDlbp44yzJs8n2gW5BItJj YmJUh7Gurk66kDI1uhQiR4cehyocpp2SkoLEpKQkS5nyLh5/v1URQohPaN2m Qut2PrhA8mIOkR/4rQxayhiaZYpgiYaVH8Hl13aJK9CtWMbfABRauZlYt2Re sGABtpV3YS9atEj5TJDlqxdkI4NEIEicDIDw39iTdePpgHUXFBSEhITIzNsT J04MDw/H53vvvVe3bnyQPEOGDHnjjTf87UuwWPfcuXOxOGrUKFmEFUv5lkgM uQr2d5LKo44AjQF6KWqK6yu/aFisW8U/x8fHJ1+NGpeGwEvjQc8O6fBY6TbW 1NToNQnQHnDG9B9xBJnARH/GQWaz15Vb0KcTtBw+7h1/wR5IV89Fok8BhZb+ CMqX8BK9bhHeyK6MjAy0dhkkR+by8nJLmRJtrqZVJISQYKB1mwqtu1sA84Ea qdeRCImJiVVVVZacMAdogMop7/Kzj8IVFhbKIB6cWVLEuqEHEBVRJrgHNMwS ixtM+RcuXFizZo0UIr4hQ+KwLwOCvWfOnAmn1UcvcTKRMmbMGD3b4MGDkaji DdLT0x944AEZ3+7Xrx+kDpaOz/qze/BYGaaGMPvblzBnzhykS4B9SUlJaGho WFiY/lLF8ePHI8O4ceP0rRoaGtBNgyJaQi/cbndeXp56nQ0uLq678liZhUPF SBw5ciTCD/oDg+ht4RilR4aWgEPQuwDttgfYLDaUVwIppO+J41UpuCnsyi2I eOtzDCKnvUwLuNGwa2m6cnegYdvLx9Gpnq8AXbe/IsfjjU7HfWRAsyeE3Exo 3aZC6+5GQOGgLuXl5VA1ffDNDq5pbW1tc3NzgDxQL5iJCrUV65Yf35GIzQNE 4QZTPvwcGZTgWYKHb01w+dBRCnAecLpwZQOfWAsozR4y7TPRcyWw2fLaFwW0 2eVydeKc6pBYtBP7zOGeTmoPEi7uby0O3+eu2wXVrq6urqmpCRyI3tjYiDsR Of3FyUiH1P5EBiGEBIbWbSq0biLo1k0MJj09PSYmJsCM1qRTaGpqWrp0aUpK igFvhiKE3GRo3aZC6yYCrfsWAfd7bm4uhNDUSWacgMvlio2NzczM5ISBhJAO QOs2FVo3EcrLy/Py8uwzIRMjKS4uNuDJVsdSWlrq81U+hBASDLRuU6F1E0II IYQ4B1q3qdC6CSGEEEKcA63bVGjdhBBCCCHOgdZtKrRuQgghhBDnQOs2FVo3 IYQQQohzoHWbCq2bEEIIIcQ50LpNhdZNCCGEEOIcaN2mQusmhBBCCHEOtG5T oXUTQgghhDgHWrep0LoJIYQQQpwDrdtUaN2EEEIIIc6B1m0qtG5CCCGEEOdA 6zYVWjchhBBCiHOgdZsKrZsQQgghxDnQuk2F1k0IIYQQ4hxo3aZC6yaEEEII cQ60blOhdRNCCCGEOAdat6nQugkhhBBCnAOt21Ro3YQQQgghzoHWbSq0bkII IYQQ50DrNhVaNyGEEEKIc6B1mwqtmxBCCCHEOdC6TYXWTQghhBDiHGjdpkLr JoQQQghxDrRuU6F1E0IIIYQ4B1q3qdC6CSGEEEKcA63bVGjdhBBCCCHOgdZt KrRuQgghhBDnQOs2FVo3IYQQQohzoHWbCq2bEEIIIcQ50LpNhdZNCCGEEOIc aN2mQusmhBBCCHEOtG5ToXUTQgghhDgHWrep0LoJIYQQQpwDrdtUaN2EEEII Ic6B1m0qtG5CCCGEEOdA6zYVWjchhBBCiHOgdZsKrZsQQgghxDnQuk2F1k0I IYQQ4hxo3aZC6yaEEEIIcQ60blOhdRNCCCGEOAdat6nQugkhhBBCnAOt21Ro 3YQQQgghzoHWbSput5vWTQghhBDiEOBmbW1tXW2IpPO5fPkyriz+dnUTI4QQ Qgi51VFi1tWGSDqf//znPwwyIYQQQghxAhJeAj3rakMkNwTpVVG8CSGEEEK6 EMgYB7rN5n//+59cZfxlqAkhhBBCyE0GAiYy5na7IWZd7YbkBoLriyv+73// Wx6uJIQQQgghNwd5fBIaBvemct8i/Pe//4V7y3UnhBBCCCE3AVg3BAwa1tUm SAghhBBCCCGEEEIIIYQQQgghhBBCCCGEEEIIIaTj/D/lThOu "], {{0, 410.}, {489., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{489., 410.}, PlotRange->{{0, 489.}, {0, 410.}}]], "Output", CellChangeTimes->{{3.763326269456345*^9, 3.763326273194248*^9}, 3.763362625272789*^9, 3.763362750595644*^9, 3.763744000807453*^9, 3.763744635100074*^9, 3.7637449120534554`*^9, 3.763810069881571*^9, 3.763815816902182*^9, 3.763914843303464*^9, 3.763989560869391*^9, 3.764081112007636*^9, 3.764589538028203*^9, {3.764595666846972*^9, 3.764595694492392*^9}, 3.798186211738844*^9}, CellLabel->"Out[27]="] }, Open ]], Cell[TextData[{ "Each head uses an ", Cell[BoxData[ ButtonBox["AttentionLayer", BaseStyle->"Link", ButtonData->"ref/AttentionLayer"]], "InlineFormula"], " at its core:" }], "Text", CellChangeTimes->{ 3.763326314199917*^9, {3.7633841279375477`*^9, 3.763384148465852*^9}, { 3.763384212287424*^9, 3.763384212288083*^9}, {3.763384338303322*^9, 3.763384338304071*^9}, {3.763916423841723*^9, 3.763916448518124*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetExtract", "[", RowBox[{"net", ",", RowBox[{"{", RowBox[{ "\"\\"", ",", "1", ",", "1", ",", "\"\\"", ",", "1"}], "}"}]}], "]"}]], "Input", CellChangeTimes->{{3.763326325137995*^9, 3.7633263301452913`*^9}, { 3.7634176270525007`*^9, 3.763417627810041*^9}}, CellLabel->"In[28]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJztnQd4FdW690GQIiqoiByKiqhcUM5RPGI7R+9nOfciIArIPWAFORZEkWKh SaihCUKUIjWUECABAlIEghhCLyFASCANQiAJaSf5knuuivj9734f17ec2Xsy 2al7z//3PPBk3lmzZk1dv1lrzexW/QZ1f/uaGjVqDK2H/7r3/fT/DBnSd3iP Rph4+cOh773z4T/6d/pw2D/e+ceQR/vVQjAd//4H6f/3718JIYQQQgghhBBC CCGk0vnpp5/y8vIyMjIuXryYTgghxMfBzTwzMxM3dtzeq7qGIaQquXr1Ki6E S5cuZWdn//Of//y/hBBC/ALc0nFjx+09NzcXt/qqrm0IqQJw5l++fDknJwdX RFFRUTEhhBA/Ajd23N5xk8etnqpDHAgkPy8vD1dBVV+LhBBCKgrc5PNcVHWd Q0il8tNPP2VnZ1NyCCHE78GtHjd8jtUhjiI/P7+goIDdVYQQ4vfgVo8bPm77 VV3zEFJ5yLCcqr74CCGEVAYyUKeqax5CKg+OzCGEEOcgo3SquuYhpPKQt8ir +sojhBBSGcib5lVd8xBSedBzCCHEOdBziNOg5xBCiHOg5xCnQc8hhBDnQM8h ToOeQwghzoGeQ5wGPYcQQpwDPYc4DXoOIYQ4B3oOcRr0HEIIcQ70HOI06DmE EOIc6DnEadBzCCHEOdBziF+Sm5ubrfHLL7+oWfQcQghxDgbPQUSvHTBZFXUU IWXixx9/HPh7UlJS1Fx6jk9w+vTpkN8TGhp66tSpqi4XIcTHMHjO8uXL9dph 6dKlVVFNEVIm/ud//gdn75YtWy78xpUrV9Rc3/Kc6Ojo0aNHv+ICf2Cy+udc LvnPmjWrhjuaNm06ceLE8i1q1bJnz54PXMybN6+qy0KIH2LwnIKCAlU1TJo0 acmSJVVQSxFSNsRzDh8+7HauD3kOHjR6mkCwOudcXvl78hxh1KhR5VXUKmfu 3LmyUc8//3xVl4UQP8RifM6XX35JzyG+iH94TnR0tFkVhDK2vVRczuWYv/Kc Nm3abN26ddOmTZ9++mnr1q2V6iQlJenpCwsLExISkH9GRkaJmSNlSkqK14uf OHEiPj5ejyQmJh45csTivEJpT5486XaWHc+xWFwnOzsbZcvPzzcE9+3bZ9hd hDgHeg7xP/zDc0aPHu3JFjCreuZcjvkrz+nYsaMKQkWU5wQHB0swNzf3gw8+ uOGGG9Ss22+/Xe8DOnDgQC0XyGrFihV33XUX0tx9991eLL5hw4Y77rhDkt1/ //2xsbH79+9v3769RBo2bGjoU7t8+XL//v2Vnt12223du3dXypGVldW8eXMs JXPr1KmDSWRrc3HQrVs3KdvKlSu7du1at25dJMN9W+aGhYXde++9NWvWVAUe PHjwhQsXbB4CQvwDeg7xP/zDc1555RVPtoBZ1TPncszfreeABx98UOKDBg2S yJtvvum2b2vatGmSYN++fRJp0qTJddddJ38rz7G/eNOmTW+66SY9zZ///OcW LVoYFly+fLksiDPt2WefNefctm1bkY1Lly6Z5zZq1Mjm4qBz584SvPPOO1UC 8ZxFixa53a7HH3/cq6NKiK9CzyH+Bz2nqnIGFy9e/Pvf/15BnpORkVG/fn2J BwUFIfLdd9/J5A033IDI2rVru3TpIhEYkSylREV49NFH+/fvL5pU2sVbtmz5 2WefPfnkkypSs2bNt99++x//+MeNN94okd69e8uCgwcPlshdd921evXqDRs2 PPLIIxJ5+eWXkSA/P3/EiBGdOnWSYOvWrTE5fvx4m4sXa54DGjdujD3ct2/f yMhIzHr66acljjzPnj0bERGhShgTE1OWQ0yIb0HPIf6Hf3iORe/P0KFDy7IJ FdRvhSIdO3YMqjBgwICy52/wnMLCwi1btnTv3l1V6ydOnEA8PT090kVcXJws mJqaqgwkLy+v+PeiMmHCBH0tpV38yJEjUpjbbrtNIqpZCTohEdXx9Mc//lEi OBUlkpaWds011yDSvHlzVQZP43PsLK48p1WrVsnJyfri99xzj8zCnVzOloSE hP0u5JshhDgEeg7xP/zDcyxG8+La3Lp168WLF8s9Z6/HIaMwKNJaF6GhoWXP X3/fSo0wUbzwwgt64suXLy9dunTIkCH9+vXr0aOHSoZ4sSYqderUyc3NNa/L 5uK33nqrWqRbt24SXLVqlUQ2b94skX/7t3/DZE5OTq1atSTynxoiKkANhHbr OTYXV54zfPhww0Yp76rh6rDr2rXr+PHjz549a3P/E+I30HOI/+EfnlPs4e3s efPmKaM4duyYd9sSFBRkzll6gkqLasYBYl+oTN02GZXXe+W9e/cuLCxUKfv3 769G3RgwiIreilLGxV966SUJYqslsn37dt1z9u/f76n8wo4dO2RBt55jc3Hl OXPmzDFsV3JysuoRU8D0Bg8ebP8oEOIH0HOI/+E3nlOsfW2vT58+AwYMkPeA UP6jR4/qalHabLds2YIL/JNPPpHv+OEPTCJY2nz0ZhwUSXYs8sFkREREuXwn sEmTJvLZ0iFDhgQHBxtaJBYsWCDJbrrpprfeeguqtnz5cvue4/XiJXpORkaG NEPVrVs3zB3qqLn1HJuLW3iOsHfv3g8++ODhhx+uXbu22i4cqVIdC0J8Gq89 58qVKzt37sQt7siRI+a5sbGxmBUTE+NVNVVqcOvb6gL3JcMsXNEyq/J/xgs3 osjIyIULF37xxRdr1qzBjvrXv/5VyWVwJhXhOZvKibJcreIPqqLXHaNUDTvI AYsYrMaQeYmYm3EsMvcCT+9bGVBDgidPniwRfThNiZ7j9eIleg5o06aNRJCD WhD3BDkT1FdulOd06NBBL5udxT15DuZGuzh06JBEUlJSkL8kHjNmjMUuJcTP 8Npz/vu//1sesgYNGpSbm2uYO3HiRMxavHixtzXV/+f48eMhISG4o1qkWbly pRRm3bp1ehwyhgdVmXXu3LmyF8Y+u3fvxmOU4VeWxo8fbzYxP+CXX37BMYLL 4ayo6rL8L/7qOWaF8K5hx63SlMpP3DbjWGTuBTY957XXXpNkDz30UHh4+IoV K1q1amXfc7xe3I7n9O/fXyJ33HEHcoai4G4pkTvvvLOgoECShYaGqjUOGDBg 0aJF9hf35DmGF9Nw94Pz3H333RLZsGGDt4eFEN+j7J4DoBmGueXoOatXr0ZW eOSxSKM85/PPP9fj8fHxqpCV6TkRERFqvXhO/Oabbz777DOZhHehjqi0klQO hYWFsnXnz5+v6rL8L/7qOcUeLKJUDTsWPmNHUTw145SYeWmx6TlYXY3fU69e PTVuuUTP8XpxO56TnZ2t3gTXQf7qI4ey07BSNVd9P8fO4hb9VkqTDMB2cAmU 8mgQ4sOUi+d8+OGHhmaKcvQcPN3Y9xyDz4SEhFS+5+DuhB2CNX700UenT5+W 4C+//IKdKSWx3hZf5NKlS/ScyvEcTyJhv2HHQmZKtBSLZpwSMy8tQUFBUi8/ +uij1ilRxTdt2lQSt2rVCk8Z6gvD8gK1GtPbsmXL8lpcveEeFhYmkR07dkik Xbt2Khn2WJ8+fdS3BK+55poHHnhAvm+js27duscff1yG0DRp0sT+4uprP+bf AMXRmTRp0i233KIb1BtvvJGVlWW9PwnxM8rFcwCeL/S51p5z5cqVCxcuHDp0 CBpgMWwGbvDjjz/OnDkTWX311VfmwquUuuesX79eLa5aUTx5DkqCqjkmJgaF kfc03Zbk559/xqyffvpJJnNzc3GTj4uLKywsNCdWcnXgwAHD5syYMUNmpaWl eZe5ofB2duPVq1eRvz40KC8vD/mjeCdOnCgoKPCUuZ09I+U/deqUbFdCQoLh GKntMmSOGzg8BAtaj5vCCYBM1K+N44/U1NQjR45g2eTkZOxSt0tVnOd8q2Ht M9+aKBfPKbZ0iRIbdko0GU+ZWzfj2My8QklyUVWLl8iZM2fgS/JBHgtw+RcV FXm9uFtw69u9e3diYqIXyxLiB5TdcyAz+P+DDz7IyMhQcz15Dmo9PC599NFH uiPhoUOv90F0dLQhjZlPP/1UpRfPmTx58kCt6wp1LiZREmldMXjOpUuXsNTH H3+s54mtQPHMVTMeLTF3+/btx48fR/76IhAw7EOVEpX+oEGDEB8+fLiqnRWo KWSpVatWeZF5aXej2pmYGxgY+Kurisdznz5wCH/jodW7PTN69GjzGCQDP/zw g74I9sm2bdsGDx6sp5kwYQJUylxyMGLECCTAnpENGTlypL7g8uXL3S7l355j rRPWDTsltre4zbzEZhybmRNCSOVTds9BbTJmzBj8sWjRIjXXrecUFBSMHTtW loIMjB8/Xmqxga6eLzzaq5R4+rCuPcFnn32m0ovnoNaTUceiNKGhoQNdI5Ml ve45e/bs0SvooUOHDhkyRE2ahxtNmTJloOu1VpVGdwzIiWotycrKkmB4eLh5 p6GWF3+YNWuWF5mXdjcKMA3MgpPk5+erBRXQxatXr3q3Z9QYbwt0z4EmTZs2 Tc9cFHSgy6Nw0M17DMXDXDzJwgPNmbtd5NeK9xw7/VMbXZiXKpfLtkSjcNuw Y7O9Rc/cTjOOULWNOYQQ4omyew7qoH379snfFy5ckLluPWfevHmSDAZS/Fs/ yJkzZ0aNGoUgZOnHH39U9VT+b0gTx+zZs/N/j97hIp6zbNky6TNav379L7/8 Mnz4cPydlJQkK9U9B38jMnLkSFSUMJOrLuLj40UYUOdeunRJL7moiIjEd999 hz2G9NhYOIbEEZSUp0+flgiEwe1+k0YnWIcXmZd2NwriOSAgIED+x/HCWmAd WMvJkyf1xKXaM8hEjoXyUliW4TDhUKr0anh2YGBgenr6ry7zUW13+F9vEhTE c+QcwNpXrFhx/vx5bGBhYWFMTIynU9fvPceOVJgbdmy2t6jMbTbjCGzMIYRU T8ruOahGr1y5Ig0FqIJlrtlzUCtJenNriXoryu2neKR2tjMOeenSpbjNSlWe nJw80NVphW00e86vrk/umPunVL9SVFSUHhcVQZ2bmpqqx7HrpJcKGiAR6I3k YPAHhVgKqmzVq2U/81+92o3Kc0QXS3z1u1R7Rjh48KDMtRiHnJOTI6038DGD ianM58yZY1hKPEcsyP7nmPx4HLLCplcoV5k/f/7MmTOhW3Yy37x5MxLjXLX5 ujobcwgh1Zaye470HRw6dEjXCbPnrFmzRur3oqIic27SXICbsHlWqTzn6tWr 0qyB9ANdg16KPXiODpbKzc09ffr0999/L4kNvU6iIjNmzDAvK6+DAfGHyMhI mYR1uF2XSq8aOuxn/qtXu1Fvz9FbV+xQ4p4R7HjO0aNHJY3bli7ZCR9//LEh rjzHrV95okI9B7t3x44dO22DxHoTUHldufbVQhp2xowZg525fPnyEqUFCZYt W4bEWKTEZhyBjTmEkGpLeXnOL7/8MmHChIGuobOq2tI954svvpD0M9whs9y+ n1Uqz8HfGzZsUM0X8nqCJ8/JysrCcysKZhgWO9DUWmKhIrjDyyLSZ6faW1D1 uy2qbOwnn3ziRebe7UblOZ7Uy4z9PSPY8ZyNGzdKGnPnFAgLC1OHTI+L57jd ORZUqOdAnjMzM/Ntg8RYpNw9p7g0doE0uDomT55s3Qmld3UhMS5/m5mzMYcQ Um0pL8/5Vet9SEpKkpfB9QpXdMUafSSzYUH7npOeni65oQyYLHbnOVevXsUj tlrvRx99NG3aNJRWDSCx7zl4YFdbjcm0tDSZNAyqUYwePRpzkaEXmXu3G5Xn 2PkUc2n3jGDHc1atWiVp3Hac7dixQ+YmJibqcfEcObL2qVDPwUGRcUc2QWIs UhGeY18wxIgSEhIsXsUyjMZB4lJlzsYcQkj1pBw9B1VkYGDgQFeTzpw5cwa6 a88ZNmzYec+Yfz/i199qdvPIDR3dc379rdcsOjr6Vw+eo17CwkMrbun6C03S gmHfc6QjCci4aLU6t4nV9/R0FbGf+a9e7cZSeU5p94ygPMeicxBVvKRx256D ilLmum3Poed4wo5jGHTIPLrY06voXmROCCHVjXL0HHDy5EkJypvmuufIzzeA rKysUtVZ4k5Tp061SGPwnNTU1AMHDki7QbE7z5HPsEyfPt084LZUngO1E6fS B5agqJ4q9PDwcJm1d+9e7zL3YjeWynNKu2eE2NhYWYX5rXaFau4zfFHnV82Q 9e48oRp6TvXptyq2pxlmXdHFRsfQn+Vd5oQQUq0oX88B+gdS3L5vhTrU/AE9 C+QFpSFDhhR7/iSvwXN0ik2eg5LLJ2IMP/r5q2s0tcxy6znjx483jONV7Rjf fPONCh4/flyCM2fO1F8sSkxMFFUYNWrUzz//7F3mXuxG+57jxZ4RVG/dxo0b PWWel5enXh8zdF2pLZ0/f75hqWroOdVkHLLC2jQsXOXChQvY4djDn332Gf7A ZDlmTggh1YRy9xz9pzMNA2IXLFgg8cmTJyckJEhdjyopNjYWtTbmul2R+kAc hEcVNTs7e+/eveqz/6XynKtXr8rH+nB7l3EvP/30U3JysryiJbj1HFGUI0eO YNsLCwsjIyPlRWlU33rrimqHAdiHJ0+evHjxImRj6NChbvdYqTL3Yjfa9xwv 9owAnZNvBkJHjx07JscFy8bFxekfOlbDqseOHYsqEiXHlkIDZEshgYYvM/9a LT2njJT7JWwtG55ERXqvVq9ePdMF/nD7/rh3mRNCSPWh3D0HyCBks+cUFRXJ V/IE1G7qreGBrnel3f7WElZk+O6ueglI1aGl8pxftVEoBkaMGCFfF/TkOW7B M7thpbm5uUp1DGzYsEH//rAXmZd2N3o3PsfmnlFs27ZNL5KUZOBv798JV65c mT17tl5UfZF9+/aZs62enmP/a4GV4DnFnn3DraWYR+NYfw+wVJkTQkh1w2vP QVUi9ZTZcxITE6XyMteJeNJHtWv4laIhQ4YEBwe7HZ4qoIKePn26oeYNDAzE nVYSyE886D8apSj+zXP0/FEM+Ib+0vSoUaPCw8PhVMuXL8dkWFiYnomoCPwN 1jFs2DC11Oeff65+kdwAssLe038lavz48W6rci8yL9VulN+3suk5pd0zCshb ZGSk4T10bI7hJ6iQLCoqSv99VZxFM2bMQFXrNlv5IhA9xxpPymFWFE9KY/Gr WPYzJ4SQaojXnlNGioqKcIeMj4/Pzc01tG94oqCgAOnPnTuXlpbm9it5peVf //pXamrqqVOnDO/4mFEq8qvLBNLT0+UHDkpcBTYNVUZSUpLFj497nfmvXu1G O9jfMwbkJ9RTUlJQqqysLIsRRMj5/PnzWEtpP11oh6r1HP0XHyrHc4rdWYfB T6x/31PwZEElZk6IkJ2djXtRVZeCkN9RVZ7jW1i8+l3NM3cmDvQcs3jocmL/ l6rc6pB15hUKtH/Pnj1HjhxxOxdxzM3JyanoYhCQnJwcEhISFRXlKUFCQsIN N9xwyy23lPjN7cqhxAITh0DPsQM9x7eo6N/x1PHkM5XsOWDevHkDBgzo06fP K6+88sknn+D6hY3YacYxY/YiZIUMkS0yxyqwIqyu4rZFcfDgwQAXMTEx5rko EmYlJSXZzxCJN2zYgHOj/MroFFatWlWjRo0XX3zRU4Ldu3fXcHHixInKLJgn SiwwcQj0HDvQc3wLB3rO0qVLe5qYPn26/R8cN2AQJGRlzh8rraDNUSjPwWWS lZVlmOuF5xw4cACLYKPKtZiOwI42rFu3rvr0ZtJziEDPsYM+hMa3Mncmjnqv HERHR5slRMBVbL8Zx4w07CATT/lj1eW7LQaU54Dg4GDDXAvPKSwsdJshPcdr fE4bfK7ApIKg59ghISHh2LFjly5d8rnMnYnTPGf06NGePGTIkCFebKwOFkcm nvLHqstrK9winjN9+vRx48bhj7179+pzzZ4Dvdm1a9fixYvHjx8/bdq01atX nz59Wmbl5eUh/eTJk7HIhAkT8PdXX31lWF1KSsq5c+cqdIt8Fzva8Ne//rVz 587yNy60J554ol+/fidPnuzbt+/dd9/dvHnzTp06QTUNS506dWrkyJGPP/54 06ZNu3fvjuNiOGnPnDmDo4llb7311o4dO7777rvp6elqrlrRnj17evXq1axZ M2Ris8DWmW/YsAGl+vDDDw2L9O7dG3G993Pp0qWvvPJKixYtOnToMHToUEPH aFBQENJHRkaOGTMGq2jYsGFFPyAQHXoO8T8qwnOqM7jBevIQzKr++VsgnrNs 2TLUEeInegVn9py1a9dK409gYCBUB3/gfxkvDc+ZMWPGpEmTJIi/Z8+era8L 1bEsm5iYWKEb5aPY0QYkaNy4sfydm5uLyTvuuKN169YSl9E7devWjYuLU4tg b8NMEK9Tpw48R9Lg+lUJzp8/D0FCsHbt2hAJSdCmTZu0tDR9Rbfccgv8QfKH tNgssHXmqamptWrVQoaZmZlqERQeaSBFBQUFEoGkyYK33XZbvXr1pDA4ndQi eFJA8MEHH5RkSAMls7XTSXlAzyH+Bz3H/zynqKhowYIF+HvevHnq8Bk8Jyoq SgxHxAYHOjw8XMb2qG4si34rnDDiOdVkGG11wzvPAQ888EBycnKxq4MVfyPy zjvvSBoc1j/+8Y+IvP322/JCOg4QFAIR1U352muvYbJbt26SID4+/v7770cE 54NhRfATrEKdHnYKXGLmUCZMoipUi0C2ERkwYIBMhoSEyFZLEw10GrOkMGoR 8RwwZswYtz+wQioUeg7xP5zmORb9VuXSr1TR+VugPKfYNVho4sSJmNy+fbvM NXjO0qVLMYlF1OI40F988QWCKSkpErHwHLgQ5mJxVL4VulE+iteec/ToUZVg +vTpiPz7v/+7TMbExGCyffv2+j7HcURQKXRGRgYOHy5blQDHHQn69+9vWNHx 48dLW+ASM1++fDkmu3TpohJ06NABEdXxhKsAk1iXSoBtuffeexFUfabiOS+/ /LJFSUjFQc8h/ofTPMdiHHK5DAOo6Pwt0D2n+DdLGTt2rLiNwXNk7M3mzZt3 aYjn7N+/X8+B45C9wDvPadiwoe4wcB5p4ZHJRYsWYfKvf/3rAo3x48cjeN99 9+k5X758OSIiAgo0a9as119/XRchWVGTJk28KHCJmefl5TVq1Eh1XUFdMLdt 27Zq2bvuugsRnIr6JogLhYSESBrxHGReYklIRUDPIf6H0zyn2MN75eX43ndF 5+8Jg+eAlStXIoIqA8dR9xxUVQGeUS8703O8xjvPadmypZ7g7NmzuucMHTq0 hgeuvfZatUifPn0aNGhgSGDwHMOKbBa4xMzBW2+9pbqupNMKJqav2hM40ySZ eE7lfHKKmKHnEP/DgZ5T7Gp1GT169Csu8Ee5t7RUdP5uMXtOTk7OtGnTEFy/ fr3uOUVFRdKrFRMTc9aEGkdKz/GaivAc6STq0aPHLhO7d++WNN26dZNFxo4d u3HjRpx4sNzy8pwSMwcojOq66tChQ82aNfWh7zKaaNOmTeZNUMnoOVULPYf4 H870HL/E7DnF2otRgqpNZKCy4UcicLjz8vLUERfP0UdTEJtUhOf88MMPmHzs scc8ZZiQkCBtO+fPn1dBHMFy8Rw7mQutW7euW7cuzkbMeuaZZ/RZf/vb3xD8 9ttvPa2lmJ5T1dBziP8hnoMa7cpv6HPpOT6EW88BERERZs/Ztm0bJqdOnare PU9OTpYXyc+cOSMRWBAmp02bpl4KVuTk5KxZsyYsLCw/P78it8lXqQjPQRp5 GVwfuzJixIjrrruuZ8+e+DszM7N27dpIoPQ1JiZG3gQvu+fYyVz4/PPPa/z2 YviiRYv0WePGjavhertKnXXQp7vuuguboIbE03OqFoPn/PLLL6pqmDlzJj2H +CLiOTrnzp1Tc+k5PoQnz8FBnD17tsFzCgsLFy5cKAOV58+fP2fOHPxh+JAy ntxlKfiP4QPL0dHRMsvTz4Y6HNGG66+/vr0J7G1JU1rPKXb9Juy1116L4EMP PfTqq6/+6U9/quH6mo0aOv7ss88i0qhRo27dunXq1AkFkE/x2PQc6wKXmLkQ Hx8vQ26QIDs7W58FYZYmnVtuuaV79+7IBLlhUr5VKNBzqhaD54SEhOi1Az2H +CJXr17FfSlO48cff1Rz6Tk+hHzTZsWKFeZZycnJ8pFk+TaLgCovLCxM3rEC kydP3rp1q+E3IKKioiQBFtfjkGGknzJlSjX5ue3qRmhoqKcBt2pcbs2aNdV7 T3l5eZh1++2365lASqVhRA+uXr36ySefrFOnjhjOY489pn8cICMjo2vXruJC Mk5m/fr1+ANSZLEimwUuMXPFX/7yF7dxkJWV1a9fPxRAMoEpjR49Wm8wlOHW Sq5IJWPwnMzMTL12wDlQFdUUIRUIPccJwFXMv/upKCoqghHpX00R/q+LCi4a cU9OTs7x48fNB0XIz8+PjY3Vv0tcjpSYOc6K++67D66ybds2i3zweJWamloB BSRlwmJ8DiF+CT2HEGKfwsLCSZMmSc9aVZeFeAM9hzgNeg4hxCbBwcE33nij dKht3ry5qotDvIGeQ5wGPYcQYpO1a9c+/PDDvXr1su6xItUZeg5xGvQcQghx DvQc4jToOYQQ4hzoOcRp0HMIIcQ50HOI06DnEEKIc6DnEKdBzyGEEOdAzyFO g55DCCHOgZ5DnAY9hxBCnAM9hzgNeg4hhDgHeg5xGvSc6s8ZQgjxQGnvJ/Qc 4jToOdWfqr6PEkKqL6W9n9BziNOg5xBCiHOg5xCnQc8hhBDnQM8hToOeQwgh zoGeQ5wGPYcQQpwDPYc4DXoOIYQ4B3oOcRr0HEIIcQ70HOI06DmEEOIc6DnE adBzfILTp0+HuPjuu++quiwlo0qrCA0NPXXqVFWXixBCzyGOg57jE8yaNauG i44dO1ZtScaMGfOBi6SkJE9pVGkNNG3adOLEiZVfHkKIgp5DnAY9xyeoPp7T okULKcnevXs9pfHkOcKoUaMquTyEEAU9hzgNeo5PYMdzEhMTjx07VlBQYJFP YWHhmTNn8L/XJSmV57Rp02br1q2bNm369NNPW7durVTH3PaCIh06dOjixYsl FiA7O/vEiRP5+fn2y0MIUdBziNOg5/gEZs85cOBALReIwCXuvfdeSVC/fv1p 06apBbt16ybJ1q5dO3jw4Lp16yJNw4YNX3zxxZiYGJUsIiJCkj366KOegqNG jWrevLlylcaNG2Ny/vz5dkoLEhIS1LLBwcH6Wp577rkbbrgB8Zo1a95///1D hw7VTUxtwsqVK7t27SqbsGTJEvvlIYQo6DnEadBzfAKzOezbt08Nern55psN fUObNm2SZJ07d5YI/MGQ5p577snIyJBk69atk+BDDz2kVmoIvv/+++ZOKN2p LEorPPjggxIfNGiQRDZv3izeYqBPnz5FRUWGTbjzzjtVAniO/fIQQhT0HOI0 6Dk+gYXnSFPG0KFDX3rppWuvvVYiPXv2lGRKEkC7du1GjBjRo0cPFenfv78k s+M5ERERWPz666+X4JtvvonJqKgoO6UFcKr69etLPCgoCJHU1NQGDRpI5JFH HkFwwIABqmyBgYHmTcCWYtP69u0bGRlpvzyEEAU9hzgNeo5PYO05O3fulOC7 774rkQ4dOkhESUKLFi3S09Ml+M4770jw/vvvl4gdzxFKNT5HSltYWLhly5bu 3burAp84cQLx1atXy2SzZs3UyJwPP/xQgs8995xhE1q1apWcnGxYF8fnEFIq 6DnEadBzfAILz7nxxhtVF8/atWsl2KZNG4koSYA/qNwOHDggwdq1a+MEKK4w z6nhGnJT4/e88MILkmzYsGESeeONN9SyUVFRErzpppsMmzB8+HDzuug5hJQK eg5xGvQcn8DCc5o3b66S7dixw5PnzJgxQyXLyMhQ1hEXF1dckZ5joHfv3mqM cZcuXSQ4cuRItWxKSopKnJqaqm/CnDlzzOui5xBSKug5xGnQc3yCsnuOGvpb bNme07p1a5Vs6dKlZfScJk2aDHQxZMiQ4ODgs2fP6skQlGR9+/ZVwT179kiw UaNGhk2g5xBSdug5xGnQc3yCsntOy5Yt1RiY9957T4Lt2rWTSGRkpGpFkRYe 8Oqrr1p4Tnh4uP3SuiUkJESSIc/MzEwJDh48WILPPPOMYROsPceiPIQQBT2H OA16jk9Qds+p4Rp1PGLEiF69eqlIv379JFlOTk6tWrWUEQ0dOrRnz54qme45 7du3V7lNnTr1yJEjdkrrluTk5Hr16knKJ554Iigo6KOPPlLjecaPH2/YBLee Y6c8hBAFPYc4DXqOT1B2z2nWrFmN39OqVSv9+8MvvfSSIQGEx+w56l0toVTf zzGzfv169S68DixLnZbWnmOnPIQQBT2HOA16jk8QFBQk9bj6XvH+/fuVjahk O3fulGDbtm0loiQB7gElqF27Nv5u0KAB4ocOHdJXkZub++KLL9apU0fSP/bY Y2FhYfL3ww8/rJJlZma+//77t956q8rWTmktWLt27VNPPaU+rQND++CDD2TU kKCGK8+bN8+8uJ3yEEIU9BziNOg5/o2hMSQ/Pz82Nlb9OJSZvLy86OjoxMTE EnMuKCjQbaSMILeDBw+mpaWVJYdyLA8h/go9hzgNeo5/Y93pQwhxGvQc4jTo Of4NPYcQokPPIU6DnuPfQG8GuOAPPxFCiuk5xHnQcwghxDnQc4jToOcQQohz oOcQp0HPIYQQ50DPIU6DnkMIIc6BnkOcBj2HEEKcAz2HOA16DiGEOAd6DnEa 9BxCCHEO9BziNOg5hBDiHOg5xGnQcwjxXcryi2AGcB8oKCiwmTgjIyMrK6u8 Vu00sJ8LCwurau30HOI06Dk+RHp6+p49e06ePFnVBSHVgvfff79GjRqjR48u e1bwlgULFuzevdtO4pSUlICAgHHjxl26dKnsq65ysrOz93jm3LlzeuLLly9v 37595cqVwcHBW7du9XQxxsfHI9myZctWrVr1/fffnz9/Xp974MCBuXPn4nKu wK3yDD2HOA16jg9x8OBB1C9Lly6tkrUnJSVt2LDh8OHDVbJ2hxATE/P5559H RETYSfzcc8/Bc/7+97+XcaXQlSlTpqBGtnkrOH36dIALCE8ZV10dSExMDPDM Dz/8oFLGxcVNnDjRkGD9+vV640xRUdGWLVsMacaPH3/kyBE9TXh4+KRJk8qx Oc4+9BziNOg5PkTVeg4eQrH2tWvXVsnaHcKKFSugLm+//badxMnJyfPnzy9j XYk7wBwX9u8DqKZjY2NR6ZdlvdWHy5cvR5rYuXPnuHHjcMLjtJdksMEJEyYg Mnfu3KioqISEhK1bt44dOxaRffv2qdyOHTsmboO5OEBIBoGUSGZmpkqGfbhw 4cJZs2bl5eVV8vbSc4jToOf4ENaegztniYeyLMeanlMJlMpzyoU1a9bgsMbH x3tKUO4jScoxQ09ZHT9+PCIiAleE27k7duzYu3evdc4QOeyWqVOn4g4pkbCw MERmzJiRn5+vkn333XcIwhJVBBcIInqLHIoBNULQsFJYEILLly/3VM4Kgp5D nAY9x4cweA6eBL/88svg4ODExET8P3HiRDyBfv311ydOnFCL4JaONCdPnoyO jsYfePycOXMmbsL62AD8jVnLli3T14WbMIKrV69WK5o8eTLWjkda/P3VV1+Z i5eSkmIYzODHBAUFPf74499//31gYOCjjz7apEmT5557Tp79FyxY8Oc//7lx 48aIY65hwfDw8Pfee69Nmza33357165dV65cKXFciU888cQ999wDz7ntttuQ eadOnfR1RUZGjhkzpmPHjg0bNsTRRPybb75BXI4RIk8//fR//ud/Xrx4Ua1r 3rx5CA4bNszTVshIG72aViQlJeFMw0HHSbVo0aJdu3ZdvnxZZhUUFOAcwJkm k6dPn/7SA3pjBc5SbDs8ASchSg5DUAqhClPi+bN9+3Zku2/fPpzDOJORFbYx JibGkGzx4sXi5GaF2LZtG2YhE4u1YCnsEyTTByx98cUXiJw6dUpPiU1ITU29 cOGCiuBKNCuNyOTOnTsNK8KORRw70Hqryxd6DnEa9BwfwuA52dnZmITeiIGg SlXjAVCnSJoNGzZgcvbs2RJXaXCfz83NlTS4USOCylRfV0JCAoKosotdnoPq adKkSTLSAH8jQ0PZoFKGVfs3Q4YMgZA89dRT+P+mm26q4aJRo0aDBg3CHw0a NKhXr54EURerpeAMEoQXIY38jf2JWfn5+TAfyeq6665r2bLln/70J31dDz74 oKRHznv27EF85MiRmER1L8lkuM6bb74pk2fPnr3++utr166t96oY2LRpk/Sw GOKoeXGgMQuVO04MObKQW2k/QVHlTJDEcXFxM36P9O8A/RyTkS3YA3K6ghUr VigPsXn+zJ8/X53k6mRGnmfOnNGT4YyF/2BWWFiYrjrS/IKTX+9CMgNxMjTm yCZjRbhbooSwvm+//fbYsWO4Bg3LwnCQcuHChWrZjIyMadOmBbgbzoR8pJAW hSl36DnEadBzfAi3niOKgntpseuRWWql0NBQSSOeA0JCQuR5PDk5GTdwRFat WiVpSvQcwbrf6vDhw7IivTXJjxH3qFWrFnZssev9GuUhw4cPLygoyMrKGjZs GCbbtWsni5w6deqaa66pW7fu9u3bi11VJxQFCe68806Vrdt+K1kXGDNmjN50 YPCcpKQk0aRt27ZhsnPnzrKIxVaIAJtH2siQEpw8Ignp6ekiFdKaYfAcA5gr J2FkZKREcJOZPn26VOhS+6OokqEa5Wvn/IFliX1BVGRUEk5dydk8chuKJb1F SnVsSg4Sw+gMjTnS8DVr1izpvdLZv3+/vnhOTg6uGnm4wA5ct26dXG5uu5ux HzALFmRRnnKHnkOcBj3Hh/DkObhbqjSoXAJcQyVlUjwHVaE+qAC1lTwOy2S5 eA7qICRACSt5sEFVIe7x+uuvqwge4RFp0aKFipw7d66mC6ncsYtQXer9Sog0 bdoUS+EQSMTCc15++WVDGQyeA5YtW4bI3XffvXjxYvzRsWNH68Ew0sZi7i3C OYY4xEBFoA3QA/nAjrXniCPpbTVnz54VwdDPjX379gW4RqeoXVHi+SMDWoDe 67p582ZPFqFUJzw83KbkgCNHjoh76N1qSsPAvHnzvv/++6ioKOl1AkePHtVz OHnypN64Ks7j9i1yPJ5Igsq8CdNziNOg5/gQbj0H1Y1el8kTohp+IJ6Dh0o9 HxxxeS6W75+Ui+c4DXEPvf8Oj/+IqEE1ArQHQdV9I6AmDQ0NRXU5ZcqUZs2a IYEaoWHhObNmzTKUwew5oFevXtL4U79+/djYWItNyMvLk0rW/MW//fv363U6 TipdPyw8BztB2j30kTlQAgnu0sA5ieD06dMtSmggOjoaiyxZskQP7ty506Lr B3teRtrYlBxcGtLGpb9OXvzbyS9Xn/qaIhKL1OmXifiblHP79u1btmz5+uuv ZXfB0wyrw+1XEquxT5UAPYc4DXqOD+HWcyZPnqynSUtLM3sOqipDVqh0VDcE PccLxD2gASri1nNatmype87cuXPbt29fw4Qdz9HXJbj1nAsXLtSuXRvxQYMG WW+C8hxzJYuqHHW0jMgSUFkra/LkOfHx8WPHjp04caLhs3iq89Qt9j/CLK8y GUbzLl++HEEZmO2WjRs3yorgJCU2Nh46dMjcmCObJpkYXkyT/izsCtkK7FJ5 G13XJDyGrF69OsDdeG8s5Uk1Kw56DnEa9BwfwmvPMYxewN1eOixksId4juEN FBkXSs/xhBees3///po1a9avX//1118PDg7esWPH4cOH27ZtW76egzpa3Om2 224r8dM6Uil7SoZa+/jx45s2bZIRJkCGgbn1nMzMTElm6MQp/m207ZIlS86a KNWodbejiWSIr7mpRICtBbj6bWXIzbp16yxUB3dCXAVIFhUVZZh16dIl2QN6 t2Ox5orSNIqyySVpWItclUDvPgbQm9LKXtmh5xCnQc/xIbz2HNzk9aMs4xyg OnI3xq3b/Fwv3QFmz1Gjlx2OF54DvcHfAwcOVHNR68lbVwbP6du3r/W6BLPn xMTE1KtX79Zbb+3Xrx9mde/e3XorZBCv4WWlYldVqJ8wqIXlRSe4WbE7zyks LJQEkCLzWuT7ybNmzdJrf/wNSTA0m1igjELvBJTxLYauW4WSHDiYGqtjoTrS WwdxMlsH9oYI1aFDh/S4PA6ogcTiObiyDOWRZp+xY8cavgp4/vx5SW9zJ5QL 9BziNOg5PoTXngM2btwoBxpLSaeVyge3femhgNvI/fncuXPy8q/uOWp8pttn z5ycnDVr1oSFhRmeWP0VLzxHFunRo4eaO3jwYEO/FXYgJtu2bWu9LsHgOTh2 Dz30ECKLFy/GUb7jjjvwd3BwsMVWyHhjQ92N4ysvhqtXn5TnSPeQ2XPWr18f 4HqZ2u3NBJsv7YebN2+WCE45+aQMiiqREs8fUQjDICX5+DDKZk6vJEfaoIp/ PyzZrDrYezNmzHDbmCPIKKMpU6aoXrlLly7JR3XkE0aSiWwpfE+tAm6zZMkS BBctWmTIU75G6Lb8FQc9hzgNeo4P4bXnSPdEYGDgV199JV+qnzp1qrr/F/82 8kFykxZ+WUT3HHn2BJAic+0pY0SB/js+fowXnhMZGVmzZk35Ek7v3r3btWuH v2+++Wbdc/CHpGnduvVbb73laV2CwXNGjx6NyWeeeUYm4R6Sv6c+neLfjpq5 L1IGCQMcaFTiIgA4N6TFz+A5auxKUFDQ4t+j2l6gTHLiwZMRhy1I+8bZs2f1 klicP/IiIVxID8rLVnB4Q+IdO3YYJEfQXzZ3uyumT5/uqQsJcTWiGEYHaREb xFr03zOVcga43rEKDQ1dtWqVNAQhsflHP2XskHoBv3Kg5xCnQc/xIeTlVuUY eASWB0w9TXp6utxjZVI8BzdS3NilosFdWj6hrC+FezhuyOrzbvKNWfMTKB5p 5QEWFmQomzQBoTCGAQz+ytChQ2ER+pM4dg4inTt31pPdfvvtCKreipCQkD/8 4Q/ShtOkSRNUl/Ai/K2PboUtSFMMFMXTuoRRo0YhLk6Lc6N27dr16tXTP9jb s2dPJOjWrZunrcjKypo4cSIqYkP7RmFh4bZt29QH/XBi4JxRziBvCaneluPH jwd4QB9ee/ToUWyv+DNORWyOLmAlnj9wBiwoH0hUyMeHzb8tC08zS44gqmN+ Dx3pzfkbgOChGHIdyVWAC8S8FmypdAgqIEhuf/MU1ymux0q+ZOg5xGnQc/wb 8RxpisexTktLsxgRgTMBCaBPFhmiQkRN4TYTw6AO4gn5eUeLfYWdfOnSJesD YQC5mYe7uA0akEEpho/dKSAqMOdy/CYSVAHnmNvhNOV4/sjgH09zsUPK8htb 2IQzZ86cPXvWesdevnwZ+oqUnnri5FFC/xph5UDPIU6DnuPf6J5DiFtCQkKm TJlS4rdlSDmSnZ39xRdfLFmypPK/q0nPIU6DnuPf0HNIieAOsHHjRlS7hu/e kApCfjk3LCysMl8nV9BziNOg5/g3J0+e3LZtm8VIVEKEgwcPOmQMeZVz9OhR iw8bVjT0HOI06DmEEOIc6DnEadBzCCHEOdBziNOg5xBCiHOg5xCnQc8hhBDn QM8hToOeQwghzoGeQ5wGPYcQQpwDPYc4DXoOIYQ4B3oOcRr0HEIIcQ70HOI0 6DmEEOIc6DnEadBzCCHEOdBziNOg51R/zhBCiAdKez+h5xCnQc+p/lT1fZQQ Un0p7f2EnkOcBj2HEEKcAz2HOA16DiGEOAd6DnEa9BxCCHEO9BziNOg5hBDi HOg5xGnQcwghxDnQc4jToOcQQohzoOcQp0HPIYQQ50DPIU6DnuNb4HitXLny o48+6tWr1yuvvPLZZ59FRkZWdaHKh9OnT4f8ntDQ0FOnTlV1uQjxK+g5xGnQ c3yIQ4cONW3atIaJxx57LDMz04sM9+zZ84GLefPmlXtpdcaMGSMrSkpK8pRm 1qxZ5k0D2OSJEydWfnkI8UvoOcRp0HN8hdOnT7uVHOGpp57Ky8srbZ5z586V xZ9//vmKKLOiRYsWsqK9e/d6SuPJc4RRo0ZVcnkI8UvoOcRp0HN8grS0tLvu ukuq5gYNGowcOfKHH37YuHHjm2++qUzgxRdfLG22Nj2nsLDwzJkz+N/r8pfK c9q0abN169ZNmzZ9+umnrVu3VhtobntBkQ4dOnTx4sUSC5CdnX3ixIn8/Hz7 5SHEL6HnEKdBz/EJAgMDVXW/bNkyfVa/fv3UrP379yMSERFRy8Wjjz6qkhmC WVlZzZs3b9iwoSxYp04dTN5///2SuFu3bpJ47dq1gwcPrlu3LtIgMVQqJiZG 5YmsJBky9xQcNWoUclYlbNy4MSbnz59v3kblOR07dlTBhIQEtWxwcLC+Oc89 99wNN9yAeM2aNVHyoUOH6iamNmHlypVdu3aVTViyZIn98hDil9BziNOg5/gE jzzyiNTLnTt3NszKy8u76aabZO7w4cMRWbdunUw+9NBDKpkheOnSJXPfUKNG jSQx1iIR+IMhzT333JORkSHJkJUEkblakSH4/vvvm1c0bdo08za69Rzw4IMP SnzQoEES2bx5s3iLgT59+hQVFRk24c4771QJ4Dn2y0OIX0LPIU6DnuMT3Hrr rVIjBwQEmOc++eSTMrdXr17F9jwnPz9/xIgRnTp1kmDr1q0xOX78eEmsJAG0 a9cOs3r06KEi/fv3l2R2PCciIgKLX3/99RJ88803MRkVFWXeCreeA6eqX7++ xIOCghBJTU1t0KCBRKB/CA4YMECVLTAw0LwJjRs37tmzZ9++fSMjI+2XhxC/ hJ5DnAY9p/pTWFioquw1a9aYE7z77rsy96mnniq25zmCp/E5ShJatGiRnp4u wXfeeUeCqnvLjucIpRqfI56Drd6yZUv37t3Vtp84cQLx1atXy2SzZs3UyJwP P/xQgs8995xhE1q1apWcnGxYF8fnEMdCzyFOg55T/SkqKrr22mulXl60aJE5 wauvvipz/+M//qO4XD0H/qCCBw4ckGDt2rVx2hRXmOfUcA25qfF7XnjhBUk2 bNgwibzxxhtq2aioKAnedNNNhk2QvjwD9BziWOg5xGnQc3wCVS8PGTLEPFeN YHn99deLy9VzZsyYoYIZGRnKOuLi4oor0nMM9O7dW40x7tKliwRHjhyplk1J SVGJU1NT9U2YM2eOxf6k5xCnQc8hToOe4xN07dpV6uX77rtPvRwtnDhxok6d OjL3yy+/LNaUpnXr1irZ0qVLvfAcNfS32LI9Z/HixSrZPffcU0bPadKkyUAX kLrg4OCzZ8/qyRCUZH379lXBPXv2SNA8lJqeQ4gOPYc4DXqOT7Bs2TLVXvH2 22+reE5OTvv27SVeq1YtacqIjIxUiaXhpVjr23LrOR06dNBXpyShZcuWagzM e++9J8F27dpJ5Omnn5ZInz59JHLq1CnV5eTWc8LDwz1to6f3rQyEhIRIMuSp vgI9ePBgCT7zzDOGTbD2HIvyEOKX0HOI06Dn+ARFRUX/9V//pexFPhfTr18/ /QvJqkKH/MB5lKggZc+ePVUy3XNCQ0NVfMCAAWrwj/6yEtY1YsSIXr16qQjW K8k+/vhjpVhY5JNPPrn99ttVMt1zlIwht6lTpx45csS8jTY9Jzk5uV69epLy iSeeCAoK+uijj5RcmV8Zc+s5dspDiF9CzyFOg57jK+BI/e1vf6vhgXHjxumJ X3rpJUMCCI/Zc86ePaucoYa7Tp9mzZoZ8mnVqpVq4YmOjjZ8xwa+oRbRPUe9 qyWU6vs5ZtavX68GZutA59TJbO05dspDiF9CzyFOg57jQ2RnZ/fr1++6667T 62h4xZQpUwwpc3NzX3zxRTVu57HHHgsLC5O/H374YT0lbOTxxx+vXbt2DdfA GAkqSYB7QAlkboMGDRA/dOiQvnh4eLjqA6pfvz7SP//88zIJG1HJMjMz33// ffUVICQzb11QUJDM1T/j7Im1a9c+9dRT6tM6bdq0+eCDD2TUkKCGK7v9iVI7 5SHEL6HnEKdBz/E5UEfv3r07NDQUdf3+/fstfnYqLy8vOjo6MTHRTrZQI/PH hKUxJD8/PzY21jD+WSchIQElQQ4lrqWgoEC3kTKC3A4ePJiWllaWHMqxPIRU f+g5xGnQc4gZ604fQojvQs8hToOeQ8zQcwjxV+g5xGnQc4gZ6M0AF/zhJ0L8 DHoOcRr0HEIIcQ70HOI06DmEEOIc6DnEadBzCCHEOdBziNOg5xBCiHOg5xCn Qc8hhBDnQM8hToOeQwghzoGeQ5wGPYcQQpwDPYc4DXoOIYQ4B3oOcRr0HEII cQ70HOI06DmEEOIc6DnEadBzCHEIuNILCgpsJs7IyMjKyqrQ8pBSgWNXWFhY 9nzoOcRp0HN8kdzc3KioqLVr1y5dunTlypXffvttUlJSVReKVGvgLQsWLNi9 e7edxCkpKQEBAePGjbt06VJFF6ySSUxM3LNnz+HDhw1x3AZjY2O3bNmyZMmS DRs2IIF9J7QDMg8PD1+8eDEuW1y8ly9ftkgcFxe3x8WFCxdU8MCBA3Pnzk1P Ty9jSeg5xGnQc3yOM2fOTJw4McBEWFiYF7lBkOSuXu7lFLZv3478IWYVlD+x A3RlypQpq1atsnmxnz59Wk4qCE9Fl63SwLbjbBw7diy2a+bMmfosKA2eFwwX 1MKFC8vlvC0sLFy+fLkh82nTpsG43KaHyUyYMEGS7d+/X8WLiopgSpMmTUpL SytLeeg5xGnQc3wLPN9NnjwZN8AvvvgC/nDo0KFdu3YtWrRI7orbtm0rbYZ4 SMSCeMasiNIClBP5X7x4sYLyJyWCa3yOC/tXOqrU2NjYuLi4iixXpQLTW7Bg gdIMg+esX78ewcDAwH379kE/oqOj5VFizZo1ZV81rlNkBXXBWk6ePBkTE/PV V18hMn36dPMRwZ6HX6ly6p6j5s6aNSsvL8/r8tBziNOg5/gQuLnh/oy735df fml40ty5c6fcGGE+pcrTjueU5Qyh51Q5qKxxCOLj4z0lKJdRHxWUoaesjh8/ HhERgXrf7dwdO3bs3btXj0yZMkUunKioKLPnTJs2zbCLjhw5ggiWMqyitOvF mS8tSHv27FFBXLyTJk1CEDJpyAGLy1PMsmXLzJ4DkpOTEV++fLmnMpQIPYc4 DXqOD4GHQdzi8KRpbrjGTU8a3pcuXSqR9PR03NVxt9ST4SaJ4OrVq4td1oS/ pXUID5v4G4+Zkgy3a0xidXiwxR+4UaNewO1dHxtQYv5QL/wt9oVHV/xtuGmn pKScO3eufHZNVXDmzBnUj506dbr11ls7duz47rvv6vsnKCjo8ccfj4yMHDNm DOY2bNgQO9MiDsLDw9977702bdrcfvvtXbt2xQFVuW3YsAFLffjhh4Yy9O7d G3GLbkcZaTNnzhzzrKSkJJwtOAHGjRu3aNGiXbt2qUEjBQUFOF5ff/21TJ4+ ffpLD+gNC4mJidiEGTNm4IT55ptvvvvuO9xeDIUp8Yhv374d2e7btw/nG846 ZDVv3ryYmBhDssWLF4ufm6v7bdu2idLowfHjx8P38vPzz549a/Cc7OxseMXs 2bP1rLKysuTUNQzGLu165TkCe8OQHv6TmppqaJbBusR/Tp06FRoa6tZzgLTf 4qCYZ9mBnkOcBj3Hh4B+4P6mV386CQkJmItqSyZxF8UkalVzmgULFhS7PAdV ktxXUQvgb9zqJZm0tGNSbvWBgYHyh96OVGL+qLCQpyw4depU/K0/6oqzAU+j FKo558+fb968eY0aNWrXrt2iRYsaLqAoSkGHDBmCyIMPPiiz6tWrJ0/0nuKQ DYk0adKkQYMG8jd2muSGvV2rVq26detmZmaqMsTFxSENLMtixOymTZuwk7du 3WqIo5bEQZemAxxEORYQXWk/gQ/IWaFWNOP3qAEk+vkgfT3YEJFnsGLFClW/ 2zzi8+fPV7026sRDnrBKPRnOXvhPgGtYmq4QkCs5UfUdJZsgf5g9xy3x8fFI hqvD4CelXe/mzZsRx5WLXQrnweT333+PxwG3d115VJHOMgvPgZEGeDser5ie Q5wHPceHkNGMuGe6nYsaRyoFeUemRA8R3PZbieeAkJAQecbHnRmugsiqVask jc38PfVbHT58WFZx4sSJ0u6H6sBrr70Gx+jWrZtU9KgW77//fkTUtovPgDFj xuhvzbiN4/n9mmuugcZADotdmoGKGGnuvPNOtWCnTp0QWbJkiYpANhAZMGCA RTlFVs0jbXAcEceBlso6PT1dpAIlKTZ5jgHMFTWKjIyUCG4j06dPl8pX2nCS kpIkwx9++EHS2DnisCyxLwiDGCNOM8k5IiLCkBh7fu7cubpyeJINHTueg2JI zuvXrzfPLdV65ZqFbUrXmGLOnDk5OTl6ytjY2ABXT1l2dnaxpedg3wa4RjJb bIIF9BziNOg5PoQ0j5jb8BVSI0inf9k9B3UBajQVRA0oj9gyWUbPQVWCVR88 eNDrYQZVS0ZGRkpKit4vg2oO1tG/f3+ZFJ95+eWXDQu6jWNvIDd9LyHStGlT pMR+lghqTEx26dJFpenQoQMiqtvLLdLGYu4tWrp0KeKooFUE1TeqaWkasvYc cSS9rUbkAXtAP5r79u0LcI0kUVtU4hGXwSdA7wGUJhHVIaujlCM8PNyO5BTb 8xxcDtIIaVARL9arum6xM6Fq2APffvutNKIiB9UQhxNJrm7sH4lYeA7OPcnT u1s3PYc4DXqOD4FnQNzccKt0OxfVh1Rq0sJfds9Zt26dHsR5Is/apWov8u9x yJcvX0blhSp41qxZr7/+OqzjlVdekVniM4gbFvEUF44cOYIKbt68eXiub9as GVKqYRh5eXmNGjVSXVeIY27btm0tiodFpEI0f/EPFajMwrq+//77pKQkXT8s PGf37t2YZXjlRwb3IrhLA+dPgGtolkUJDcDZsIjeZlX82xh7T900UA65LuxI TrENz5FhNhMmTLDuX7O5XrGXcePGSUOZgJyl408d3K1bt2Jy0aJFKo2F5+Cm Leu1/giPJ+g5xGnQc3wIPDzi5rZx40a3c/EILHc/aYQpu+eg+jOsAhWZ6tpw uOeguuzTp48aSKMweA4swrCgpzie7tu3b1/DhD7c9K233lJdV9Jp5anJRVCe Y64QCwoKtm/fLg0Lwtdff61e//HkOfHx8WPHjoVOnz9/Xo+rjk632P/gnjSk QGz0oHT9WDRb4YqQFa1atarE5kFrz/nhhx8wF9topzvVznrlJXH8b4jLRkl7 Wlpa2jgX+l618BzsT0/6agd6DnEa9BwfQp5258+f73buoUOHAlyDh2VSPMTw 9oeMBbXpOYYREaq9SEaV2MzfXz2nW7du0IwHHngAdSLqOxwaSKDXnoPqrGbN mvXr13/99deDg4N37Nhx+PDhtm3bGjxn165dquuqQ4cOWKTE72Cj9sT+9/Rl OYjQ8ePHN23aJIOvQEZGRrEHz8nMzJRkR48eNeQjI2MhYGdNlGqcudvRRDKy JTk52e0isDXxFvkozbp166xVx8Jz5AoKsPdxBpvrDQsLC3D3HR5xJGmkkhYk UU2FXGvY4cgfx0hfVr0L5t0Xm+k5xGnQc3wINXrB/PV+VE9SB6k7KtTC/Cwv XQBmz1GjiwXxHNxg9XND1o7br9zPbeYvnmN4/Pd1EhIS4BvXXnutvl3YTK89 R/q8Bg4cqCIwDWksMrw+3Lp167p16x48eBCznnnmmRKLKkO2DC8rFbsqO/3g osaUF51klLvZcwoLCyUBpMi8Fvl+MkxPr+vxNzzK8Gq5Bar1Sf82lIxFQUnc fkhHyQYcTI2ZsVYdT54TGxsrH7px24Ti9XrleziGkUtAXtrat29f8W+dVhYY ioSzTq7EEsvpFnoOcRr0HN9CDAQ3ZNR0apBwamqq3Gnx5CvP48WuWkZ6JeAe UkecO3dOXvjVPUS+h4YF9WdD1Q2Bp045PbKzs6XTSg0HtZm/LBUVFWXYkJyc HCgZnmf1oc6+Amq32rVrwzSw9yQSExMjr5l75zkS7NGjh4oMHjzY3G8FPv/8 c/Vauj6cwxMy3tjQQIFjLeNDVO+M8hzpHjJ7jnwxeOHChW5vF6jrpf1h8+bN EsHpId8nXLx4sURKPOJxcXEiS3rw2LFjAR7aMJVsqHNeHx7sSXXceg7EVcae 6V/z80Sp1osdK+1RuKYMOQT8NtwaabJNSMcWnmjwt8Hx5M0sT+26JULPIU6D nuNb4GAFBwfLTRJ3ZhiFPLDL853h14hktEOA66M60rouvRi6h8izYYDrUyHI WYLiOZI4MDAQy8qj7tSpU9W93Wb+8m6OtA7p/R3SBweUKvgWzz77LEyjUaNG 3bp169Sp0/XXX9+4cWOvPScyMrJmzZoiML17927Xrh3+vvnmm82eEx8fL/6D NcoLyNbIfjZ/71oGCQMc9NWrV8twWRxHaZ0zeI58TCbANRxr8e9RbS9QJjlJ UK0jLt8fRgReoZfE4ohjJwSYunjkZSvzmDT5lpQuG4L+0rfbtZg9Jy8vT41T mmLC0GfkxXpVdxh2MuxFvYHl6QMRglw4bhuXpM9LvdRfWug5xGnQc3wO1EGo pNSH2gQ83OGZ1JASz4m4W6qU8l3ZgN+/1lHsellGepdgKRIRz8GNFDdtqbxQ 5aFCNIy1sJP/xYsXcW+XTPQ7szT+oB7x0aE7qOa6du167bXXinV06dJl/fr1 +OPVV1+VBEOHDsWk+aHbUzwkJOQPf/iD+lTgwoUL5YM55t9r+Mtf/qKvyJqs rCwIMNzD0M5QWFi4bds29UE/HEQcX1V3yxs9qmcEdX2AB/ShsPBYFFtcF0cc 26gPqinxiMvIW0OLili9+YPPuATMsiGIcrh9D734t4/P6K1G0EVPW2e2Mu/W e+rUKf37OXg2KfGH4yGfAR5erpw9ezauR68vHHoOcRr0HB8FD6GoRE6ePIl6 UF709gSOb1pamqcvgQioBHGXVkMpxHOkswlBLG4xysJO/jAiJDBUtYYhIr4I nDM2NrbEd5ntg2MKX7XYLZh13333wXPs/2arvELuadgJRCU9Pb0cv2KEfYLz we1wmnI84jL4x9NcnK7l/qNdZV8vzCQuLq6MYi+PEiVqkgX0HOI06DnEjO45 pPqAOnTSpEmQnIceeqhUC4aEhEyZMqUcfYxUCfJTXEuWLCmLl9JziNOg5xAz 9JxqSHBw8I033ljD9YtaarivTXCNb9y4EVWkn7345ijkl3PDwsK8e51cQc8h ToOeQ8ycPHly27Ztnr5YQqqEtWvXPvzww7169bLfY2Xg4MGDPjrqmxS7hj9Z /8aHTeg5xGnQcwghxDnQc4jToOcQQohzoOcQp0HPIYQQ50DPIU6DnkMIIc6B nkOcBj2HEEKcAz2HOA16DiGEOAd6DnEa9BxCCHEO9BziNOg5hBDiHOg5xGnQ cwghxDnQc4jToOcQQohzoOcQp0HPIYQQ50DPIU6DnkMIIc6BnkOcBj2HEEKc Az2HOA16DiGEOAd6DnEa9BxCCHEO9BziNAoLC+k5hBDiEHDPLygoqOqah5DK o6ioCOc8/q/qi48QQkjFom74VV3zEFJ5/Pzzz+y6IoQQJyCdVrjtV3XNQ0il IoZP1SGEED8GN3k25hBncvXqVTn/8T87sAghxM/AjV1u8oWFhbjhV3WdQ0gV gDMf18I///lPGZZMCCHEP5CBx7i9w3YoOcThXLlyBbYjVwQhhBA/AJ6DGztu 71VdwxBCCCGEEEIIIYQQQgghhBBCCCGEEELc8/8AdDAztA== "], {{0, 114.}, {380., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{380., 114.}, PlotRange->{{0, 380.}, {0, 114.}}]], "Output", CellChangeTimes->{{3.763326325835087*^9, 3.7633263305523653`*^9}, 3.7633626287166967`*^9, 3.763362750941792*^9, {3.763417624946739*^9, 3.7634176283312263`*^9}, 3.7637440008712387`*^9, 3.7637446351549597`*^9, 3.763744912102482*^9, 3.76381007005124*^9, 3.763815817038168*^9, 3.763914843445978*^9, 3.763989563194501*^9, 3.764081112137546*^9, 3.7645895418761597`*^9, 3.7981862171902227`*^9}, CellLabel->"Out[28]="] }, Open ]], Cell["\<\ BERT uses self-attention, where the embedding of a given subword depends on \ the full input text. The following figure compares self-attention (lower \ left) to other types of connectivity patterns that are popular in deep \ learning:\ \>", "Text", CellChangeTimes->{{3.764589561438443*^9, 3.764589609454968*^9}, { 3.76458964479923*^9, 3.7645896942634163`*^9}, {3.7645897305833817`*^9, 3.764589813151984*^9}, {3.7645898737050543`*^9, 3.764589920897312*^9}, { 3.764616315317905*^9, 3.764616323211269*^9}}], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsvWtQlVme5ltxOs6piZ4PcybOnPPlRPR0x0x0xKmZiYmp6O6ajo7o6uqu 6q571fTpPh0zX870dHdmVlammqlpppmmKaiggiiKIAgKgohcFAVEuclFBBFB bgIiUiACgoDcFO/nd/iHb+zcKL4osrd7P78PxGbv97LW/11rvc963vWu9Xv/ Y+Vfv/M/feMb31j9z/jz13//6fc+/vjv1/3f/yv//O2K1b96d8U//eOPVqz5 p3f/6eP//D9+iy9j2PbP/pdvfOP///xUCCGEEEL4iImJidjY2PXr1//BH/zB t771re985zvf/va3v/nNb35DCCGEECKYQP8gh/74j//43/27f/eHf/iHGzZs 2LdvH0rJ12JNCCGEECKoefz48Z07d3p7e/fs2fPll1/+5V/+5Z/92Z/9/Oc/ //GPf/z7v//7/1oIIYQQIphA//z0pz/95S9/+b3vfe+HP/zhhg0bYmJibty4 MTExgWrytXATQgghhAhSbORVeHh4SEjI1q1bi4qKzp8/j0gbHBz8jRBCCCFE 8IEKQgtVV1efPn16y5YtGzZsQCNpFJYQQgghhE948uTJ5ORkX19feHj4559/ HhkZiTBrbm5Gtj18+NDXqRNCCCGE8BlooZ6enqamppiYmK1bt65fv37btm2o pgUsrMdzLGcihRBCCCH8hDcqhCYnJ1NTU/fs2RMSEhIZGdnb24skm52dffDg wRs6oxBCCCHE2wKKCF2EOkIjbdu27auvvoqNjU1ISHA2sEeB4+Pjg4ODN5/h 67FjQgghhBA+wNFC6CLUERoJpbRUqowDRkZGaki8EEIIIcQC3LlzZ+/evWFh YRvncL63R4GoqZ/+9Kff//73f/GLX2j6UCGEEEIEITZ96H/5L//lBz/4wc9/ /vMdO3akpaVNTU0tlRgbGBj42c9+9ld/9Vdnzpxpbm6+f//+Uh1ZCCGEECJg QCNdvny5sLDwRz/60R/90R891fI3QgghhBAevOnlb27evPn9738fuXX+/Pnf aM4rIYQQQojngUa6fv36uXPn/vzP//zf//t//1TL3wghhBBCfJ03t/zN48eP +/v7fzlHb2+v5rwSQgghhHgRKCX0EqrpO9/5DjJMy98IIYQQQnjxouVvXnMu LPOvfjLHwMDAEiZYCCGEECLwQC+hmr797W/HxsZq+RshhBBCiPk8d/mb9PT0 15wL6ze/+c3vz8GHpUqqEEIIIURAYsLpm9/85vr167X8jRBCCCHEAnguf7Nz 507+fZ2jIcNsoi35V0IIIYQQC2PC6Rvf+MYf/MEfaPkbIYQQQogF8Fz+5he/ +MVrvvcn/0oIIYQQwiWOf/Wtb31Ly98IIYQQQiyA5/I3f/mXf3nz5s3XOZr8 KyGEEEIIlzj+1Xe+8x0tfyOEEEIIsTDO8jfQ39//+PHjVz6U/CshhBBCCJc4 /tW3v/1tLX8jhBBCCPFSbPkbkH8lhBBCCLE8OP7VN7/5TS1/I4QQQgjxUpZq 3UD5V0IIIYQQLnH8K5B8EkIIIYR4KUvlO8m/EkIIIYRwifwrIYQQQohFIf9K CCGEEGKZkX8lhBBCCLEo5F8JIYQQQiwz8q+EEEIIIRaF/CshhBBCiGVG/pUQ QgghxKKQfyVEIPHkyZPHjx/PzMzcdsHo6OjY2Nj4+Pj09PTs7OzDhw99nXwh hAgW5F8JIYQQQiwK+VdCBBL37t0bGxvLy8tb9zI+n2Pz5s1RUVHZ2dl1dXX9 /f2+Tr4QQgQL8q+EEEIIIRaF/CshAom7d+/evn376NGj77jg3Xff/eijj774 4ov4+PjCwsLOzs4HDx48evTI15kQQojAR/6VEEIIIcSikH8lRCCxWP/qvffe +9WvfrVixYqPP/748OHDHR0dIyMjvs6EEEIEPvKvhBBCCCEWhfwrIQKJ+f7V r3/961WrVq1evfrTr/PJJ5/w5YcffvjuHGwZHx9fX19/8+ZNX2dCCCECH/lX QgghhBCLQv6VEIHEfP8qPDycfysrK7uecfXq1c7OzosXL5aWlu7evdvxr0JD Qw8ePNjY2OjrTAghROAj/0oI/+Hhw4f37t2bmZmZnJy8c+fO2NjY6Bzj4+MT ExPT09P8qmVujPv37xMQooTgtPjMzs4u29m5EJyUS8PZOTWX7MGDB8t2diGE z5F/JUQgMd+/2rFjR35+flNT09gz7KZ//fr1y5cvHzhw4IMPPvjVr37Fll9+ +eXevXvr6uq8jokwQB5MTU05ig7lwGe+WUDO2UqISBo2Q2Cwi7Mv/5JOr7m2 2JhDoYic5RH5zJccx9nG5CXa0lk/keM4v9rc9ZY7trk3hyWA7zkaZ+QInNTr LHZYPpAwr2NaqvjGEbQGW3JYDu6VQv5d+PieR+AbRKDn7vPP64TOIA2WNc4y f8cnc3DM58bcNB5HfnkxEkK8eeRfCeEncOukAlZVVZ04cQJdFB0dvXnz5tDQ 0JCQkMjIyNjY2GPHjtXU1PT19Xnd9IOTK1eu5OTk7NmzZ926deHh4cTn4sWL S3h8k0DGfLVTXV3NRdm0aRNn37dv3+nTp69du7aEZxdC+Dnyr4QIJOb7V7t3 7y4rK+vu7n78dVAF9+/fP3v27JYtW9auXcuW/A0LC6usrHSOZptdvXoVeZCW lrZr1y6EyoYNG/jL56ysLOTcjRs3zDbxSom5SXV1denp6XFxcdu2beNEGzdu jIiI4N/CwkL0z/DwsLM9G5NIjumskJifn8+Xnm4S5+KMycnJ/Lp+/fqvvvqK 9Du/VlRUoGdIHr8mJSWRETZGU2VkZHBqxGdHRwcN1NDQEALVOQvfc1jSyZcI IVJYUlIyOztr3tqdO3fQq+Xl5Rxw586dqFk0LUfbu3cv+aqtrSULMzMzThqm pqa8jk+Orl+/jtzKzs5OSEggdASZI3Auvmlubn6uB2jRICWJiYmEmnzZeZGL hw4d4micZXp62msvkj0xMXHp0iUKQHx8PDFnL3K0Y8cOTl1UVMSlHB0dfbWi JYRYWuRfCeFzUC/cgrnXX758OTc3l3s9KgUh8dFHH61atWrFihWfffYZt2Du xfYocHx8nLu2r1PtY4gVUgSBgXT89NNPt27deu7cuSU5sulJk1IGqgax6vm4 s7S0FAXIBeLs27dvR8W1t7cvydmFEG8F8q+ECCSe61+dPXu2p6fnudsjObZt 24Y8Y8tPPvkENVJRUeF5tNHR0QsXLhw+fJjjoOjY5sMPP+Tvl19+GR8fj5xr aWlBaaAunL1s5BU7Xrt2rbCwcM+ePWaRffzxx0jBdevWcZbU1NSqqqrr1687 soRkd3Z2JicnOyknF3zp6Q6RC/LCAfn1V7/6FSk5deqU8+uZM2c4xQcffMCv UVFR/HT69Gm+3Lt3L99zRpJKkm7evEl2nLMcPHgQ5VNUVHTgwIGwsDCOefLk ydnZWRL24MEDNkan5eTkoJFM0HKo1atXo2ZJBqcgzcPDw87zwcnJSa/jJyQk NDQ0ECjCxfHXrFnD7hyEz3xTWVlJHj09Og7FqXt7e+vq6jIzM7k6hPqjOdgr JCRk586dpNDcP+dBsEk+dDUtcHFxcWxsbHh4OKqSc61cufKLL77g3yNHjtTU 1LCBlxQUQvgE+VdC+BZuhQiY7u5ubscIFe6z3DFREe+///57771nsysgNn79 61+jXviVbdgSUYRICOZ3Cd+cf2UD/pGOzkPAjIwMRJrn0kLyr4QIcuRfCRFI vIJ/hfBAfrDl559/HhUVdf78efvpyZMnFy9eRKohUVB0iLcPP/wQwfDJJ5+s WrUKdWf/RkdH5+bmeooHtAdKo6Cg4Kuvvlq7di2bof1+NYctd2hTyq9Zs8Y8 pVu3bj1dav8qJCQkMTGRrNkc9Zw3KSmpubl5vn9F7pBhbEkGSRhi9fjx40hT UkV6EE4Eh2xyWEszf/n84RzkbuPGjeSUjW001Hz/im1QvGgwO/57zzA9TMBz cnKuXLni5GJoaKipqYnIsCPnZTNCbaEDPnN28w8579jYmD0ItpFXRUVFmzZt +uyzz0gbB1/1DDs1H9iRzKI8OcsSljohxCsg/0oI33L//n3uho2NjWFhYTYQ 3aQFN981z+DWyTfcfJ1h6txqvQaHBxtvzr+ymbXQRY6IQssh3kwoGpxr+/bt yDxObeP5u7q6luTsQoi3AvlXQgQSr/D+IArkk08+YcsNGzbExsba/FePHj2a nZ3lVxN17777LnLu888/37hxY3h4+JdffunpFCUlJV26dIldbG6l8fHxnp6e I0eOmG2CGkT+ITM+++yzdevWrV692plxKyYmpqqqqre39+lS+1fr16+Pjo7+ 6quvzMbh+7S0tLa2NuLg5S+RF5IRGhq6cuVK27KgoIC82wio+Ph485o4F4m3 VyA5OLkwB4+9ONq1a9dQs0+f519xTOJG3gmjrfxo7h+HtTnziV5DQ4OTC5rQ iooKkkTM7RSciyNwdvbldE56MjIybty4MTExYecdGBjIzs7mMvETwUF1k1Su Dqfg7I47R/QW8DOFEMuG/CshfMvw8PCxY8e40ZtWMfHAXTIrKwtxUl1dff78 eTaIi4tDTvCr3X937dqFTrh69aqvk+8zfOtf3blzB4V2/fr1rq4uVJDXIHYh RMAj/0qIQGK+f2Vv0rW0tEw8w+Yhp6ryZXJysuMmmdOFJHj6bBTQgQMHTNS9 ++67bNnZ2dnf38++ra2tRUVFSDj2ev/991esWMEZ+cm8FJRGWlrali1b2GvV qlU2B1R3dzdnRGmcPn06MjKSL+1RZnh4OCrx6VL7V+b8bNq0CV165MgRpGZj YyP6Z3Bw0MtfMoOIJPElkoktOzo6Hj58WFlZuX37dtJJLhCu+/btI53kHbnL EU6ePMmX7IiaRcI56zbO968++eQTkpGdnd3W1kYGicPx48dDQkLY19wtTlFS UvL02WxjXILQ0NA1a9Y4Lhy/kmzyTsLICwck4KSK5Dljt/iLxo6IiCA9BIEd T5w4MTAwQGpHR0dJOXEw+c2vGzZs4Jiay10I3yL/Sggf8ujRo76+PpvwyhEV 3FjT09O5EXNXbW9v566NGMjMzGQbRywhbxA5aKT5B7x//z6ixRFa8xdemX/b tSVXnLX82NIWuLHj2EJ7tmiOLcLivP7/5MkTvnEWi2HH577VaBNBOKv1eU25 yUE4oK2M45lmZ9mX5044sLB/NT9VXov1zF/mxiIDhALdkpGR4Ygo1FddXR1X ykmJPbBDc/b29iJWnzsjmee1cL8Ijl0Ltve6Fs5iOoTRDqIFcYTwIfKvhAgk 5vtX6IqsrKzq6urrz7h27VpXV1dDQ0N5eXlMTMx7771nL8cdOHCgvr7+5s2b HAepwC6xsbGOqMvOzh4ZGTE3CeVw6dKl/fv32wQRbJOSksJhbW5wlAaH/fLL L995NqdWUVGRaQZ2v3DhQmJiIl9yTGRPSEgIyXi61P4Vv/IhMjKSvPN9bW0t WXbcJ09/aeXKlTYE/fTp06SELWnEkFuk2ZlgwSaPamlpefpsDUS2ROXavGFf fPHFrl272PHp8/wreyuzuLgYlUV2iAPC2HHwSCSn4NTOkcmRTTL2zty7jQSW UKOpODLx51cbT8UG27Zt40TNzc3sy9Xk8vGTiUkSTKBMyqLQWltbCSZfOmfM z8/XUuBC+Bb5V0L4Cu6qg4OD3LjXrl1rsmHDhg0oAZSPra384BnIKr7hlhoR EWE3bmQDd2Fu617HvHXrFjqhsLAwISGB+z535C1btqDBUETcyisrK50h057Y Iz+25+A7duxgy5KSkvb2do6DBkNpfPXVVxwkOjranq/ZTFA2aQCSwIZnAzui r/r7+50jm4eDmEGl2GZJSUmoF88B2GSNOJDrQ4cO7d27lxMhPBBmpCQ+Pp4z Xr58mQ280rywf0X6ndmrUlNT0TCIRs8zPncZHc5C1i5evIhs44COiCIxqE1E CzrHUtLW1oYiTU9PP3jwIClEuHoe3/NaEB+va0EeF7gWTU1NXIudO3d6XQuO w7WgADjXgvRwLRB1Ly9qQoilRv6VEIHEfP/KJj5avXr1p1/nk08+scm933// /W3btrELN27n4R3iJDk5edOmTRyB+3hkZCTixJkt3JyWvLw8duf4bLNnz56K igqr/p4+EgLA7vK2L6BP+JcvEQDsha6rqalBItp8U0vlX5mnhERBKXEEyxdp QK54+Utk7eTJk62trWxG9EwTDg8PZ2ZmOm/5HThw4MqVKyYaLRdXr17l1Igi G8G1Zs0aJOLT5/lX6GEUcl9fn/N40bJPZPj13Xff5RQkgOSNj48jqJBkNuCN X4mGs2KgzYrf3d1dWlp65MgRokeqkHA2/gqJhTrlQrMXEis3N5ezOJHhCF1d XRZbOyNHIJ2kdnmKpRBiPvKvhPAVaBhqXHl5uY1ntgdVGRkZ9qBqPvbMjm2Q GWvXrkVjcC92fuUOjn7gPssBU1NT2Wz9+vU2IwFaKzQ0FLVw/PhxVNbAwIDn GCqor6/ft2+fDZA26YL8qK6uTktLswktbdLLDRs2HDp06Pz58729vYgB0o9W 4ZjOqDBUAUmyIfQGJ5qenj527JjzqNFzmD0HYYPBwcG2tjZEiK2wzIlIsy37 EhYWlpKSwjHRZjbcyDnywv4VR3MkUGxsLGlG2zi/zpdJRIxr0d/fT2IIoDNc 3HkIiF7i0jjTTdgDO45MVBGKzvdOrp1rQQy5FuTFZDC5CwkJYS80EteCZHiN obp48aLXtbDnv3YtyKnXtUDaOcJYCLFsyL8SIpCY718tjK2tgzbwmoOdm7gz 4QPyKSIioqSkxBnsPTJHTk6O41+hfM6cOXPt2jXu4/n5+c5UEvNnjzfpYnrA aGxsXHL/auPGjQkJCQhOr/jMF04cDZHj2XDN3wbVSiK7u7udCPAvCi0yMvKd Z29QkhLyPn/fpKSk1tbW4eFh5/hIRwQhUtPZ5sSJE6hfDsspkIvO9+np6Wjd qakpZ1/+RbmRU+KGKHWmLT19+rSTd3uvEEnspNbeFU1MTPSUiwRzfHz89Yuc EOLVkH8lhK/g9ldZWZmRkWEaBg4dOrTAbdEeA9XW1hYUFHC75+Z79epV531A u63v27fPHgs6C68Y/Gsjnz/99FNu621tbZ6SwNO/8pwwk8/OBA785YA2lSja g8Qg9pAcFRUVzhSmNhmp53SalmZEhS2kyDYoB2fckf2amZnpTExqz+zsXMDZ yQs/7d2712uO9DfhX9l8VqQQQWWz5Rs2nN5zwDmajczawDnEJwrK84HdwtfC FsFBLz33Wnj6V57XwhbTsWthK+nY1XSuxWuURCHEopF/JUQg8Qr+lc3ghMJB kNjgoqdfX5fQxAO3+3Vfh9u381AvPDyckyKcUCbZ2dmOWEpMTHRWGDScyRam nnH//v2nS/3+YGRkpJekMeYLp4MHD3oNAie1ra2t6EBnG9QU+pBoOHnnX9Mz FkPIyckhLxzH6/hoPNSUpyTu7++/cOFCXFycl3+FwEPHxsfHO9/bSoieD2rt dQB7nWF6etoZM5abm+tcC9PJpNBJLSlHgJEL58jmyM1/KUAIsWzIvxLCV4yO jhYVFXGDtpVNuIEiOZxJEubDDXdoaAi1UPsMG1ZtFhaaISsra9u2bY44WbNm jfNmH8rErCFOFBsbW11dzb7OkT39K9tx9RzctT/77DMbEGXCxoysjIyMgYEB Gz5dU1Ozc+fOL7744p1nM8+jLpwj20Sm9ujKXJfMzEwnj2iVixcvJiQkOIaV +WMkmAPaTJvmI23ZsgVVw6Ecv+5N+FcIvO7ublJIrleuXOn8ajGxoVBtbW0c obS01JneYfv27ceOHbPHrzZM3a6FvYHouQiOXQ72MjvrudfC07+y89rlIBpe 18KMviNHjng9ZBRCLAPyr4QIJJ77/qCXmwHciJ0nUzY/eVxcHLLHGdZeWVmJ YrHlpE32cMte8WIiIyNRHZcuXbpz5w7KwTm7vXnn+XjrRSytf2Wz1s9fUnm+ cEKDIZls9UDDJN/+/fudbWyE1cLk5uZOTEygDJ8rzAiLc3wTjWgkL/8KEfVc X2vhCULtXU70m7OX+Y0EZ4HUcl2cqSSEED5B/pUQvuLWrVtoldjYWHOWuC0W FBQs8C6YPXpDI80+wyZbsNnREQBoKpu40hZ8QasMDAxwW79+/Tq/2rIsyC1k FeKqvLzcubl7+lfmpCFg8vPz6+rqent7kVVoG1suxzbwdF2uXr3qTEfA2VF6 RUVFzoNIz1/RgevXry8pKXEWoT537pxN48kxSTwJQEgghzgy2qCiogIVFBIS 8s6zaShycnLQctPT00/fjH+FDCMm/CUBaWlpzq979+7l4NeuXePU9g7ji/wr z2thUvBF14JAPfdaePpXnteC751rsXPnzneePbic74AJIZYB+VdCBBLz/Sub gQqx4bxNhjJBS1RVVaEZuHc7XlB6ejo3d3NykC7OKni2Yh3HSXwx2dnZKIor V66Mj497+VfoCps5amGW1r9C7Nn7jF5nea5w8nplgPg0NjYmJCQ422zcuDE+ Pp5vFogACu3OnTuv41+hjmpra82/MmmECHzp1Ao21YOnf7Vu3bqIiAiitEBq CY4zo5cQwifIvxLCVwwNDaF5du/ebaNxnIVUFouJiiNHjnAQG2G1bdu2zMxM m1B0enp6YmICQRUdHb1+/fp35oZzf/rpp4WFhc4SKp7+lT0u5F8URXd3N7vT LLAB8sMZYu05z5VN4WWiyB60ofecB5FtbW3OqLC1a9eGhYVVVlY+fTbzVXFx seMC2XTl6Cu0Cie9f/9+a2srumL79u3vPgNh09fXZ2LmTfhXdmROTQJycnKc X1Eszc3NnsP4X+Rfzb8WJMy5FrYWoV0LG7Fmy/d4XgtP/8rNtaD8eM05JoRY BuRfCRFIzPev5s9AZRONImycB3PzvSZEjjOpgq0hGB8ff/wFIB4QQrW1tdeu XbPxV877gzb706u9P5iRkeE87DPIRVlZGTly418VFRUhObzis7BwMuaPv0IC EdLs7Oz5GXdAzr3++Ku6ujr73gKYm5vrNbmo1/uDd59BYhxBtWHDBhQj510g tdXV1Vwsz1FnQohlRv6VEL6CG3FCQgI3dxux/OWXX6IuXuE4zmLN3H9tpDpS wfPRG5pncHAQhWCiwt4BTE9Pd+ah8vSvVq9eHRISUlJSYo4K+6IBbLkc520+ W3TGpkcwScPR3nk2Xgj54Rz5/PnzzmLH5uSgx54+excSiWU+D78i8C5dusSh nEdmNpYJtWDn5eB79uxx5n73T//K81rYZbVrYfNT2bA0z2th7wB6XgtP/2rN mjXOteAqsK8Ndyd3XtcC7foKJUcI8crIvxIikHDjXz2du49zs+7v7/eUENz0 a2pqTGbU1dXt3bt3w4YNfI+ui4mJQZxMfx3ExugcnHF8fHxqasoma+Ju7ogi z3UJDTYbGBjgRCQycw6UCapgeHjYy79C0vT29no6P2yAijPP7c35V3zmvHzv bJOSktLV1YXscfJOLjgUuXZGtSFWn7u+oXv/yiJw8OBB5/sjR45wUs+pFUx6 Id6IHnEuLi62OfMLCwttKcl35ubS59+2trb518tJrdmGnjNrCSGWGflXQviK pfKvbLZwe6xm79nl5+c7bxcaSAU0mJeocCYu8PSv1q5dGx4eXlVV5XkKu8V/ 9NFHzhItx48f93xjLjs721ku2ZYttik9PR9E2qQK/MT3CBWvd/RID1rCFjs2 kAfkoqCgwHkv0vMI/ulfeS6O43ktPKWOXQsS7xzfcxIJT/+KfHldC3PATp8+ 7Ux/GhER4VwLIcSyIf9KiEDCpX/13C1jYmI8lydOSkoKDQ19Z25SUA5SU1Pj PJjjA9oMCTQwMIAO6e/vNwPHHlGhcJx1CU3wdHZ22r4wMjLCvwgAVA06Yf/+ /QgGjsb3tlCOkx6bWZ3vbcgW27S0tGRmZtq0nG/Ov5qamhocHLRHkzaiyaSd 52ile/fuscvw8DB5JwhDQ0M2TszN8V/kXyEd2RJJ6YykMiFqktImJkWecTUJ AtFjS2eOejLrzNA+f+4ve2jIcUjtzTlIDyHVos9C+BD5V0L4Cu7y3Kmjo6Nt 9vI1a9YgIV7hOG1tbY4s4SCopvLycmcGKsNuwcePH3eGpiMAkAHci59+3b8y uVVbW+t1lrKysi+//BKF4zg2V65cefpMGPAru69evdrET3FxMcphdnb25MmT KCJbDNrTofKc193Izs52BsN74jkValhYWHp6uq0A6J/+VWtr63OvhWeOPK+F HT8+Pt65Fp7+lV0Lz/nwPa/F/LMLIZYN+VdCBBLu/Svu4F5zVbGlMzicvyUl JfYYyyYFzc3NdZatsQdYCIwvvvjC1hf2XIbYc0EckxCnT5+2AVokDzGAGkH2 cNh169YhjWxCBnONjhw54qTHc+ICMkXbgk5wRMub86/sHb3CwkLHEYqIiCD7 5hSZd9fQ0IB427FjB3knCwTKBOfr+Fc2P0NeXp6TC0J34MABxO2DBw9QdChP UhUSEoJcJO+kigttrwPU1dWhEm28HFckKioKMUku7AUEElBRUUHYSS2/EkOO Q1Lni1UhxLIh/0oIX3Hr1i1uoDExMZ7zX3n5Tp7Yrd+mtDLsBtrS0oJuCQ8P f2dusgVH0szflxu941/FxcWhGczV8fSvbLj7fM8EFcf9fb5/ZXC737ZtG/d3 kyucqLm5GU2FSLOXCm2lPLSHrVqI1kLDeE7yaYsde44Zm39kZBt6Bkn21F/9 K7sWYWFh78yNZHvptbDj27Xo7+9/+nX/yq4F4srrCOhkr2sh/0qIZUb+lRCB hHv/Cq3itVZgdHR0SUmJeT7cylFQyBubz2HlypVpaWmdnZ18Pz4+fv36dWcS S9sgOTnZ8a8aGxsPHDgQGhqKPLDpMTMzM69du0bjwO5IRGdCBv7yubq6mr1m ZmZGRkZIjzMMfuPGjfv37y8vL0fqcOpLly6lpKSsWbPGhrK/Of8KefPo0SPO u2nTJpvBng9ksKamZmxsjPCiowiUeXTk0fwifnV5/Bf5Vw8ePEC5FRcXO9LI jsw3Q0NDvb29V69eRZuRpBUrVhAiQue4aqg7TopsIz1IaBJcUFCARuWKjI6O EjqbiPWduclj2SAvL0/+lRC+Rf6VEL6COyOaAVFhqzBzS83Ozp6YmEAaPXd7 7v7ImKqqqqxncNtl48uXL2dkZJh/5TlHuic2hvzN+Vft7e3OLR5ZgpZDANiE DCbDSBiZJQ02naZ7/wp5ZmsUOkIIgffUj/0rr2vh9Sbm02cCjxTKvxLi7UX+ lRCBhHv/CpFgj+ecLT0nN7A53ouKiriPr1mzBmm3atUqNMzGjRsRKty7P/nk kxUrViAAUBGIJUSOM58SsrCrqystLe2DDz5wnmyy77p161BWHM0GtLOv57Bt eyhWUlLiyBJbTGf16tXsixThjPxFKSHG3qh/ZdiYJdQL6UTfclLSj1oLDQ0l JiTmg2ccPHiwqalpaGjI5fFf5F+ZsCTNxcXFhNSZgJSIERMb58ZV4FrwJR8Q 287sWDZn15EjR4gJv5JgUkg6Q0JCSDNhd6ZrcKZ+1cuDQvgW+VdC+Irx8XFu 8dw07c4I6enpXhNOemILE+fn53PvtvWIL1y4MDs7y93f8521TZs2cVivcVw2 l5TX+4N1dXXmmby+f4VsQPAgexw3KSsry3Go2IuDcwRn+/nvDyIFnRFlnlRV VYWHh3PMd15j/BX6sK+vz/l1/uSrb+L9QXuQ99z3B7kWubm5jn9l12L++4Py r4TwW+RfCRFIuPevHjx4MDMz47lu3fxJFWpra9l948aNK1asMDNq1apVa9eu 5S+fP/zwQ/RDREREWloaksYRbLa0DQKGu7/5POZivf8Mm1eTnzIyMpyJRu2M 6JwdO3agTNjGPC4b+s652GXdunUIEv6SHk7tNWHF0vpXJvD4lXRyarOwVq9e zecVc5Akc5YImnOE1/ev+BU1m5KS4pzXsBjygfMiF0NCQsiv87DYJnEtKCgw a5EgsBmJ5AjExKYj4y8JRgqiMz3FpBDCJ8i/EsJXcNNEIZSVldmTOHu4k5eX 96K15GxgeVhYmI1hduZ7dx45vfPi+dvtAdOL5kt/ff/Kc354e/CHjsrOzt62 bZuX72Q8d/52Zw5zwyZSOHXqlDN/O8dEYyx2/vb5YgzJx0G8FutZEv/KroX5 eC+av/25C/Q4E5zKvxLirUD+lRCBhHv/yvyiM2fOOG/kzX90iIpAMyBLDh48 iHT54osvbNgVfxEPsbGxJ06cQMagne7du+d5ZNSCraaH2kHXcVh2MSOFg4SH h2dkZKDQUFAPHjzwlBY2dgsNiSpgSxKGdlq3bh0HYRdSi7rg14SEBCQNqbKh 7MbS+lcm3viptrbWRqSTHrODSFJoaChJIiVtbW1DQ0OO+/T6/pU9HOR6nT9/ 3mbV8DwvkdyzZ09hYSFR4kI7M+rb1eQbvkdwesacfTkCqhuxWllZSUxsYe5X LmNCiCVB/pUQvoL7bH9/P/fZtWvXmmzgppmSksLdGeVjd0nD5rxCGjkTe3qO aELGoJH27t377rvv2nQK2dnZKBnEmJ2IuzMiobm52cY72cjq9PT0GzduTExM PF0K/8oeGtqiM8bmzZsRSOTona+vVzh/e5uuAU3V0NAwMDDgyD8iMDY2xomQ fOSLZCMmS0tLbYrUhf0rpJHzYNRzyLcJFc6CcvN8e/Gl/pUzxN14kX/V29tL MpCmzvB1juNcC0uA57Wwx6Oe10L+lRBvBfKvhAgkTHIgHtY9wyZu8px8wIFb eWVlZXh4+MaNG21ok81G5QgYG/Te1taG/DALKywsjBs3fxFymZmZSAWqvDOp gic2n1VtbW1aWhpyAnmD1AkJCYmKikK3FBcXey3W7Ow1PDxcVVWVlJTEKUgY 5yJVyBt2QeaxFydFk6DHnNX3DOQlWSB55IXdyZqtpejJ1NQU6oUdnfigPRzp Mh/z04qKikgzKSf9nGLLli2oGvKF9PV63cDN8W1k1+HDh51tOL7jXxm3b9/u 7Ow8c+aMnTc0NJToEQoiidZC3LKBI48dbBSWE3OurMWcCxcfH19YWOjMUSaE 8Dnyr4TwIUgXRAI3R3vj7P3331+5cmV0dDSCh9toT08PVZIN+Mx9nNuos5af PYNraWl5OjeWiVs892Xzgt59991t27ZlZWXx64MHD7gpowqQNLt3716/fv07 c1NQrl279tSpU86DpNf3r+yhIRrp888/X7VqFWkgqR999JH5csih1tZWz/FL iA3ShvCw7d+ZmzVr165diKjJyUnSTMJQC6gs1JdlCtAVBGR8fPzpy/wrzwej KJyIiIiysjKOzL5Il/LycvJITl36VwSHgNi7lsaL/KsXXYvm5mbyS8C5FihD r2uBNHKuhfwrId4K5F8JEUjY462ZmZnbz+CGvsB4G/O7RkdH2ZK/XiOpDFv8 juPwq23MB5QGSmCBI1tKZmdnES1sbDvavhwKgeQ18spzLw5rp2N7dnTWLuRo JkK8FgCanxdO+ty02dSd7OjEh89886LJoPjJJKiTfYN/Lfte+7o5vo3sIoXO NvOdqIXPSxyckVfPjZ4Tc+d6sSOlgmPOdxqFED5B/pUQvuXWrVuZmZmxsbE2 ZcE7c8v+Hjx48NSpU/X19ZcuXWpoaMjPz9+/f39ISIiN2GHL3bt3nzlz5tq1 a0+fPao7fvy4857dV199FR8fX1FRMTQ01N/fT70+ceKEjYh+9913165du3nz 5rNnzzoPrV7fvzLsER5ncUZ5mY2Tmpp6/fp1z3cDTSpUVVVt3br1008/ZeM1 a9aQwdzc3J6eHtJMWPg1MTHRcm3v4mVnZ5Mje2C3sH/l6S8Rlo0bN9qRiVhn Zyefw8PDicNz/Sub2oIMOvNT2dJC7GjG2tMX+1ee18K8O07NtSgvL3euxcmT J51rwV+ywK/OtZB/JcRbgfwrIYQQQohlRv6VEL7l/v37AwMD9fX1jvNj80d9 9NFHnzzj448/tnfoHK+mrKzMeXxmXlBLS8vRo0edcVzONAuff/75unXrOILN YPnee+/FxcWdP3/ecwrKpfKvWltbndX3zMKyUVgFBQXO8jqe2DLT8fHxzvSk bGzr7JBymyzCch0WFpaent7U1OQMtl/Yv2poaLBFqM3xsyN/Ngcb89cZ9zXf v7J45uXlOUtR29ynpNPxDF/kX3leC4uD57UgXxyHvbyuhefrCfKvhHgrkH8l hBBCCLHMyL8Swrc8evRoYmLi6tWrsbGxmzZtWrVq1YcffmgLprz3DFs5ZeXK lR9//HFYWNj+/fvr6+u9hkDbzOFxcXGrV6+2BVO8Fl754IMP2P2zzz47cuTI lStXPF/kXyr/ilzk5+fv2LHD8YVs3WQS9twB2zZNwdGjR9nGBiw5S+046bdl awjO6dOnu7q6nH0X9q9IGymMiopasWIFB3EiaQvQcLotW7bYQjxGRkZGb2+v 419BUVHR2rVrObuzF0dzZvF6kX/l5lrYHGWe12JkZMTZV/6VEG8F8q+EEEII IZYZ+VdC+JwnT548ePBgfHy8oaEhOzs7ISFh27Zt69evX7ly5UcfffTxxx/z OTw8PDk5+cyZM21tbXfu3Jk/zcLDhw/v3r179erVs2fPpqamhoWFff7552Z5 2at5UVFRubm5LS0tg4ODXpMnLJV/ZX6U57p+Gzdu3L9//6VLl56bcZsFi/S0 traeOHHCZvj0XGqHXKSkpJSWlnJYm3/A2Xdh/8qm4qyqqoqPj9+yZcvHz2B7 s8Kc6T0NthwZGeEUzhGuXbtWVFQUFxdH9Gwh5sjIyJycHMv1wv6Vzb3Q1dVF 0EgkuSAvdjWda0F+uRYDAwNekyrIvxLirUD+lRBCCCHEMiP/Sgg/4fHjx9ev Xz979uyxY8cSExN37twZGhpqy7Xs2rUrISEhLy+vvr7+5s2bzgI387l161Zz c3NBQUF8fPyOHTtCQkI2b94cHh4eExOTkpJSVVXF7pOTk157sUtaWlpUVNS6 devYiy0vX77stU1NTQ0bkB57k85rNJRhs5eTfmdpmD179mRnZ7e1tS2Q8amp qYGBAdLGeffu3esstWPLvpAXEjM4OOi115UrV3Jycjg+ZyGDsbGxFy9edH61 RRvJF2c3C8uIi4s7evTohQsXhoeHCWbOMy5dujQ2NubpChIoNsjMzOTgXIgv v/yS45w6dcpyXV1dHRkZadHYt28f0bD3Cj2xdQbZZVHXoqmpiWvB1V/gWpw/ f55rwdEWOLsQ4o0i/0oIIYQQYpmRfyWE/2BjqKampiYmJpwFUGwFGVsAZXZ2 9kUL1hi2OAtbeh2Bz7amzHMXa2YXTso2t2/f5lxsyYm8tmFfW8mFbdjSazSU wZFJnufyMWzptczNfGztQq9lXzxz/dy5s2yVQCfNbOmZZrP4bBt+eu7CN2zv uRCP1xuO5oCxsbMMjbMGjRMNW6zH8/sXXQvPRXA8r8V8H9Kuha2TaNdifvTc nF0I8UaRfyWEEEIIsczIvxJCCCGEWBTyr4QQQgghlhn5V0IIIYQQi0L+lRBC CCHEMiP/SgghhBBiUci/EkIIIYRYZuRfCSGEEEIsCvlXQgghhBDLjPwrIYQQ QohFIf9KCCGEEGKZkX8lhBBCCLEo5F8JIYQQQiwz8q+EEEIIIRaF/CshhBBC iGVG/pUQQgghxKKQfyWEEEIIsczIvxJCCCGEWBTyr4QQQgghlhn5V0IIIYQQ i0L+lRBCCCHEMiP/SgghhBBiUci/EkIIIYRYZuRfCSGEEEIsCvlXQgghhBDL jPwrIYQQQohFIf9KCCGEEGKZkX8lhBBCCLEo5F8JIYQQQiwz8q+EEEIIIRaF /CshhBBCiGVG/pUQQgghxKKQfyWEEEIIsczIvxJCCCGEWBTyr4QQQgghlhn5 V0IIIYQQi0L+lRBCCCHEMiP/SgghhBBiUci/EkIIIYRYZuRfCSGEEEIsCvlX QgghhBDLjPwrf+D+/ftDQ0ODi2RgYOA3i4Fdnjx54uu8vi5vOlaBEaU7d+4s qmwY/f39txbD9PS0rzO6BMzOzk4snrt37z5aDAFQqIBcTy+SqakpwjXuAgrt w4cPfZ3F14ULPTo6utgGCnp7e91XVTam3Po6r8HOb+RfCSGEEEIsL7+Rf+UH tLW1/ff//t//n0Xywx/+8N/8m3/zr13zd3/3d3QkfZ3X16W1tfW//bf/9stF 8hd/8RcuYxUYUdq5c6f7guHw3e9+9/9dDMePH/d1RpeAurq62MVz9uzZRbkT U1NTvs7o6/LgwYOKioqcRZKVlRUXF7fTBUT1xo0bvs7l6zIzM7Nx48bFNuZ/ /dd//R/+w39wX1XZuKamxtd5DXbkXwkhhBBCLDPyr/yBxsbGH//4x3+6SP7j f/yPv/Vbv/UN13z3u9+9c+eOr/P6ujQ0NHz/+9//z4vkW9/6lstYBUaU6EG7 LxgO//bf/tsfLYa0tDRfZ3QJqKqq2rp4Tp061b8YJicnfZ3R1+XBgwdnzpw5 tEgOHjxIuL5yQXh4eE9Pj69z+bpMTU2tWLFisY35n/zJn/yLf/Ev3FdVNi4v L/d1XoMd+VdCCCGEEMuM/Ct/QP6Ve+RfuUH+lXvkX7lE/pUb5F8FD/KvhBBC CCGWGflX/oD8K/fIv3KD/Cv3yL9yifwrN8i/Ch783796/Pixy9nnvGaiu3v3 7j0XzM7OBsDMfpqA1D2agNQNmoDUPZqA1D2agPSlaALS4OE38q/8APlX7pF/ 5Qb5V+6Rf+US+VdukH8VPPzG7/0rei5JSUluZp/zmomO0lXjAm7HASDjNQGp ezQBqRs0Aal7NAGpSzQBqRs0AWnwIP/KH5B/5R75V26Qf+Ue+VcukX/lBvlX wYP/+1fj4+N0TNzUPq+aWFhYeNYFaPh79+4tbZqXHwkw90iAuUECzD0SYC6R AHODBFjwIP/KH5B8co/kkxskn9wj+eQSySc3SD4FD/Kv5F+5r7OBIS0kwNwg AeYeCTCXSIC5QQIseJB/5Q9IPrlH8skNkk/ukXxyieSTGySfggf5V/Kv3NfZ wJAWEmBukABzjwSYSyTA3CABFjzIv/IHJJ/cI/nkBskn90g+uUTyyQ2ST8GD /Cv5V+7rbGBICwkwN0iAuUcCzCUSYG6QAAse5F/5A5JP7pF8coPkk3skn1wi +eQGyafgQf6V/Cv3dTYwpIUEmBskwNwjAeYSCTA3SIAFD/Kv/AHJJ/dIPrlB 8sk9kk8ukXxyg+RT8CD/Sv6V+zobGNJCAswNEmDukQBziQSYGyTAggf5V/6A 5JN7JJ/cIPnkHsknl0g+uUHyKXiQfyX/yn2dDQxpIQHmBgkw90iAuUQCzA0S YMGD/Ct/QPLJPZJPbpB8co/kk0skn9wg+RQ8yL+Sf+W+zvqVtHgyxyvsGIQC 7BViFWwC7JWL09PgE2CvHKugEmCvHCUJsOBB/pU/0NHR8etf//p/LJKf/OQn //yf//P/2TV/8Rd/4W/C4BVoaWn5r//1v/58kXzve9/7nd/5nf/TBX/zN38z MTHh61y+Lnv27Pn9xfOHf/iHP1gM3Bx9ndEl4OTJk+8snoiIiOOLob293dcZ fV3oqO7fv//zxfPFF1+sd8HmzZv9Rz69MtPT059++umPF8kPf/jD//Sf/tP/ 5Zo/+qM/qqur83Vegx35V/7mX9HlmZ3j/v37dPcezfHSrlCw+VePHz+mmSIx g4ODQ0NDY2NjfL579y5Be2msgs2/IlC3njE6Ooo4JFBWtBbYK9j8K6JBTGht +vv7KVG3b98mUDMzM5QoCtvCJSqo/CtCMTU1RaAoTgSKEmVVjybrpXZNsPlX lBwK0sDAwPDwsJUoQkeJWrjqyb8KHuRf+QNUTHTguUWyb9++3/3d3/3fXfOL X/wiAJyZvr6+6Ojo7YskNDT0/ffff88FO3bs4G7i61y+LhcuXNi9eP7pn/7p jxdDYmKirzO6BGzatOmfLZ5/9a/+1e8uhl27dvk6o68LQuuv/uqvfnuR/Mt/ +S//7u/+7iMXrFu37tq1a77O5etC6xEVFfWrRfLBBx9QQpJck5qaSt/T13kN duRf+ZV/RZeHxPTNQe2g10P26bc+fPhw4S5PsPlXtFFdXV11dXV0b9PT00tK Ss6fP0/be/PmTfrRhGuBfYPNv+ro6MjNzc3JycnKyiouLiZoV69epWhNT08v sFew+VfUO65adXV1fHx8RkYGrQctQ3t7+8DAAIVt4RIVVP4VDVFzc3NFRUV2 dvaRI0c8qx4xfPz48QL7BpV/RSgICAUpJSWF2pefn0+npqWlZWhoiJK2QHsu /yp4kH/lD1DjLl++fGmR0C793u/93v/hml/+8pcB4F9xJ4qNjd21SGiZ6SCv cMHu3bsDwL+ieOxfPO+///6iRCm7+DqjSwA37leQmr/927/9vy2Gbdu2+Tqj rwuygc7FYgP1zW9+82//9m+Dyr+iwflgkaxatQrlf8Q1qN9bt275Oq/Bjvyr 5/pXj3zBwzlmZmb6+vp6e3tv3LhBl9Ae3NN1nZ6eJp2zs7P0Ac3O8uwBLb9/ 9eTJE5Jx//79KV9ATNrb26urq/ft25eQkJCVlXXy5MmqqqqLFy92dXVdv37d xjyY+2eD2ZxYLb9/RU+WRpUr65NYNTU1HT9+PD09PSkpKTMz89SpU5WVlRSY zs5Oytjg4ODIyAiBYksCRaEyC2L5/Ssbr0IafBKl0dFRQlFWVoY4p2PCHSov L48OPkKUQPX09FiJGhsbY2NqItXQGW60/P4VVY9AUaiWP1CUbWJCoA4dOuRU PWpifX391atXaf8tUGxmVc8zUD7xr3wSJaCvSs2iFJH+lJSU1NTUgoKC0tJS otfW1kaguKZsYCWKxoFralVvSv5V0CD/yh+Ykn/lGvlXbpB/5R75Vy6Rf+UG +VfBg/yr+f4VvS16XhM+gu7M0NAQIqGrq4v+YNczuru76Zr19fVRa+j12MtN jou1/P4VvS2SSl+1paWledlpampCcNKJjoqKom+7adOmzZs3b9myJSwsbPv2 7dHR0fQWc3JyKisr6VYTNDrUzltgy+9f0aJyX+jo6Fj+QEF1dfWZM2cOHDgQ GhrqGajIyMg9e/YQqJMnT1ZUVBDS3t5eUm7VYfn9qwcPHlCkKfk+iVLzXKE6 depUYmIitz/PQNGYIKoJ4LFjxyhylJ/r16+Pjo46hsPy+1dU/xs3btA++CRK tJnl5eXEJCQkxAIVNgclitBxTal6xAQNTyJv377NlXUu8XL6V9R3mkdazuWP kkGzXFBQkJmZuWPHDqf2kUiyQ9VLSEjIzc0tKiqiKevs7KQ5paGgRMm/Ch7k X/kD8q/cI//KDfKv3CP/yiXyr9wg/yp4kH8137+iB0FX4raPGB4eHhgYsM5p pwf8axYWP7EB3Wci48z4RJ96mf0rzksyurq6KioqynxBaWkpvb+oqChuTBs9 oJPIN3FxcQcOHDh+/Dj9ZaQp0SNc09PTxAppscz+1eTkJCelpFHelj9Qp0+f PnnyJI3zxq9DV5pAoUUPHTpEoIinOTMjIyOzs7Pr169f7I3yG6/nX3FS5H1H R8fyh8ghJycnMTGRQuUVKwKFrqbDwm3L3sGk5FNbZ2ZmaC4qKyuX2b/iGhGo ixcv+iRKNK2UqB07dnhFacuWLTS8VD0CdeLECat6BIrm1AK1/P4VZ6RU+yRK ZXNtFDUrIyNj+/btXrGKiIhAZaWkpBw9erSkpKS6uhrpaC+r0lJ9+OGH8q+C AflX/oD8K/fIv3KD/Cv3yL9yifwrN8i/Ch7kX3n5V9bl6evru+4j6H/19vYS Ri//ytPIopFxxmKNjo7S8adDvcz+1fT0dEtLCz2gmJiYnT7CpBH96JCQEK/u Id+EhoaGhYXxa3JyMq3NhQsX2tvbiRW9/mX2r7hM9FIPHDgQHR29/FHau3dv QkLCfLfBM1DUFzZAkOfm5tbW1lICuZ0t9kb5jdfzr+hEDA4OFhcXL3+IHFDm SUlJ8/2rkDlsCA2/Es9jx45VVFR0dXXNzMyUlZUts3/V3d1NoA4fPuzDWG3e vHl+vYPQOShRERERVL2srCxqXEdHB4FaZv/K3tslAb4KEfWdLgYVn/Q/t0TZ iKydcwXv+PHjRUVF1rDTkZF/FQzIv/IH5F+5R/6VG+RfuUf+lUvkX7lB/lXw 4P/+FXWWDnjYIqGPWVJS4mYRGfpWs7OzzulsWif60YvqWi4hN+bo7e11/Csz rMDcLSJMB2dgYMBZ/Ov+/futra3LvAA0rQSd6MbGRrqHGcsODQiddwQkLRXd ZOsP2ntMO3bsoD8YExOzb98+NmCz/Pz80tLSlpYWokcPGqW6zAtAj42NURpJ xtGjR5c/VmlpaSkpKXSlPUde0aG2t70IFO12cnJyenq6zfhEoCj/CM5lXgCa EnX79u26urrlD5EDgSIaRMbT2aOAOSUqMTGRQOXm5p49exaNSk28d+/esWPH lnkBaBqKCxcunD59evlDRNXjqpkn4/hX9qIlVY8SRXNNoChRFDzKPCXfpnu6 N8f+ZVwA2qZTo+4vf5QcUlNTzX9zPCtK1Pbt2wkUPTUKGyWKkObk5FCiuKY3 51izZo0WgA4G5F/5A/Kv3CP/yg3yr9wj/8ol8q/cIP8qePB//2pycpIuQOwi SUpKqqmpaXBBa2ur5+TexmPfQReP/ovjX9ls5PbO4NjYmC1VT4Jt4ydzPPXR AtAk4NGjR7O+YHp6enh4uLGxEV1knWi6rrRaNEFe7wxOzS1Y/+DBA2eusOVf ANoG9YFPYkUc6MIjtq0HvWnTJrLvOZsTXf6RkREKnhMoElxbW7v8C0Cbe+yT KAERqKysjIqKolARq7CwsIiICCSirdtoSwPcvn3bKVHO/FfLvwA057XVE5Y/ SmZcU2xiYmKcoURc+oSEhBMnThCo5uZmVJC9M+iUKAuUTxaA9kmUDIpKUVER 9y97x3nLli1bt27l3oT+oepRxWjbBwcHp+aWTrD1OwgUcdMC0EGC/Ct/QP6V e+RfuUH+lXvkX7lE/pUb5F8FD/7vXyEt0tLS9i0SpMWFCxcuu6Ctrc2ZXtgf oLdFt+LmzZtEsre3F7XA55GRkdHR0cnJSXsHh26O117BJsAIwu3bt7l2XOu9 e/dyH7eXBPPy8s6dO0fWenp6COPs3PJnXvsGmwCjIFVUVBAcck2gUlJS+FxY WEigmpqabMIrypXjhRrBJsAePXpEUbl48SJJSpiDQGVkZBQUFBC9hoaGrq6u oaEhruz82hdUAuzx48eUqObmZprlmJgYwkUzkpOTQ9Wrrq4mULRaFijPBVKN oBJg1CYa8/Ly8qysrKSkJNQRJerw4cP2YjtVr7OzEwlETLwCJQEWPMi/8geC TT69DsEmn16NYJNPr0NQyafXIajk0ysj+RQ8yL/yK//KBuqMjo7aourWB7Tx MA7P3THYBNi9e/dsiJpd5cHBQZsHjA7jw4cPbbyHDfmYT7AJsL6+vnPnztXX 11PUyTtFheiZC2qBem65CjYBxiWjCLW3t9tSjATNKVHOCKIX1b6gEmDEoaur q3EOqt7AwICzkISNIHJGW80nqAQYcaCiUY/Ky8tbW1u7u7udQDm177lVTwIs eJB/5Q8Em3x6HYJNPr0awSafXoegkk+vQ1DJp1dG8il4kH/lb/4V/RqaqcnJ Ses7L+BZeRJsAsxWP+zr6+vs7Lx+/Tqpcq95gk2A0cy2tLR0dXXduHGDHrTz 1tvCBJsAu3fvHvWOEkWsaMRGR0fdX8SgEmAUHmoQxam7u5tAUfUInZsS9TTI BNijR49opghUU1NTb28v1ZAm3U2gJMCCB/lX/kCwyafXIdjk06sRbPLpdQgq +fQ6BJV8emUkn4IH+Vd+5V8ZLj0rT4JQgNkIK8/xaS53DDYBRmQsSguMIJpP sAkwK0JWohYVqKfBJ8A85+tb1I5BKMBeoYGSAAse5F/5A0Eon16ZYJNPr0aw yafXIdjk0ysThPLpFZB8Ch7kX/mhf/UKSIC5RwLMDRJg7pEAc4kEmBskwIIH +Vf+gOSTeySf3CD55B7JJ5dIPrlB8il4kH8l/0oCTAJsPhJg7pEAc4kEmBsk wIIH+Vf+gOSTeySf3CD55B7JJ5dIPrlB8il4kH8l/0oCTAJsPhJg7pEAc4kE mBskwIIH+Vf+gOSTeySf3CD55B7JJ5dIPrlB8il4kH8l/0oCTAJsPhJg7pEA c4kEmBskwIIH+Vf+gOSTeySf3CD55B7JJ5dIPrlB8il4kH8l/0oCTAJsPhJg 7pEAc4kEmBskwIIH+Vf+gOSTeySf3CD55B7JJ5dIPrlB8il4kH8l/0oCTAJs PhJg7pEAc4kEmBskwIIH+Vf+ADWuq6urc5EUFBT8zd/8zU9/+tOf/exnP//5 z3/xi1/8ckG++OKLmZkZX+f1dRkeHs7Nzc1yTWZmJq1NcnJy5BzR0dG7FyQn J+f+/fu+zuXrcvXq1fzFs3379r//+7//h3/4h3/8x3985513fvUy2MXXGV0C 6IZ8dzH86Rx//ud//oMf/OBHP/rRz+b4+cugEPo6o68LvbyVK1cuKlZ/8id/ wt/PPvssKiqKyhWzIAkJCTdv3vR1Ll8XWg+ancX27wgOiogmvaioqKSkpPRl VFRUjI+P+zqvwY78K/lX8q/kX81H/pV75F+5RP6VG+RfBQ/yr/yER4vh4RyI ru7u7uvXr9+4cWNwcJDGbWJBpqenfZ3LJeDJkyf3F8PMzMzIyAj94vb2dmJF 0Pjm7ouZnZ31dRaXAArJoqJkUIroj1y9erWnp4fGeeplBIDRB/fu3buzGEZH RwlOV1dXQ0MDFfD27dvj4+MLVz0IjHLFRXcfqLGxsd7e3q45CBTf3HsZjx8/ 9nUWlwCu9QItzHxomScnJ/v6+migiBLfULMevAxaQl9nNNiRfyX/Sv6V/Kv5 yL9yj/wrl8i/coP8q+BB/tVbB92WR48eWaebXiENjllY/BsAw6uWEKJEr3B0 dJRSTQ+6ubm5vb395s2bIyMjdA8fPnzo6wT6BVacCBStcU9PT2trK1EiXARq bGyMe8GTOXydTL+AnhoBofxQojo6OujydHZ29vX13b59m+9VogxKy+PHj+mp ERareu1zWNWj4aK8+TqN/gIlanZ2lqpHA06JunLlCu0VzTjfU5xU7/wc+Vfy r+Rfyb+aj/wr98i/con8KzfIvwoe5F+9XdCjob8zOTlp/Z2SkpKCgoKjR4+e OHHi4sWL9KbV6zEIAlqxoaHh3Llzx48fP3bsGCHib2pqan5+/sDAwPj4uAL1 dO51J4oTgcrKysrNzaU4EZ+8vLwzZ86cPXu2u7ubEhUYo2Jen9HR0a6urrKy spSUlMzMTALFX3oxfEOgxsbGfJ1Av4ACg4SgOSouLrbidOrUKT5QAfnc19dH b1EW1tO5NooSRX+Q6nbo0CGaKQLV3t7ON7dv30asKkp+jvwr+Vfyr+RfzUf+ lXvkX7lE/pUb5F8FD/Kv3hbo7NjDehoxejeoiCtXruTm5lKPcnJy+FBYWFhV VXXjxo1gHl9kUZqZmRkcHLx69eqZM2es40zf0JkIKysrC3VBP5HOI8L18ePH wWlkUUjIvr1ZWVpampqaSogoSCdPnszPzze7r6ampre3l/J27969oC1RlBAC NTw8TKCoYgQn/RmHDx82R7S6urqjo4OqNz09HcwlCvFAiUIFUfVomsyzysvL 40PGHPX19V1dXVb1gtOfcUoUQaANJyAEKiUlheLEh+bmZu7CQ0NDVpZo8IO2 OPk/8q/kX8m/kn81H/lX7pF/5RL5V26QfxU8yL96KzBbhv4O/Zru7u6enp7B wUF6OnFxcYioc+fOFRYWbt26NSIiIjk5mQ6jTYcVhL2ehw8fEqXOzk46g9zc w8LCYmJiKisri4qKEhISEhMTMzMzU1NTt2/fjgYrKSlpaGighxic/WgKSWNj 4/Hjx6Oiomz2bDrRGRkZlJ/y8nJ6PZSonTt3UsaKi4spcsE5X7TZodS13Nxc StSWLVsoPxUVFfzl86FDh86ePctfPvPriRMn2HJ6ejoA+nSL5fHjx5Sorq4u itDevXt3zEHQSktL8/Lyjh07Ru2jmEVHR/OXoFmgfJ1qH0BrQ5tDy0P7QytE W0RjRXyQT/x74cIFYnj16lX+cju+ceOGTTQXhI25/yP/Sv6V/Cv5V/ORf+Ue +VcukX/lBvlXwYP8Kz+HXqG9jzM5OTk0NIR4oF9z/fp1uopNTU3c6xMSEior KwsLCyMjIyMiIugbolfr6+vb29vHxsYQZkHS8aFXSPNOiFpaWggIbRHBoW+I Gi8vLyc+e/bsIVZ8yMrK2rJlC51rPhQVFXV3dw8MDNicyb7OxBvHipOVJRT7 mTNnkpOTw8LC4uLijh49SslBk+fk5BQXF9NEE8CdO3cSqMzMzPPnzwfb+CLK A5m1gXz2ziAB2bRp0+HDh2tqaohVaGiofeYv3/Mr25SWlnZ2dlKiAmZ++4Wh MFjVGx4epkRR9ahllBkaIqresWPHCgoKqGgUJyJ26NChqKgolCRljEAhliiH NnVYkJSoe/fuUaJoc2h5CAuBoi0iGtREwhUdHU0AacEoQtbO2+hHm/9fc7b7 G/7vX1G5KGknFsmpU6euXLnS5YKenp4AGJf7OgtA/+xnP3vp0s/BvAC0QeOP lqDx1wLQC4CGd7n0czAvAG18//vf/9GPfvSTn/zkpUs/B/MC0H/6p39KoGh5 du3aRQ9IC0C/CFoexOrp06fdrP6sBaD9BPlX/gx9Q5QVnZe+vj6Eoj2XB/o1 9PsaGxvDwsJQBdQ46h19Z3pAm+bYunVrbGwstYweJdU54McXESgak5qaGiQl XcLw8PCQkJDt27fTXz569Ki9yrR582bi09HRcf78eTYIDQ0lUHSl4+LiMjMz iS39yoDvHtJ9JlB1dXWHDx/eu3cvMSEOGzduJAJtbW3Hjx8nIESJez0alUKV mJhIJAkUvWw+5+Xltba22vzSvs7KG+fWrVstLS305ri7ERYCxV+qG1GiWmVk ZFjc2IZYWalztuGbpqYmKqmvM/HGsbdQz507R6Dom1BOCALlylxQooQ63TNH eXl5fX09Mom4UaJou5KSkqieVL2xsbEA6AUvDG0LJaq7uxtxRRWjnBAEYkWx MXOYhoi4UXIqKytp5Ht7e69du2YDsdiLsjQ6OioLy6/wf//q6VwNfQUeu2bJ E+wTFrUAtDE9PW0Os5v1Z4N2AWiDewQidmBgYOHVn4N8Aejbt2/b0s+Tc2gB 6AWgLHF/RLS7qXrBuQA07RKaYWRkZGgO9tUC0C+Cdol6Zw3US5d+1gLQfoL8 K//EXhi0hp32h5saXZjOzk4zr/iXK3XhwgX6gHScbcKivLy8I0eO0Ceie0i3 iO+zs7PpQrIjjTyHCoybnScmpWiW+/v729vbCcLhw4eJiQWBCBCTgoICuoRp aWl8SQ+RzWpra+lNR0REWD86KioqMTGxqqqqqalpeHjYpk0OjGbcE5p0buI3 btxobW09deoUPejIyEjHSThx4gRigL98jo6OTkpKInQUMOIZMgdb8n1qaurZ s2cphJQo1HjglSgbTUSJMvOqtLT00KFD4eHhVqJ27dp18OBBihmRoUTxTXp6 en19Pd9QhChITjxTUlKoemZhIUQDcg58mhQCRYmiPNAEEaht27ZZ40MEioqK zCal0lETCc758+fb2trsrV4rVLt377aqR6gJVEDOWO6sxkgz3tzcXF1dfeDA gZ07d1qJohWiLaqpqSEysbGxfJmRkXH69Gmz0502n39p4qws2XC1wCtObyNv hX8l3hDURG6aNIC0WurLLACNFWqB1oxGTA3XAnAHpAWwVXoD/oHOa0KUOjo6 bt++7euE+DUzMzNoD5TDwMBA4Mn1JYR2iZac9pyIqTF/W5B/5YdQlWZnZ2mf 7RGD58h2/uXL3t7exsZGeoibN2+mc00f8NixY/xkjtbGZ9j4on379mVlZdmr OgFWMWmQyRRdP/rOdJO3bNlivUKDjCOZmpqa6DDSSaS/HBMTc/HiRXqLdLft /S8243s+sG90dPSJEyfOnTtHJzEwHtY4cN1pnGtra48ePbp9+3YKCVk2V2rH jh0JCQllZWVsQFis2FCoiouLx8fH+ctnipkTKEoUgWLLlpaWwHsB3Fy+S5cu ESgrUVZIjMOHD1PFiFVeXl58fDzf7N+//9SpU2fPnqVQpaamOlvaGKS9e/fa LmNjYyhSX2duKaGN6uvro+qRwd27dzsWH9ig0Pb29uHh4eTkZGJIyeHLhoYG dqEtio2N5UsrUcC+FMKCgoLz58+Pjo4GmIVl86dRBigntEKeVY+I0QrR5lRX V5upzvcEh4LEjZgeH3+dlt9crJ6eHmJIEaXvrJ6gz5F/FczQd87Pz6+srAyS AcmvhplXqIvCwkJaeNkyC4Cwr6qqam1t7e/vD4Cp0t4oRAmRxl9fJ8R/oerZ YjqXL19GfdGv8XWK/BfaJVubaXBw8L6mG31LkH/lb1Bx6ERPTU0NDAzQVaHb 4jnlwvXr10dGRrjN0eWhE00fmZ5gTExMZmYmt7ympiZ6jvQHPS0svqEXefHi Rbo/qKyA6R6SERrkK1eu0OzQ6YuKirJeobkH9vIgrTdd7Li4OPs1Ojq6oqLi 3LlzaM6TJ0+aLeOwbds2eo50M4kk/eiAacFsiBpxQEAmJSXZW0tOrglddnY2 MaR4HDt2zPFezpw5MzY2RqDYwMaqGfy0devWw4cP81N3dzdlNTAU6ZM5xsfH aQPLysr279+/Y8cOLz/q+PHj1EGKEMLJ3oMjOFlZWXzT0dFB9Jz3MY3IyMh9 +/bZGMgAK1F02WyiOXsVzsm11TKiR0CGhoYOHDhgvhbNVG1tLVKqoKAgPT2d 6unZRoWFhR05coTyScuGbg+YNspKFC05tYkMeubaqhJteNUcZ8+e3bVrF9Hj L9Hr6ekhwrT/9gqh0/5TkCifNiNWkEwa5s/IvwpmaN8yMjJotaiPgfFu4JsA zUl8aMS4UZaWlsroWwBbt7eurs6W6PV1cvwUm8SVHk1iYmJ9fb1GP74IAkWH sbm5mS4PGuPOnTu+TpH/QrtUVFREe47ECiQJGtjIv/I3bAU9W2fQa05RNAA9 mtnZ2cHBwRMnTqSkpFi3kb8HDx6ko8SvmZmZtOqenWhnJiiE1sjIyNTUlK+z uATQvNDI0DgfPXqU7h5dYMe8MicqLS2N5oiYtLe3b9261X6NiIgwEXXr1i1a dZuCxtOjsPney8rKmpqaAqYFs1fhTp48SdeYUGz8OtnZ2R0dHQ0NDdXV1Z4j iHJzc22YX0lJyYEDB7z28hy1FRgPCrnc3MKuXLlCzSJrXiWKuBE9QsQ258+f 3717t70rR4HZs2cP3/A9v3q5x+Z6UT4JPkdmmwAYM4NWHB4e7unp8RxJ5WlG HT9+vLy8nHpHexUXF+d4NadOnaqtraXeUbn27t3ruRehpkTxJSqrra0tYDqD FKrLly/T4BAoipBns2yDY4kG1ZMqRrttLrENcSR0d+/e5S5gY9q97gJUzJs3 b9qoD0l3HyL/Kmih3lENo6OjuTnStuvNuBdBM8W9gB50TEwMtwy9xLQA3BkR 6qiFyspKNQUvgjvj2NjY6dOnuVci8icnJ1Wongt9yfr6+uLi4pw5kG2+TpGf QmN+7969I0eOIOBramroegfAEqjBgPwrf8MW8xoZGeFamIVlLpbNfDUwMECf hc4LdY1etvMGXGJiIv2gvr4+Giu6QmbL2AtfNqNRbm4ufcPx8fHAqJjOoHQb zkGPz15TMs8hKiqK76urq2mx6SE671TShUxJSeHGRxyIKp3KyMhI24ttbObt gwcPXrx4kWgHjH/FvZ4uMDf6ffv2mcFiA8/sPUEbb9bY2EjJ8fSpkJrIzpaW loaGhoyMDM+Xnmy9Qr6sra2l1AWG22AP9QhUWVlZamoqkbHXUS3Xu3btSk5O vnTpEkoJbUmxsUJl775Rs/geqcA2bOk4V/xKoTp8+HBFRQWVNzCmLXLGFFHF bD1Bm8/Kck2WS0pKLl++TMmhUDk+FZFEQdnkaTRrlDRbZGHj3PuGtt6EVT0C FTDvWtKG0HRfuHAhKSnJZvh3mimKEFmmFSKYhYWFlBzzlu2FytbWVr7nLmBT vlv7b68QUkRp/2nZbI4U+Vc+RP5VcGKD5KmMVOr4+Hgb3B4AbfubYHJyEgmB uuBOgVSbmpoKjAHbSw6FioJkq+dwR6BE+TpFfsrExASS1Sa7QIQEjARdcqho dILo+iFo09LSBgcHfZ0iPwX1TqEiShEREaWlpfQZVaLeCuRf+SG27CBNdF9f H1eEDp29OUL7Mzo6yq+9vb3OCBCDfiKVzqb5pR9tS6Hxl15SXFxcdnY234+N jQVYZwcNeePGDTrLdA/37NnjzH+FpOzp6SFK7e3tFRUVzhARYkIDlZGRQdfv 9u3bp0+fthnv6UEjGzgIP9F8zczMBJLEshmkKRtnzpzJyspyZhrnL1m+dOkS vxJDWm8KlVOiDh48WFZWZuNAysvLt2/fbmG0d+IIXUdHB/1re+3O11lcMmzF BG76CQkJMTExdE/Mp7J52qmJ4+Pj5N0ZmmVehDmi1FC2YUtnvFZsbCzRrq2t DZilggyu+KNHjyhR5JcqQ3WzKJHr6OhovqfPghQvKirauXOnU/X2799/9OhR W0siPz+ff83Cou2isKHYq6qqaKMCrBv44MEDtBChQEBSqMisNVOUk1OnTtlU vcSQdskZncUGlZWVtPkUKtoimwXLnCtr1uyJcyDVu7cU+VfBCbWPdoxbJ+2e PSy7cOFCIGmGJQShRdtuQosGECFKs6a2ywt7fIbQovFHyiYnJzc2Nvo6UX7K wMAAVc+EForCxjD7OlH+CM2UzdSKaEeJcXMJMHG1JNAW0UYRHAJFG5WZmVlS UoIQ9XW6xMuRf+Vv2KM9eig35rh586bNgkXfmVpG55r+Dn0ZG/bguA3c8ugm t7a2ss2VK1d2797NBnSf9+3bRz+RKkkjz8UNmPmviBK3+4mJCfJFtzclJQVp RK537dpFTzA1NXVoaIjQcZs7c+aMM8+VjW2gI2nLgdXV1dGFJIw070fnOHTo EFrUlkILGInFXYzihBzKycnJzs7OysqKi4ujoaa5PnbsGKWFnNbX1xNAzxmf +JV7X3t7O2UGcc6/hJfYUqgQV/x0/vx5etMcPGBKFHDboqJx/yIaiYmJBw4c iI6Opvzk5uZS6ahBaCfyvvHr8A2FjV/Zhi1tSjoKIbvbW5ZU4QArUVNTU1Qu skZxIqfUu+3bt5sTRWbv3r1rD/5sjUurevxKPSVQNoaNomijAfmSg5w4cYJu Di3e9PR0wKgsK1Fk+eTJk9SaI3NQiSghlJmamhpaIdoiWiRnNJq9a0mrRYW1 af+JSXd3t605C729vXQAbRhDwJSotxT5V8EJtZLbBE0cFdkmJaioqNDMTs9l eHgY4cGtMDQ0FDlqixer4fLCHnMUFxcjzJBhqCzUaYA9HFwqbJZ7ZAP3Sm6d 3Ea5Rfo6UX4HIopmiqqHgt26dStKzB5GB4y4WioIiK1fT7eINspe4EWY+Tpd 4uXIv/I3uJHdunWLLg8CiWZ5cnJydHSU/gufZ2dnZ2ZmhoaGGhoaPM2rjXPz EdEBtHl42P348eN0HsvLy4uKiqiPSUlJ1E2aMg4VGO8P0ne7c+dOR0dH7Bzp 6el0h8ksXeCjR4+WlpbSR/YcYeUZq/j4eMo5UaIbaF4fne6WlhbuiTTyMTEx 7FVfXx8wj1MpSxcvXszIyKBfjNJubm6mbOzcubOgoIA7GvKSnCK/KVGegaJE 7d+//9KlS0/nBENlZSWxJcjsy450xrdv3843HDwwljV59OiR8w4gGd+7dy8V iuxTlRAAtnAStzmqHiXKy7/iG7bkVw5iC8nl5eW1trbanGMc7fDhw5cvXw6M 9wfJAh0QSg6tCmWA+Nj0TeSRimPThBJJZ4SVEyX6ehQqmjV+pTVDMLAL7dLZ s2cvXLhALd62bRufiVtgDN6m62GzpVGQyJ3Nn0buCAvtDB/GxsZohahctpal A9WQEkUwbc5/pJS9MEiDNjAwQOjMxbJhourg+BD5V8HJxMQE7TxNnLOADs1g II2wXUK4Le7bt88W0EFrnTt3rq2tLTCeeS0haFHujJQie9bDLaCsrCxgFsdZ Wig/6FibmgD9cOzYMbSErxPld1B4KFQHDhxwZmlAZdn8yb5Omn9BW0Tvjx4Q Ot9ZQAfF5et0iZcj/8p/sP4OvRLqDp0Umhr6iXfv3rWxWPR0uJfZoupoJ3uf K/wZVLqMjAx6fxyBVqu8vJz62NTURH/8zJkztPC09unp6R0dHdbLfqt7PfSg p6amOjs76RsiHWltioqKyPLFixdramqKi4v5QKzo8aEHUlJS6Bc7gUJwJiQk cL8jvASTbiBxQ4gSVcJ18OBBGvyjR4/SJQ+Aue7toR63+1OnTh0+fJhYEShy WllZmZaWRm+a7jB9ZLYpLS3lBkd8tm7daoFCcLI9JY1oDw0N0cJz+0NTVVVV 8cEGz1CuzLcJAGeGLgm1jDigiCgzZJD4dHd3U4+oOM3NzTZhL9kndBYrCJuD 2PI9gWUbW9aZfdkYrX7kyJHk5GSKHMehOgdAiaJFIo/k117apc1paGigvpw+ fbquro5yQnt17949qh7lJyIiYuszbKAjTRBBoHIRbZqmnJycxsZGviSqHO3E iRNUZMrb2/6gkJJAiaLRLiwsJGuUKP4SHAoJkUlNTe3p6SEO1B3aMVowz8ac 9ort2YwWjOPcuXPn9u3btOr2weY5tFFY3Cze6ii97ci/Ck7GxsZo92i7Ns3B rZAONZVRtownNkieIk3DbpNdECgUCJpBtowXtshOZmbmxmfz1tqkBAEzFeQS cvnyZe6PNtHo7t27uZki3X2dKP+CqocMQ0chLbwmJaCZ8nXq/Aub5Z7qZtOq 2MSktFpqzP0f+Vf+gy00TCf32rVr9FDo3djNy+Yqt3FT9IkuXLhAv8/GHSU+ Iz4+HpFw7Nix+/fvcxy6hzbUCjgO3fCDBw/SiabLmZ+f/1Yv2GHRIETk2iat ysvLo2dHs0y+uOPTaNu4hZs3b5JlonRgDrYkStz1+EAfubm5mQbKZtBCPPCZ QNGPppG3N6EqKipQFG91I0b/t62tDZm9efPmI0eO1NXVkVkCaCP6bC41Ikat P3XqFO02uT7wDIRBWFiYzQZGOSRQBJlAEWQKGNFDsRNJFGlZWdlbPa6PIHCV yREh4nZPjsg1NcheZLNY2evwV65cIZjoJcrV/jkoUXFxccnJyShPQv30mQ9G rNjXJASKndgSK+dtTV/n+BVxJgpAPe7YsSMjI4NiY746JY3qZk2WzSFGlOiz EB+n9kXNQfPFEcyeYq/+/n4+U6KQoBQkSmBERMT58+ff9lnciYnNWmwmXk1N TU9Pj42YIsuWO0qdOYGUH1pvYuW057Ra/MuWT+fuCzTX7PhoDv7lHmHvEtqr mrKwfIX8q+CERs8WPnYW0KF948u3/fHE0mKD5GninJcFuCNwX6BV1LuWXnBH QHWg0p1RuNnZ2Z2dndxwfZ00P8JeqKyurrZ38G1YEeIKweDrpPkXNkkymtZZ QMcmwbDh8b5OnX9BW2SvKZnHTmO+detWWi0C9Vb3/oIB+Vf+gD2ooidCx4Ru IC0PH2ZnZ+0ubzNimeOEQKJRojdEW0T/kb6zTW+VlZWFoKIDaDNc0XGm9/1w DptrND8/nw327dvH32vXrtGtpol767rSJJgo0ROsq6uji0feT548yc3LBlrz l19pdkxG2vgrC44t14jI5F9CcfHiRbQBgWKvsbGxiYkJjkygCP7ly5c5LFsS 4crKyrd0xUbyZfOknT17lnKya9cue1vQ5BD9aIJjb2mZ30IM2QxtiYLiL+Gy QSMEio1tkiviwMYEmVhxCSoqKthl586dXAJae0oU27yNXWlKC4nnWicmJpLl w4cP19fXExyrfXxwFrgxkXnq1CnKFfc7m9Dp2LFjlKgzZ85QrZ7OLe7suHnE ygxnNkuegzhzCd7GEmVmL7WGqmf2Jlnms9Ud/lKPnGpI+3P69GmK0OE5iJUF 1ibis0Hszl42eI/YNjY2stmePXsKCwtra2tHRkZsM19nfdFQWrjKFBVaHrPv rly5QnYoUTYjFrXPGmpq6KVLlyhROTk5FD/a8/T0dGvMKVe24LU1U1a5zGul gA0MDPTOQXMnC8tXyL8KQuzRGC0V7b8zZ118fDyV0R6c+TqB/oLnLPcWpfDw cMQVt8u3tGF/c3BjzcvLQ6N6TkrADVFrxnli6qK8vJyCZJNdbN68eevWrWjR gJmIdUkgFB0dHTaXxfxJCXydOj/COte0SFQ9p5niQ0NDA52Ct3eYR5Ag/8of sAdV9FZsjmgbTfSiLen+sAGVjv5yREQEXR7qGrc/m9DJujleEz+yFz+1t7fb qC36lUVFRfR63q4G35oaVCLpt74eXTwbJ+O1mWFTYNmcRSUlJbt27aIBp2dN xh3Pytne9iV0fH/58uXi4mJui5ziLV0a26ZTOHnyJDf61NTUs2fP8q+nv+Tk 2gZWTU1NERazJuhK0/WmtDjfO09Lnb3MH6uvr0fG22xR586dI7Zv18v1lh1u 9CdOnCDj1Cbu714zenkWD2ol5Q1JSaFCbdJ5sSXkqLmOx+W5vX22EW7cIilR ycnJnMKcrrcI85q4oVODoqOjDx06RCEZGBgwQ+bp13NNvaNs2MR9VVVVBOro 0aOZmZnUXELEcShX9v6I517mj124cIFAxcTE7Ny5015NfbtK1NO5jLS1tVEd qBTbtm1DWlOVyLhXqXj6LKq0RfaSuL3pjDhvbW0l42bfPbflsbnaiL8NwWJj Aq7+4PIj/yrYsEeN3Ex37NjhrGsMNFlvqVR4cyAempqakFKeC+jQN0SQ0B6+ XeLzTcPNDh1lbzAZ/Iu6uHr1qq+T5kdQolBf8xfQQTMEwOQMSwhtVE1NDfrf WZLJFtBBuckR9YRWiLaIFol2yXkYQcNOh7G5uTkwJmINYORf+RxqEL0PusDW H6H7TK15Ua/NXnOz8RtsvH379n379lHREFR0hexB/HwFxTf8dOPGDXqdaWlp 3Bn5QMfn7XpcaO/6NTY2kv7ExESyQN/QGSfjBfmyponAEp+ysjJ6xBn/H3tv /pTYl6V7/+tvVZlf2zDTVwUxBIohmG43EAJehwaNkkYJJ1oMFb2CdJcDEAzq 6w8amdFR+X4iVxThzcHcMhwOsp8fMjJJPGfvtdda+1mPe/hf/4uHvLThj2As ZH0X2Z4Xra2tUVMPkTIjtS2+kc1m/+3f/o1ex+NxLFCr1X76fTlyjR/BSSBR ss3tb3/7W6lU+vwNPz1EVMwLufr3f//3YDA4OzsL14K9N5vNIVrXR8dpMBO9 LBDyeDw7OzsiDv/0+3Lvnhy7DWVipqP7/98/8aufkrdsbW3htLwoFApR6bzy FrNB9qwRFBQjWImxpu+ZTAY7/HSgZb8boOM4oaz8xBXJb7KK71c3DMovDRkO nJC3QCEI9vv7+7ZKZn7I+k/GGupIL+x2ez6ff0WFk/3dot3hG//yL/8ih1Tg MHxOx39lKEJSznW/+gZJUMNipXcDrV+NGuT4grOzM6jXy+tOZmZm2ocSDLqN ZgGZkPy/tLTU1q9kHQjz7FAfONBbkLRxKlkqg1O1DQX/DIfDciiBhkCuh5ZT 7l+CT9qHPGh8/cbBSEfwrrZH/elPf6IWgJDIESKDbqBZINslyEgvkzn5Sq6H htUPuoEar0HrV4MFbIcYoRIplUpUiGRgyhbFMuSlfvX3v//9t0mJnCZiDrWh 1WqNRCLUSpSZQ1H1YCjqX6pgUg1cMRQKVatVCNJvGy87Ll/qV1//74UfPwJL MhBSek9MTJDzYRHDkvbpL4ZaW1vDN9xuN104Pz9X3Nn3n//5n239Cm/87ffl fHgMtbCwIKLi7u7u8/PzsJzOCudJpVJer3d8fDyRSECNfqXJvITsKHypX72+ +kWcjRhnICBaHz588Pv9mUyGmO11h/oC0d9IFzBq3CMQCNByuvxbj5JfArb1 K4z223fxI7wO6gV1t9lspKmTkxPePhTnpWAQPCqXy83Pz//Lv/xLMpkkoOQk tN/+7Hf6lcohq6Ru2JecpsWf5EO93N1gaP1q1CArwHd2dr67AHpycpIUR+Ia ikxlDJjy4vE406ucEtYG6fHi4oJ8NRSEqt+QpewbGxtYqb0C5P/5dlaYxWKB UA26gSbC1dUVbNPn832nX/HJ9vY2/zvoBpoFMAFoPP7zco0oDoZTwS70gQMC aLlcpU0R/dKdMBRZi9xFBht0GzVeg9avBghyCHSI6qxWq/3Xf/0XNYjsblP8 8bfqV3IuFsVOKBSiVvJ4PNTsiirQYEE2xkRUsm63mwo6GAymUimarXLC80/1 q9+C+vHs7CyTyVBBO53Ov/3tb0dHR7JOqeve9As4AFzo9vaWqpnJi/5GIhH4 j5wCrYK36le8Ebp+enpKqodCzM3NMS5yJs/rCuHAIWeM53I5ig5iwW63/8d/ /Ee9XlfRDX7Ur1ScitBjaPBD2Du2isVieJTJb1IQj8J/iAW/308s4BsMMYZS WWbQgX4l23MIWDyQuOONjAsOhpFNvrBBTv3a29tLJpO0nFgg9C4vLxXzeQf6 layLw6pyWS1/ymJUM8fdO4PWr0YNBCbhSbE8MzMD+/rjjz+I2Y8fP87OzpIh 19fXtX7VBqkJBkIlODU1ha3kduPJyUmXy5XP5yEYuoj++s8DHDY2NrAMVpqY mOBP/g5/s1gsTKDv4HLnXoHiBQYF///06ROGIvqwFX8PBAIYkP8ddANNAVnl LhdAE3riV4Lp6WnY1P39/bD8irmvaJ9yT0Yih8sd2VgJo1Edh8NhFcqqMUBo /WpQkIXocrsZZm82m+0TmRTxVv3q67fykMqdYmd/f5/5kfxGdUmdZeZsRpup l2lnKBQisVDlvemg1M70q6/fZoFGowEjhYOR3Nxud6lUIqGZtjyUpbAU/nLH opyVLcdEKz7hrfqVAI9iQkyn09AtKndo/OHhoclJl/wiDyL04cOHaDRaKBTw McWR7Uy/km0ClUoF4/zrv/7r2NgYf2JwMx9H8PnzZ3qXSCTkhlNmeZKV7HdT +fEO9Kv2e3kLtiXY5+fn7Xb7xcWFmSUsRlZUPgxFTsZiW1tbd3d36iHQgX7V /kEcEn9maPAuORm+035ovA1avxo1yJ446EQ8HienkZrIisyYchuvPpn8JZja IAOktbZ9qAphU6RHCEY+nzdtPjcScDZmt83NTdgI5vL5fNDU9kEf29vbkKuh voq3h2CGjcVihB4uFAgEqH34k3/iYPgVLG7QDTQF4E7QAExEaqJoIk3xp3gX f+7t7b2JYLxjyLXjVC7kbVwIc5HPMRd/J1ORuCBUOpmbGVq/Ggioa6jR5CCd crlM9fHKme2/Qgf61dd/ngBPBiM83W43CY06sdFoqO9bNBK0lvmd1jJPeb9h eXlZWqv4hI71K/nZw8PD9fV1i8VCcoO4MkWacH+cnEYlS9SYtqanp+kpjOi/ /uu/3rQOqjP9ShbAHx0dwVFhX0wBvJrM//IUdPNAtJHj4+OlpSUmdJvNtrGx IUsfFZ/QmX4lkAMc1tbWZmdnJfSgZAS+2TQHUdvwqFKphC9hpX/913+lywyr +o7jjvUrSY/7+/sYyul08vadnR1a8quDswYLOelLiBDh014z9qbDPzvWr/Ac fpaRkiVYt7e3T09PWN6Eyfz9QetXowaCmrmDiRIyQPaen58n5PlLMpkk0fGh XgDZBswNy6ysrPBnPB6HGJDi4AZYCVpVLBa1fvX1myIKe8cg8JBYLAaDgkdB Tv73N+BsQ3T4ar9BevyPfwIrQcv5M5PJMPO+ibK+b0AA7u/vt7a28B88Cj4P AYPFEYn4GCWMnP8w6GYOHpBJWGU+n4de4lEejweqSQGVSqUoMyGfzWZTJ3Mz Q+tXxoOoIXtQ9lLJknIpPeQ39W+NlM70q6/filPqIyJ3e3t7cnJybm6O5G/O o0exjKyAop0QRTk2gXaq26ob/Up2MzFA8usenuD3+ykSVU5JMhJU0JAc5nE5 c97r9R4cHHSgs3WmXwnwqHq9zmBNTU05nU54xcnJiQnPcmc0j46O/u3f/m1s bAx/wFCU/G9qZzf6lehCRCuhhy/RBqx9dnZmto32sgSd3kGn5dJSGvnWZecd 61cCWddH6OHS1D6wC3zSbEc3yOpQMjnN+/jxI5lKfX/lS3SsXwkwC36IX5HV ZUIxm8b+LqH1q1GDUAIBUQ8tgYBRAfF38pWIV2ab8gYFuQOaXMSECwGTC6Dl Wgo+N9uvbAYF+eUj7oRBYCOynA9zFYtFrMTnioeXjgLkciJYQaVSgVf86U9/ Wl1dbbVaIvGZsH4ZCCQFycVM8Mz2BdAQXT6U3KVzlEBCD8Dq5QJouTD64eFB 7rTShjIztH5lMIT/UGs0m03KMSqON53Z/hId61dfX5wo9de//lVmzM3NTVpi nmWlJBZyyOXlJSVhIBCgnYuLi0xbbz2tqxv9SiBbBlZWViwWC1UqHEOohRmm S9FD8CWaxFQ+OzuLrf793/9dfrn5VtrTjX4l65owjt/vZyJwu91MnRTy5rmC lrmbMv///J//k0qlgsHgzMwMY0q9/9Zra7rRrwSYpVAowCvwSSg9oUczzLOu D6eCE15dXRF6drudAY1Go9fX12/VjrrUr+Ri1q2trVgs5nA4SHS5XI5ERzNM Qipk5eH5+fnu7i5ZVO5MJIg6WHnYpX4lZ7lDUJkUSJLkdr0KywBo/WoE0Vao fkrAtH71HX5KwH56TfbIou0zPxIwIXjaowRiqJ9eoKOt1EbbnX5KwHTovUTb Vj8lYDqZmxxavzIS8gspqpv//gbKjWaz2XHi7Ua/+votj9GS4+NjqMXExITN ZqMKI3LNIMt8/adq1L4BkHbSxzetvBJ0r1/JqN3c3DALuN1ui8XCXFCv19X3 MPYPUkHv7e05nU6stLCwwITF/N7ZqeDd6FcCuMT19fXq6qrcfri8vHxxcdHB c/oBYo3GxGKxDx8++Hw+5ikiqIPo616/4qV4cqlUktsPmTR5Go3B59/0nH5A LpVIp9PhcNhqtc7OzhI+nZ2q1KV+JZDDHGjM3Nycy+UKBoNEohk26dAAxosc HggEpqen/X5/PB6vVqudta1L/UqAZ97f3+NIuBZGo3kmSebvFVq/GmV0ScBG BN0TsNFB9wRsFNA9ARsR9ISAjQh6QsA0DIbWrwyDrLx6fHxstVrY+erqqoMz 21+ie/okl0HncrmFhQXmzZWVFdKd/Dqjsyb1BHJPIvaRxEvbqF7pbKPR6MBW PaFPssYJUkGJSkVPojs5OcF0gz2lHGtQrtKMVCpFde/z+WheoVB405ntL9E9 fcJz7u7ustmsy+XCSnJc5MAPIGWYGPdyuZzJZPAoPAFDMYLU+B08rVf0qVKp YPDFxUXag4fv7OwwlINd1ycedXt7m0gkPB6P1+sNhULYjXjsoFU9oU+yrg+H 5AlOpxP/JJzlGK4OntYryN1/WIYRxFAWi2V5eVnO2u0sn/eEPslZ7rIKq1qt kjCJOy1h9Q9avxplaP1KBVq/UofWr1Sg9StFaP1KHVq/GkZo/coYiHhF4oXn iHglZ6R0w3l6Qp/4qYeHh/39/dXVVYLXDGQMwnN0dLS+vk57bDYbbSsWi1SL nTWpV/SJt5PTIBVkObvdPj8/v7i4OEBlhvbgUfl8Xu6poYJOp9Ndtqcn9AlX l/2M8Xh8fHyc5slJ8h23qnvg4UTcysrKxMSEx+NZW1u7uLjoYC2foFf0SUbw 9PQUR8Lsk5OTPK1Wqw1wxyXt2dnZWVpawp2mpqYIPeyGR3VmqF7Rp/ZKJ9n1 KRdaYagB5ijGqNFoyKowPIrEgqG6uXqsV/SJBsivSIi4UqlEJHamPWqoQOtX owytX6lA61fq0PqVCrR+pQitX6lD61fDCK1fGQO4DVn37u4OC0N1yCT39/dd btnuFX16enqinM9ms1ar1el0bmxs5HK5jvegdQPRiCqVCm2gqKc9gUCAtvFJ x73rFX0SBZKBW11dlbqValra1rES0g3oF6/e3NyUa32YnvL5fJdnKPWKPuE8 tVqNti0sLPh8vmg0StsGco2OrJO5urra29uLRCJTU1M05uDgANN1/Mze0ifa tr29zfDhn/g8hqJtxq/rw4HxqEajsba2RtC5XC6v18s83s1J4D2kT3KlDtb2 +/1utxuHLxQKne1q7BKiOl5fX9MAGkOOisVipBfZCd7xY3tIn+Qsd1ool4O0 Wi05ibTjB2r8Clq/GmVo/UoFWr9Sh9avVKD1K0Vo/UodWr8aRmj9ygBQRBMO FKqy+KparcqtIl0+tof0icZQf21ublJBT05OUr3KTX8GyzKipGUyGfo1NzeX TqePj487Xnkl6C19klPlyW9yoDSl9MrKCvOpkXW0HOl5enoaiURgO1NTU+vr 6z05Kb2H9IlGUjszfDRybGwsFArt7+/f3t522cK3ApvIrYifPn2ia9Fo9OTk pEt1qLf0Sa7UOTw8DAaDTqdzZmaGMDR+XR/NkFuiiDsywPb2dqlUImt1kwF6 S59oSaPRkAWQ+DzPXF1d7WxPcTdgrMkAch283PVZLpcxVJfVa2/pk6xYw/Mx F83DnbpvocaP0PrVKEPrVyrQ+pU6tH6lAq1fKULrV+rQ+tUwQutX/YasKZJz iq6urqrVaqvV6smVuL2lT0IzqJ0tFovL5aKApdJ/enoyRpmh8RTsFFxbW1vL y8u0gcKQ9sgJ2908ubf0icKQJ9ze3lLABoNBjM+8QIVIyw2ro3EnXscAeTwe Uq7f79/f36eb3a+X6y19ooKWs7lkHd3Kysrp6alh6/pknUylUmH0GSxGn0k8 k8nQpC6f3Fv6JGpkqVRaX1/Ho6anp3ny2dkZrzBsXR8eRVIi9hn6v34DsV+r 1bocqZ7TJznLnQeK5xPOxWKxmxVib4KMFAmcl4ZCIZvNFg6H8Wr61f1Nfz2n T9gKy8gqLIzGBGSGQ+/fGbR+NcrQ+pUKtH6lDq1fqUDrV4rQ+pU6tH41jND6 VV8BpXl6eqKIkJVXcuxwr04j6Tl9omGkOCoyaMbU1BThzGMpbA2oesgeV1dX +/v7vNdqta6urvJ3PuzeVv2gT1SCzWaTGn/uGwKBwMbGhmFHJV9cXCwvL1PC M/oMFt4ltwB0P0y9pU9S79/c3Ozt7VHvj42NRSIRjGbMNMpwPDw8pNPp2dlZ h8Ph8/lyuVxPPKof9AmPIlFgKJfLZbfbbTbb9vZ2TzTJ3wKDnJ+f7+zs8NKJ iQki5fj4+PHxsft80nP6hEfR2tvbWxrsdDpp7eLiohz71uWTVUB3YDVra2te r3d+fp5hguf06o6/ntMniT6yd7VaZaYgnMVRtYTVQ2j9apSh9SsVaP1KHVq/ UoHWrxSh9St1aP1qGKH1q/5B1hRRCf79GyqVSrPZ7MnOQUE/6BNzAUU0k4LF YnG73fv7+xcXFz2p+n8F6imKdOqsg4ODZDIp793d3SWH9OSl/aBPcpoZ9mdS 8Pl8UA7q6Ovr635rfbgTpTqDEgwGSbYulyubzd7d3T09PfXk+f2gT9hEBLfp 6elQKMRA8M8uz+l6HVK51+v18/Nz5m7eGwgEcK1SqdTliXOCftAnGoa34/MY Co+anZ1dW1u7vLw0wKNoPySH9zqdzoWFhVwuRzLpSZj0gz61b0iMRqN2ux2P isViIs70VT2W3YuEfDgcxkry3pubm14pQv2gT+2z3K+urmi5XHfb/VIxjTa0 fjXK0PqVCrR+pQ6tX6lA61eK0PqVOrR+NYzQ+lWfIBvNKpWK7OCg4KJU7O1e vH7QJ5pNI2k25SHhTB0N5SDvUQf1qerhdTw/n8/PzMxYrVbeSwJ5enrqVTXa P/r0+fNnDAXloOVzc3Owjq2trb5u0mHENzc3fT7f+Ph4PB4vFouib/Tqjf2g T7Jm5vLykmdKy5eWlsrl8t3dXU+e/yMYAmafbDZLaNjtdqfTubu7yye9ir7+ 0Seew9PwIlne43K5erUK8acgY+BRR0dHGIqJWxY9Pjw89KoQ6xN9Eo/CP0ka L1tOjurJ838EBmEUyCGybZBgPzk5aTabPcznfaJPouU2Go2bm5v//gbyod5I 2Cto/WqUofUrFWj9Sh1av1KB1q8UofUrdWj9ahih9at+QMQr6qnrb6B2qNfr PVx5JegTfaLxd3d3mUyGvGexWAhqOYeKqqe37Zcb/Zh9eD7ztdVqZe6WM9t7 +KL+0Sc5y71UKgUCATmQZ3l5uVar8WHPy8PPnz/jToeHh9Fo1OfzLSwsYCje Re96+Jb+0SemTjmHanp6OhwOZ7NZnt/DvbQCkfLw3nK5DLHBo/x+/9LS0tnZ WQ9v9OsffZKz3PF/hgB3stls6+vrV1dXffIonCeXy21sbDgcDmzFoFxcXPRq Ld/XftInrEFGZZSJ6Pn5+Ugksra2RpplaHpeRYpH8fBYLOZyuYg++kI+7O27 +kef5Cz3RqPBHEHQ0RfGXa/C6gm0fjXK0PqVCrR+pQ6tX6lA61eK0PqVOrR+ NYzQ+lU/IGuKbm9vqbCur6+JhX6cY9M/+iQ7H3lmIBBgPp2bm6NCbLVaPaxt v37LrhRW5+fntm/gXalUqud3+fWVPlEDYhMGYmdn5+PHj+2Tu3q+OQ5f4rG0 f2xsLBwOHx8f42A9XHkl6B99kjUzjHX75C4MRad6u2ZG9J+DgwNe4XA48Fsm I3yst4Peb/qERxFreOz09DS9IC7y+XwP9bev3wK8Wq1eXFww3IwFI765udls NnsrJ/aVPtEFEgUexSsY6ImJCZz25OSk+1s4X0LWo2az2VgsRnQzIrlcDr/t eT7vK32S3anYXzYSkth5vpawuofWr0YZWr9Sgdav1KH1KxVo/UoRWr9Sh9av hhFav+otpN7B+alxrq+vqRfkGq9+nPbTV/pEvVOv11OpFJOp3B9HqXhzc9Or O9FoLcSGREFtyPOp05PJ5MHBAbbqbUf6TZ8Y7kajUSgU3G63y+XiFRiNT3q1 MornM1+fnp6urKwwCtgKQzHcDw8PPXn+S/SbPuE/u7u7kUiEseAtuVyOT3qo zMBkCLqNjY35+Xmv18uLGPqe77/rN32Ss9wPDw9DoZDH47Hb7el0ulqt9tCj eP7JyUkmk2G4LRYLw42heruW72v/6ZPcDnB2dubz+ehFPB7f3Nwk39K7Xm2m ZnDlSgtewVjwZ7FYJPR6ns/7Sp9kF+H9/b1IWGQP/s4rdMXdJbR+NcrQ+pUK tH6lDq1fqUDrV4rQ+pU6tH41jND6VW8hN9NRbJZKJSoF5u7n5+c+vavf9Imq n/ZfXl663W6KaCZWMuHj42P3aw/k/HNM5HQ6aT9PXlpa6vk6GYEB9InykO6c n5+T+hiRubm5Hu6CrNfrp6en0Wh0bGwsFArt7e1dX1/35BzyH9Fv+iRrZhiO cDjMixgRxqUn6/pkhQlPxkQOh4Mnb2xsNBqNfkxDxtAn/LZWqyWTyY8fP8qg FAqF7gedJ5CgSB1M1lNTU0TEysoK3enHEVsG0CfxqHw+z3ATd5OTk9lstlwu 9yTr0n4iIpVKWa1WnkxfCAp8tR+hZwB9YoixlVxHCJieeiX0jSy0fjXK0PqV CrR+pQ6tX6lA61eK0PqVOrR+NYzQ+lWvIKc5URFQINze3sJn+AtB0dvdcC/R b/okNwPSC0mAvIg/S6USZXU3+onsuWN23tvbs9lsDocjkUhQdfbpnGpj6BOG YtCpo+kODCQWi5EG7+/vu1FmcCfK8GKxCJmh5dPT08vLyxcXF81ms4ctfwkD 6BOjz5PT6TQ9YkQY+pOTEzyKoe/GgUXtweZOp9Pr9QYCgYODg57vRRUYQ5/k LPfd3V264/kGYoSh78ajMDKPZXYm9Nxut8Viof38nRf1sOVtGEOfcJtyuYzr YiJ6tLq6urOz02g0urwZkFHG2hsbG+Fw2PUNhUKBx/YpnxtAn0TjbbValUpF rsR9eHjQElY30PrVKEPrVyrQ+pU6tH6lAq1fKULrV+rQ+tUwQutXvYKc5lSv 14vFImbk7/0OAWPoE/26v78/OzuTU6r8fn8ymez4fkA52Z7W8hyn0zk/P89k TVXY21NrXsIw+oT9mUnz+XwqlSIN0rX9/X2coeNxwewU5olEYnx83OfzQWn4 Z5c6z+swhj6JLndwcOD1euWUqq2trcfHx86GRg4BOz8/j8fjLpdrcnJybW2t 5ydrvYQx9En6hQ8Q5rwLj/J4PHStVCp1/EBsQstDodDMzAyTdTQarVQq/buN zjD6JLrc7u4u77JarUS6bE3teFx44OnpKW4pJ2ul02kSIOPevwOjDKNPckYc 4y5LsLASwagPwuoMWr8aZWj9SgVav1KH1q9UoPUrRWj9Sh1avxpGaP2qJ6Ao oAqo1+tSF1A+k2P7nVSNoU889unp6fr6mmpXRCemVwqfVqv11qqH72MlHnV4 eOhwOHgUz6Q27NM6GYGR9Ineyboyyl46uLa2Rk3N9PrW3UxSYzK+2WyWeYeW x+NxelGr1frUcoEx9EnOcr+4uIB7BIPB2dnZlZWVYrHYbDY7WNfHRMMPMk0z +8hSpZ2dnT7tRRUYSZ+kd1tbWwQLHQyFQvv7+/f39x14FCGGR+FFWAmb0/5M JnN3d9c/7cJI+iS3A/A6HJjera+vk2QwVAfqHOkO/8HmiUSCp9nt9nw+340a pgIj9SvGpdFoMEldfYNIx1rC6gBavxplaP1KBVq/UofWr1Sg9StFaP1KHVq/ GkZo/ap7UBGQQqkISqVS+4JyA8oBI+kTZWC9XudFFotFriMkzD9//qzeTdlf WalUSKQej2dmZoZivFarYbq+2spg+sQo4ADZbJZZY2JiguL35OSEIvFNo0NF SXW5trbG+Lrdbh5Fbd7XlVcCI+mTnEvP9IqJRMzc3d19fn5+68nYl5eXqVTK 6/WOj48nEolyudxXTearsfRJVmGRW3CAxcXFDx8++P3+TCZDvL/pOXKnIWFL 8DLEgUCAJ2DtvnqUwfSJ18E6Njc3w+GwzWazWq2EXrPZfJM2jkHwqFwuNz8/ D5lJJpMExcPDQ79Dz2D6ROrGJUgyhPnNzU21Wu3HDbnvHlq/GmVo/UoFWr9S h9avVKD1K0Vo/UodWr8aRmj9qktAWvB8qqRarSa1wOPjY/fnUavASPr0P//z P/SrXC6HQiHC3OPxJBIJqh4qO0WtgPoIE1FRut1u5uhgMJhKpQyYeoynT6S+ s7OzTCZDBe10Onl1Pp/nQ5UKkUFk0sGLqKChMbQ5Eons7u7ySb+b/dVY+iQn bxcKBRzJ5/PNzc3hD3Imj4g2v33C8/MzX8ZQTM34pN1uZ+qp1+v9nnqMp0+i Z+JI8/Pz2CoWix0dHSnepCC6Mf5zfn4u91cyvpgaQ/XjuLmXMJg+SU8JdryX uKOn+MPp6SkDpNhTxhGr7u3tJZNJnoBPEnqXl5cG5HOD6ZOsx2NEcIzrb5BF sHoV1pug9atRhtavVKD1K3Vo/UoFWr9ShNav1KH1q2GE1q+6gdyj12q1qHEw XbPZ7PdqopcwmD7J7r+rq6v9/X2oCOUhiZH6TmXBDD9Lvby0tBQMBqempqjB O9uB2AEGQp8oDxuNxvr6eiQSmZycdLlcxWKxVqv9tr+yTkZaC5Nh0qHx1JX9 lhoExtMnPOr+/j6dTlssFofDgWMcHh4q9hcX2t7eDgQCHz58iEajhUIBHzPA o4ynT3SK6K5UKhgHKvKXv/wFT2awqtXqb3+W8KSFhB4ju7CwgCuSrAgKAzxq IPTp8+fP9I5xgY3Mz8/T5fPz84eHh9/2Fwvf3NycnZ05nU7yaiKR2Nrauru7 M6YsHQh94qU4M3GES+BdhE//9nG/S2j9apSh9SsVaP1KHVq/UoHWrxSh9St1 aP1qGKH1q44BXaFWosAhLZTLZaoAA85sfwnj6RP5kP4S3US62+0OBoNMH41G 4/X9kvxUq9Xip/x+v8fj8Xq9y8vLzWazf2e2v8Sg6BPvPTw8XF9ft1gsLpdr e3v79PT09f1xlNiyRC0ej0trNzc3oTGK65G6h/H0Sc5yPzo6CofDONXCwgKv /u3p66JRMKyJRAKnstlsGxsbsvTRgDYPij4ReuSZtbW12dlZCb2zszMSzq80 B3EbklKpVILAzM/P8yfNrlQqb9r52zEGQp8kLe/v72MouR2ANhSLRUbtlQzJ j2BJPCqdThMC7bVbxuSorwOiT3gO7yXnMHldX18Td09PT4yaXoWlCK1fjTK0 fqUCrV+pQ+tXKtD6lSK0fqUOrV8NI7R+1RngKtTdZE4qSiYaSgD5Tb2RzH8g 9IkOEtpUxNvb25OTk5SHTLU04JXlDVimvRKJ2fni4qJarfJ9Y2w1KPoku5lw jEQiEQgEeLvf76dI/NXpTLJDM5PJyNnvXq/34OCgg/OgusGg6BMeVa/XcZKp qSmn0xkOh09OTl6JJqyaz+eZkcfGxhhTDEXpbVj0DYo+ySosIp3Qw5foOyN1 dnbWaDR++n3Zi4qhROIDfPn+/t4wjxogfZJ1fYQeoWS32z0eD/789PT00yQp q0PJ5Hzt48ePuF8ymTRgf+VLDJA+YRZ8GL8iq8tEZmTOGWpo/WqUofUrFWj9 Sh1av1KB1q8UofUrdWj9ahih9asOINIEabPZbDLLQF0MO7P9JQZFn9onWTHV ulwuZtvNzU0s8GPIUwA+PDxcXl5GIpFAIMD3FxcXK5WK+qlZ3WOw9ImeZrPZ lZUVi8VCdXxwcFAsFjHgd6WxbETlv1ZXV2dnZ7EV8zL/NODM9pcYFH2S9VQY x+/3e71et9u9vb1dr9d/XP1C3FFun5+fJ5PJYDA4MzODbam7cT/DWjtY+oRZ CoXC0tIS/hwKhQg9uv9TnROTVqtVvuBwODBsNBq9vr7+lYbTDwyQPsmFsFtb W7FYjO6TJHO5HEmS7n+XeWQFIB61u7tLNiP6UqkUXzZmLV8bA6RP7bPcmVBI zuR2vQpLEVq/GmVo/UoFWr9Sh9avVKD1K0Vo/UodWr8aRmj9qgNA+Klu/vsb oP3NZtPglVeCAdIn3oUFjo+PoSUTExM2m43qj6j/TpYR9aZ9Ex/fp52GrbwS DJY+0VO85ebmhhnE7XZbLBbmkXq9Tqva35G1fFjG6XRipYWFBSYd5mjjbwQb LH2Ch1xfX6+ursqti8vLyxcXF999p9Vq8aHcxOfz+ZhriAKDo2+w9InOEkGl UkluXWTCpSUYgVhrf0d2DhYKhWg0ilPNzs4SAsafbjRw+kSU0etwODw3N+dy uYLBIJH48pRy2V9JDg8EAtPT036/Px6PV6tV408yHzh9wqvv7+9xJFwLo6mc GKah9atRhtavVKD1K3Vo/UoFWr9SxMAJ2BBh4ARMowNo/epNkC08j4+P1NHY 6urqyuAz219isPSJeOeluVxuYWGBOXdlZYVUiSmen5+//nM3HPaR5Ml3qCJp cKPRMNhWA6dPjAtOAiGhNJ6dnSVJnpycYDomFxFe5FR8qAtVts/n42uFQsGw M9tfYrD0CTvc3d1ls1mXy4WVgsHg3t4eNXXboxi7y8vLTCaDRzGaGApLUmsb 3E4z0KdKpcJgLS4uYgcia2dnB49qr+tj8iXQtre33W63x+MJhULlcpl4NNij Bk6fZF0fzszbnU4nvk0qkOO/vv7zDj4sgyWxksViWV5exmi4nPH5fOD0Sc5y l1VY1WoV/yHutIT1OrR+NcrQ+pUKBk7Ahghav1KBGQjYUGDgBGyIMHACptEB tH71JshmE7iKiFdyRsqgeMvA6RNvfHh42N/fX11dJfBpydHRkayHgbTw9/X1 dT632Wx8p1gsUi0a30gz0Cd6TT6EkJAh7Xb7/Pz84uKi7LjEoyio26uz0ul0 W7ExHgOnT//4xz+azSauEo/Hx8fH/X6/nGD/9dvqLCJueXl5YmLC6/Wura1d XFwYvJZPYAb6JGdbnZ6exmIxl8s1OTlJS2q1muy4xDIrKysejwdbYTHshkcZ bygz0Kf2CivZbUoA4t4YSnbsNhoNWZ1F9IVCIQzFUA7klxFmoE90XH41Q8SV SiUi0XjNc7ig9atRxsAJ2FDADARsWDBwAjYUMAMBGwqYgYANC8xAwDTeCq1f qUOOLidbYiXoCtng/v5+IDsHBWagT09PTxTL2WzWarU6HI6NjY39/X0KHyrB 9fX1xcVFPg8EAnynUqkMpIVmoE9yYBoOs7q6KvUy1TRJEopCZX18fEz57HK5 +K+joyODz2x/CTPQJ8hJrVbb3NxcWFjw+XzRaJRW3d3dUVDv7e1FIhHGkRYe HBzgUYNqoUnoE1FG6OE22IRYy+fzjFqr1drd3cXBmI4Zze3t7UGdyG0S+kSO IlEzUn6/n0DDLKenp9VqtVwuFwoFPiRHYT1yl+wEH0gjTUKf5Cz36+truZQE X/r8+bMuzH8FrV+NMsxAwMwPMxCwYYEZCJj5YR4CZnKYhIANBUxCwDTeBK1f KeJ//ud/5CAjWXxF+UNyGOzvpk1CnzACdV86naYGpD02m43Jl5xJKqAw5PPj 4+OBrLwSmIc+YSim2ouLC4fDMT8/L6U0M4vP55ueng4EAj89sdxImIQ+/eMf /2CqPTk5iUQiY2NjXq+X4cM+nz59onnRaJT/kt2XA2meeeiT7IDL5/PBYNDp dM7MzMjJ/7jWH3/8EYvFcDZic1ACu3noU/uGQSJuamoKc+HkHo/HbrfLXZ/8 F4xlgBWoeeiTrFjDXGQAzHJ/fz9Yy5gZWr8aZZiEgJkc5iFg5odJCJjJYR4C ZnKYh4CZH+YhYBrq0PqVCqD0T09PcqATRKVarbZarQHuHBSYhz4xodAG5lwq 6Lm5uWg0GgqFoCsul6u9o3BQbTMPfZKz3CuVSjwep4LGUDabTY7lIXOurq7S VOPPbH8J89An2S2YTCZnZ2fdbnckEsFQHz9+xK8ymQzePsC2mYc+yTntl5eX 6+vrou/Z7XacHIuJjEymenlTgMEwD33CSnKWeyKRkDPBAAEo9ymsrKzQtsHe uGcq+oSt7u7uZBUWRru/vzf+QPuhgNavRhnmIWBmhnkImPlhHgJmZpiHgJkc 5iFg5oepCJiGIrR+9VtAS2Dvt7e3svIKVt8+J3mwMBV9wiCVSiUcDjP/fvjw YWpqKhqNZjKZgR8CbDb6xJzSarVo0qdPn8bHx8fGxhwOR7FYrNVqg23YVzPR J7kogTak02mmlb/85S94FK6+s7Mz8OgzG30iOz09PW1tbc3MzJAQsJUcfX9z czPA3c1fTUafsANuQ/Y+PT3Fkf785z/DVbBYoVAww117pqJPEn2kqWq1yixD GIqTawnrO2j9apRhKgJmWpiNgJkZ5iFgZobZCJhpYSoCZnKYioBpKELrV68D xo5j48xXV1cYp1KpyG/qB17vfDUTfZKVRZgoEAgsLCx8+PDh06dPFNHr6+tU 1lq/eglZgpXL5SYnJ8fGxv7yl79gsZOTE6w32IZ9NRN9kgq6WCyura3RJKyE uSwWy9bWFsH45cuXAbbNbPSJtz8+Pm5ubk5NTU1MTGArJuJ0Os0IDlZwMBt9 wqNKpdLBwYHNZvvzn//8xx9/kKaOjo4ajcZgPeqr+ehT+yx3mfjkmt3BLlEz IbR+NcowDwEzM8xGwMwM8xAwM8NsBMy0MBsBMzPMRsA0VKD1q1cgx27XarWb mxtmE5g8Xj3YHV4vYR76RPVHbszn81JB/7/fIOfMYD3qoAFWPWajTw8PD5lM JhaLyeIr/pydnZW1arIdbIBtMw99wqOItWw2a7PZsA9jJ8pMPB4vl8t3d3cD bJvZ6FO9Xi8UCpFI5E9/+hPuRNzNzc05HI7d3V0aNkD12FT0ifSINTY2NgKB QHuh2ocPH1ZWVvb398lRA2zbV1PSJ9GQG40G099/fwO5VG8kfAmtX40yzEPA zAyzETAzwzwEzMwwGwEzLUxFwEwOExIwjd9C61evQFZe3d7eXl1dweGr1erA 9y69hBnok0h8zCBy/pXVarXZbLAUv98/OztLQuBz2knVMyi7mYc+YQHKZEYq kUjQEvs3yGk8Xq93eXm52WwONm2ahD7hVHd3d+VyGXKCR7lcLuZfGjY9PR0O h7PZLG0bYCSahz4Res/Pz8VikfEKhULYx+l08hcsRhiur6+TuB4eHvT57eJR 5PBYLIZxvN9A5pyfn49EImtra/wXw6rPb/8OcpZ7o9EgaxF02JB0qldhtaH1 q1GGGQiY+WEeAmZ+mISAmRzmIWAmh3kImPlhTgKm8Tq0fvUrwEbq9XqlUmEe wSzkTLMdAGIG+kSGpMA5Pz+3fUMgEEgmk/f39+Vymb8zF8/NzVEhtlqtp6cn 45v31Uz06fHxEX6CfUiSFM78vVAoYKidnZ2PHz9arVb+q1QqDap5X81Bn+Si z1wuJ9fD4T9MKHgUg7i8vMyH+Pzq6urt7e2g1syYhz5hFoZpaWlpfHzc6/Wu r6+fnJw0m028fXp62uFwEIP5fH5QgoNJ6BN9//LlSzabjcViRBmWwbuurq7w KJqHg01MTODwmG6At3+alj4xrRCSjJ1sJGSioW1awhJo/WqUYQYCZn6Yh4CZ H2YgYOaHeQiYyWESAjYUMC0B03gFWr/6Kah3np+fq9Xqzc0NvJ1imX8O/IyU 7zBw+iQLigh2akMKQ+rlZDK5v7+Prer1eiqVYiLmc4ro8/NzLMn3ja96zECf GBrmXCyAfbCJzWYLhUKlUun6+lo2f3k8HrfbzSxDZf3w8IABjW/kV3PQJ9gI ZtnY2Jifn2dCiUQiDB+jhrfv7u7yT8aRFmIovkZIGu/2ZqBPovJhE/hJOBzG JouLi4eHhySrp6cn/oKDiQCYTqdrtdpAdj2bgT6RcBggXr2ysuLz+bAJfxaL xbu7O6x3dnbGPy0WSzwe39zcxFBYbyB1qGnpk+wivL+/FwmLiYa/0zxdrX/V +tVoY+AEbChgBgI2LDADATM/zEDAhgJmIGDDAtMSMI1XoPWrHyFXLxHsl5eX 5XL58fGRosaEv24eLH2igoaZlEolp9NJG1wuF3ny4eFBphJRbDCg2+2miGZS 5n+xpPF1tBnoE8mwUChQIE9MTMzNzWWzWQpn0V7wKwxFwsxkMh8+fAgEAtvb 24M6y32w9EluiDs6OgqFQg6Hg/Ha2NhoNBoylcg1oAyl3HEp/9tsNo1f12cG +kQoVSoV3j45OUnoJRIJPKqtD+PztVotmUx+/PgRWy0tLdXrdeMzmBnoE23I 5XKYwmq1Yivag2NLPhePyufzOBJRyf8SmKSsgawUNTl9wrWwlVxHCEQm1QW7 1q9GGVq/UoEZCNiwQOtXKjADARsKmIGADQtMTsA0fgqtX30HQh5PJtIpD+Ek Nzc3EHXznNn+EgOkT1R/mIUZdm9vz2azORwOamRSZfu8aNmzQxEtyZNG8mep VOITEW0Ma+pg6ROdfX5+xp3IiphIzgQrFArX19ft8cK7bm9vqaOxpN/vTyQS NJhmG+91g6VP0BLcA0M5nU6PxxMIBHK5HB+27YDb0Kp0Os04MprxePzk5IQf MXhd32DpE25DjiIvMViLi4vYIRwOE3p4VPs7tIdW7e7uer1eLIlTMSM3m02D V5AOnD4xUvR6Y2ODt+PYLpeL0Ht52yDGLJfLWBIrWSyWlZUVGswXjN8qbnL6 JLsIW62WTItAloma8Nc6RkLrV6MMrV+pQOtX6tD6lQq0fqWIgROwIYLJCZjG T6H1q5eAgZAMYemlUuny8vL+/p7J17QUfVD0CYMQ7LyRutjpdM7PzzPhUvT9 eHoM+RMbnp2dzX8D308mk09PT0aevD1Y+kSJxzDJCVdzc3Pr6+uHh4c/3gon jpfL5ZhoxsbGmGvy+Xy1WjWyqV8HSp9wKvwkHo+7XK7Jycm1tbWfnnAlJ5Yf HBx4vV6Hw4FJM5mMwev6BkufCJ+7uzsGSE5rx6WPjo6+20cpF1kSevge7cSe kUgklUqR2Qxr59dB0ydC7PT0dGtrS064SqfTOBhj910+52sM3+7uLu20Wq1k CcLw5ubG4EQxFPRJdq1WKhVZgsVfRvwgLK1fjTK0fqUCrV+pQ+tXKtD6lSK0 fqWOoSBgGt9B61dtyBaJRqNRrVZhI/BzZl4z/4p5IPQJa2CT6+vrw8NDh8NB BR2NRqkNmVN+XN1Bkyi3+TIzciAQ4Mv8BcNSRxtm1UHRJzpIrddsNimN19bW LBaLx+OhNL68vPzpSGFV/mt1dXVmZiYYDCaTScptsqiRa2YGRZ9wEgzFVMsM IkuGdnZ2fnoOmOwxvLi4gLfgUbOzs1isWCzy44at6xsUfWqfpF0oFJaWlqan p/FnRqpcLv+04zgPZtna2mJMsVU4HMbBsCqsxoDWfh0ofcKj6Cl9TyQSdN9u t+fz+V+pUlj1/PycpvJNPEpE5vv7e0LPsDQ1FPQJQzGmTJHtO3lp5yjXDlq/ GmVo/UoFWr9Sh9avVKD1K0Vo/UodQ0HANL6D1q8E1CmPj493d3eUeKVSqdVq kRVNTkiMp09YiZRYqVRIhh6PZ2ZmJhQK1Wo1bPVKoUcZWK/XaaTFYpHrCEkR nz9/NqY2HBR9koUKEJLp6Wl6HY/Hs9nsjyuvXoIRZEwzmYzf7//w4cPi4qJs +zKmwV8HR58IulQq5fV6x8fHl5aWyuUykSiLiH76fTwKDsPUbLfbRUTd29t7 fn42RpkZFH3Co3BmWVBEx91ut9yV8KteiwEbjcbx8THfn5iYoM044cPDgwGt /To4+kQc4VG5XG5+fh5CkkwmcWx6/UqGpKkwFtyedtpsNqvVKjc5GqYeDxF9 wg+fnp5ub29JEfzJsBq5ntZU0PrVKEPrVyrQ+pU6tH6lAq1fKULrV+oYIgKm 0YbWr77+80gZObNdrggnJeLApl15JTCePlG51Go1KjtqZ+bZYDCYSqV+O33I NYXlchn2IleAJRKJarVKRWmAhY2nT6Ly0Ts5Gtpisbhcru3t7UKh8NtCr16v n56eYh9ay6RDLqUSp83GKDPG0ye6hqFyuRzTK9OH3W6nyxjhx72oLyEnb2NP DOX3++fm5vBDuRnNAI8ynj7JYePNZrNUKq2urs7OzgYCAf7CPPvbs78IPRJF OBxeWFjAo2j21dUVHxqgOQyEPjEW9G5vby+ZTDqdTnxjd3f3t6eyS8weHR2R KPgpq9WKHxKJDK4x4swQ0SdZqMxomvlyXmOg9atRhtavVKD1K3Vo/UoFWr9S hNav1DFEBEyjDa1fUbngq+TAcrlMmdNoNKiCh4KKGEyfMFS9Xl9aWgoGg1NT Uz6fT30nYHvX4cHBATSG8pCkSl1pgCxjPH2iU3d3dxS/1M42m425AzaCj6l0 tr0/bm1tzePxkE4xlJyW3O9mfx0EfcKFtre3A4HAhw8feG+hUJCb8lScCo8i VNPptMVicTgcOCTt/+4kqH7AePokp37t7e0xOvR0YWEhm80qnvAvNykUi8X9 /X28cXx8PJVKHR0d8eN9bfPXQdAnhv7m5ubs7MzpdJIbE4nE1tYWwajoErLC jTGFyczPz2Pn8/NzQs8ACWvo6BMNxlZkCebNarXKvDkUk2ZvofWrUYbWr1Sg 9St1aP1KBVq/UoTWr9QxdARM4+vI61ey8ooqGPrNfEH3TX5m+0sYSZ9kfRpx 7ff7PR6P1+tdXl5uNpuvr5P57gkUkjxBlmAFg0GmHszeb2sbTJ9ksVmhUCAN Wq1W6uiNjY18Pv+mg44rlcrBwUE0GqXNkUhkd3f36urqy5cv/a6jjaRPohUc Hx/LGiqbzYahbm5uYCOKTxBV5+joKBwOM/UsLCzQ7J+e+t5bGEmfRMp7eHgg wFOpFFYifGKx2OnpKQZU9AcyAx51cXERCATwycXFRUzNJ8zR/U4aRtInDEKP CPZ0Oo0bt9dQqecorMFD9vf319bW5HYA2l8sFnlCv4vToaNPpCPZRSgHYVWr VcOWiZoHWr8aZWj9SgVav1KH1q9UoPUrRWj9Sh1DR8A0vo62fiWLgr77JfJv 9+OYB0bSp7u7u/X19UgkMjk56XK5qIUx11tPzJalbqVSaXt7m+dQHsJqjo6O +irLGEmfRLyi4KX4ZVw8Hs/KykoHt+OJrHpychKNRuEznz59SqVSzD7qlXhn MJI+1Wq1fD7PrDo2Nsa45HI5auEOzmDHJvV6Heecmpqi8eFwGLv19Sx3I+mT eAKGCgQCONXMzAxDQzD+eLL9b5/z5csXLCOy6sTERCaT6fc0bSR9ktWhl5eX brf748ePuEEymeSTDnKLrOtLJBIY3G63E8XMDv3W+oaRPg3v0uVeQetXowyt X6lA61fq0PqVCrR+pQitX6ljGAmYxsjqV3KkDKmv1WrRayaLoTiz/SWMoU8U gA8PD5QnkUiEIprpNRaLdXx6VfsErb9+AzP15uYmlu9fujCMPmF/JlYMtbu7 a7PZnE7n6urq3t7e62e2vwLGNJPJMPXQcmx+dHQkF9b3z0WNoU904enp6fz8 PJlM4lEzMzMYCrvd3d118DRZx7W/v+/3+5mA3G731tZWvV7vn0cZSZ/oGrFG jNjtdq/XiwMfHh5ivQ4WuuA2eBSxwHMsFsvS0hLehaGen5/75FGG0SdZiXd2 drazs+NyuWZnZ1OpVC6X62wlHtHK0/Aigk6E6IODA5ItZu+fKDqk9Ilm09Tr 62vsg6PiTiO1CkvrV6MMrV+pQOtX6tD6lQq0fqUIrV+pY0gJ2IhjZPUr2Wxy dXXVvkTpTTu8zABj6NPDw0M2myUHTkxMUEQfHx/zrm5WqfGDsnEMSsMzbTbb 7u6unETd25YLjKFPGJ+aF8uQA+WEIorfu7u710+Nfh0YBC89PDz0+/08k6o8 nU4zHG9de6MOY+hTq9W6uLhYXFz88OEDXctkMnhylyum4DAU0SsrK0SE2+3G XXlFD9v8EsbQJ9k5WCgUotGo0+n8+PFj92vw8CiaenBwsLa2ZrVap6am9vb2 cNo+BYUx9En2V5LDA4HA9PQ0HhWPx6vV6pcvX7rxKPJGvV4Ph8Nzc3MulysY DN7c3HT5zFcwvPTp5dW9QH4NNFwzacfQ+tUoQ+tXKtD6lTq0fqUCrV8pQutX 6hheAjbKGEH9SlZeMave39/DOujykG586Dd9kt1wV1dXkgAXFhao5sRcXZYn 5AoanMvleCbz9crKCmmWaagfyowB9Ek2omIo6Ifdbnc6nYuLi5lMprN1Mi8f y5iWSqVUKkX5TG3OQJyfn1NW92l/XL/pk+gnFLntdWWJROL09JQedflkTE0F nc1mXS4X01AgENjb2yOo++FRxtAnJlACbXt72+12ezweOkWPOtiL+h3kdgAe xWMtFgvt39/fb7VaPLbnHmUAfZK78MrlMq6LlejR8vIyRuv+JkpZ10cg0HIi mrggjVQqlS7t/ysML31qb8MnAZLVa7Vas9k04LA+M0DrV6MMrV+pQOtX6tD6 lQq0fqUIrV+pY3gJ2ChjBPUr2WwiWx6g3JQkBtxZ1g/0mz5BPI6OjtbX1wlq m81GnVssFntlKx7y8PDAM1dXV3k+veBdsg6n+4e/RL/pE+UbHoUXMUe43W6r 1cpMUa/Xe3UuPQYnl+7u7lJELywszM/PZ7NZZvB+1NH9pk8wDSJueXl5YmLC 6/Wura1dXFxQ53Y/6LJaqdVq4aLxeHx8fNzn821ubvYjoRlDn7DMysqKx+PB VliMAO9ekxHI/jim6Wg0Ojc3NzMzk8/n6RGe1v3DX8IA+sRYNBoNWSVF9IVC IRyM4eiJodoru4LBIFaC3hAatVqtH5PFsNMnyefNZrNUKl1eXg7RNSjdQOtX owytX6lA61fq0PqVCrR+pQitX6lj2AnYaGLU9Kv2kbOy8kp+WTxEZ7a/RP/o E4/CSpVKZWNjY3Fx0Wq1BgIBCmo+6eFbnp6eeGY2m+X5DoeDd+VyOdhOb5WZ ftMnWlutVo+PjymfXS4X0wTv6n6dTBtygvfZ2dnS0pLP57NYLGtra3JalIg2 PXmLoH/0Sc4Pv7q62tvbi0QijAVvOTg4uL297eFbGGsiOp1OLywsYKtoNHp4 eNhzetNv+iTnh2OoYDDIlMqIbG9vt1qtbvaifgcynpzlLvf0MdzECB3p7R7q vtIniYvr6+tCoSBxIfcqks97m6MYC0bZ7/cT4AwHr6vX6z3X+oadPuE2xIVc lMnEhIlIUH09rM8M0PrVKEPrVyrQ+pU6tH6lAq1fKULrV+oYdgI2mhgp/UoO FLq5uRGOTe2jfge9CdE/+kTwEsKZTGZiYoLyNp1OHx8f92OVGsan3uT51J70 pX2zYQ+L6L7SJwxSq9VisRjl7fT0dCAQwLt6tU7mJWg20zQjMjc3Z7fbKaUP Dg567r39o09COVKp1KdPn3hFNBo9OTnpeXkrgh4v4uGRSGRsbCwUCu3t7fVW Jes3fSKumUPxqD/++APXIiKIkX5s7iPMGRGc1vYNZ2dnuG4PD9/uK33C5g8P D/F43P4N/bslUG425OFerxfvpS+rq6vd76H+Du+DPmH8VqvFQGMuEsjj42Nf D70fOLR+NcrQ+pUKtH6lDq1fqUDrV4rQ+pU63gcBGzWMjn4Fi4ZLy4FOkI1K pUJVMtS/IO4HfZJzyLFMJpNZWlqyWCxUbX3a2SdgMqL9zNdWq9Xlcm1tbZ2c nDBSvVrh0D/6RAspY6n6/X4/VvL5fBS2XZ7Z/iuI9Hp8fByJRHiXzWZbX1+/ vr5m7u7hW/pBn2SdDOGGF8ViMUaByRTvwmN78vwf8fDwQIynUilZN7iysnJ6 eoqb9cqj+kefeA4JCkMtLi7iToT25uZmtVrFh3vy/JcgH2KQg4ODZDLpcDiw 1fb29vn5OQ3olSjaJ/pEy2khZikWi8FgkFgIh8OMcp/u4JD7JkiJiURClmCR Rni1rC/q1VveB33CVnKlr/yGqNlsEoxDPcm+Dq1fjTK0fqUCrV+pQ+tXKtD6 lSK0fqWO90HARg0jol/J9qXb29v2yqvn5+dhv+a7H/RJTlbf29ubmpqipF1b W6PC5cO+rlLj4aRWKlDoDe8ljdAGiqCelKL9o0+0kNmTeeHTp08ej+fy8pIi t6+LDZ6enigJ6cv09LTD4QiFQnStJ+dHCfpBn2SdTDqdnp2dpc1+v5+39HXd o5x7f3Nzs7+/j4nGxsYikcjJyUn3p8QL+kefSFCEG15Km8Ph8NHRUbVa7dNZ /QIZHfricrnIJE6ns1QqETI9GZ0+0SdZOba6ukrQzc/P2+12+Aa96KtH8XBG 5/z8HBNNTEwsLi7i0vf39716xbuhT2IrYq1SqZBDmKSIl16dSGY2aP1qlKH1 KxVo/UodWr9SgdavFKH1K3W8GwI2UhgF/QrmjHPikFdXV3QQXk0gD/XOQUFv 6RNWwiatVouqn0nBYrG43e69vT05YbtXbf4VmIN4V/u9tIH39kQ36wd9Yl6g Xr68vGRSkINxEokETvX4+NjXMk3OcofkBINBr9dLKU0R3cP1Ob2lT6IjNRoN qn7mUNlfmUqlSqVSXzUZgZzlvrKyMjMzEwqFcAA8qssbIQX9oE8ysoVCYXV1 FStZrdZkMkmA42ZdPvl1yI2QOzs7dMfhcNhstt3d3XK53JNtX/2gT7QKj8Iy 4XB4YWGBkY3FYjc3N/1WSHj4/f397e1tNBq12+3yXmKE0e9JenxP9Kl9ljvD xMR0d3dHYjQg5I2H1q9GGVq/UoHWr9Sh9SsVaP1KEVq/Usd7ImCjg3evX8n1 cLVajRqHGeHq6grP7NMd6Aajt/SJCpr8ls/nqfcpnynTCGTKWGNUPtnNVKlU mLt9Pp9oDj1RhPpBn6jOMplMPB6fnJx0Op2FQgHvMuwWAGxCHb22tsbbXS4X dfTZ2VlPznLvLX1iQGEa29vbskgGQ+3s7BB9PT/7+qcgHBiRcrm8ubnp9/vH x8cTiYSce9/lk/tBn+r1Ol5E0I2NjQWDQQx1fX1tWMnPoDSbTcadkXI4HAQg +aR7Rajn9AmD0CpimeRgs9nm5uaOj48bjYYxHiUri4gLkiRZF6qzurq6t7fX kz2874w+veNp9yW0fjXK0PqVCrR+pQ6tX6lA61eK0PqVOt4ZARsRvHv9SlZe Ue9DoeHS1Wr1Hay8EvSKPkmtwSzQPoeKOVTObDfSULLCIZPJkG/l3C3aQx8p TrtpRm/pEy3BULQqkUjwNLvdHo1Gr6+ve7XbUQWy52tnZ8fj8WAlsi6TVLPZ 7D7l9pA+YQ1scnl5uba2xmj6fL6lpaWzszODD8Ohgmb04/H49PR0OBzOZrOM XZcZoLf0CYM8Pz8Xi0VsHgqFaCdPPj8/Z0A7fuZbIUojIY8zO51Oh8NxeHgo m6y78ere0idacnd3Rw6PxWIulwuP4pmMZq92Oyq2Qc5yJ/bn5+cjkcjq6iqG 6r4N748+EeaMF+aSXYTkdqz0zlZhaf1qlKH1KxVo/UodWr9SgdavFKH1K3W8 PwI2Cnjf+hWM4h0fxNEr+iS74SiZ5Roy2eTVw/Ou1SGnx9MX2sA8Pjc3R4XY arW6ORG9t/Tp8fExl8sxaZLoKGDhGxSzhq28akNqeV6dSCTGx8e9Xm8ymSyV Sl0+tlf0SU6bx1Aej8dutzOOmUwGH2Nwu2zhWyGrsJiPlpeX3W438bK2tnZ7 e8s4dvzM3tIn6npMvbS0JOO4vr7OOA7Eo+hIrVYLhUIzMzMMHISHzNnNieg9 pE+yPjObzcZiMavVOj09jXcxjsYv6ZGjFMmWdA3HnpiYIFhOTk7wim4e+y7p U/vYSdm5f319TTeH/djJl9D61ShD61cq0PqVOrR+pQKtXylC61fqeJcE7N3j HetXMGfq5Wq1enNzA3+GRfNP4zWZ/qEn9IlKmVqegKU2pDB0OBzJZPLg4IDC cCB8TM7+TaVSTOJyfxylYje783pFn7AG8yYtwT60zWazUemXSiVK/oEooiRY DEWyZchIvMxQVPRdakS9ok8wCmrVjY2N+fl52haJRBiCAZaujNru7i7NwAfo HYaieaSCzjy8V/RJVD6iGI4RDodp2+Li4uHhISmrg6d1D3pBXwi9YDCIG7jd 7pOTE9y7Y42oV/RJtDW54sHn83k8Hv4sFotyt11nz+wGcjvA2dkZzbBYLPF4 fHNzE0M9PT11nDPfJX2SHZfNZhPjMEMxCzPX9ORoNZNA61ejDK1fqUDrV+rQ +pUKtH6lCK1fqeNdErB3j/eqX8ElWq0WAXt5eVkul0lxOOS7oc2C7ukTxQXs olQqOZ1OnuNyuch1VIWDnQ5EKWLgqKDtdjsTOq2i8Omsju4VfZLjtSlUJyYm 5ubmstksBWzHGkhPILuZaAaE58OHD4FAYHt7myKx4wd2T5+kYj06OgqFQg6H A5tvbGw0Go3BTgc0iZE6Pj4Oh8N0UFpFWd3Zur5e0Sc8s1qt8gQ5xyyRSDCU xq+8egnZ9YkL+Xy+qampYDC4tLTU8a2avaJPPCeXyyWTSavViq0IvV6dMN8x ZGVRPp/HkWgVeZhWkbI6Xin6jumT3BFwfX39399wc3Mz2LTZQ2j9apSh9SsV aP1KHVq/UoHWrxSh9St1vGMC9o7xLvUrwhZvJForlQq8AsKMN76/FNclfaL6 o9piltzb27PZbA6Hg1qVdNeTW/+6gewVqtVqknjpIH/KSqcOjk/pnj7Jrkbc icyGiWZnZ0l0hUJBTth+69N6i8fHR9ybIpoR9Pv9iUSCztLlzrS+7ukT1IJh wlBOp9Pj8QQCgVwux4eDPbpZDrcnXtLpND6AJ8TjcVlc1IFe1D19wm3IUbe3 t4eHh4uLi7QnHA4TenjUm57TD9C7ZrO5vLzs9XoZQZyK2ZxPOli52hP6JO3B w3kCzulyuU5PT+v1+sBX0jKI5XKZkMFKFotlZWWFzjYajc62qL9j+iRan2zk Jwavrq5IUF0erWYSaP1qlKH1KxVo/UodWr9SgdavFKH1K3W8YwL2jvH+9CtY BAmt1WqVSqXLy0s5NvYdUOUf0Q19wiAELD9Ffep0Oufn55k0Kb66PMWlhyD3 MnZnZ2fz30A7k8lkB/chdk+fKLUw9c7OzsePH+fm5tbX1w8PDweu8gkYR8b9 9vY2l8sxSY2NjTFP5fP5zvagdUmfaAzjFY/HXS7X5ORk9ydN9RZyUvrBwYHX 63U4HHIqVwfkp3v6hBvf3d1h5OnpaaIPtzw6OjLJohTxKPzn4uJCxjESiaRS KTLqWx/VPX0ixE5PT7e2tuSkqXQ6jYNhf5Pkc1lZRGZYWlqyWq1kGMLw5uam gyTz7umTRJ+swrq6uuryaDWTQOtXowytX6lA61fq0PqVCrR+pQitX6nj3ROw d4l3pl/JRqFGo0H9JQduvJtf9f6IjukT1pBS4vDwkEKeCjoajVIbDuTM9l+B 7lDm00hm80AgQCP5CwP61pv+uqFPvOjz58/NZnN3d5eJ0mKxeDweOMbl5aWp yOrd3R1NWl1dnZmZCQaDyWSSMp8M/NbR7IY+MVgYiumSWUCW7lDXD+TM9l9B znIvFosMJR41Ozu7srJycXFBs9+0rq8b+iRt4McLhcLS0tL09DQ+ibXL5bJ5 chQtYeBIobFYDH/AVuFwGAd763lTXdInPIo3bm1tJRIJmmG32/P5fGfqUP/A aBJr2WyWFuJRIm7f398Tem8a0HdPn+S0t1qtdnt7+24WRWv9apSh9SsVaP1K HVq/UoHWrxSh9St1vHsC9i7xnvQr6oXHx0ep5UulUqvVIrO9Y1LRGX3CSqS1 SqVCQvN4PDMzM6FQiLICW5mngm5D1Mjj42OLxTI3N0cdnU6nqYPUm9oNfeJF GApSMT09zdupo82wv/JHyJoZ/CGTyfj9/g8fPiwuLsq2rzc9pxv6RNAlk0k8 anx8fGlpqVwuE4myce9Nz+k3GFP4D+Not9sd37Czs/P09KSuzHRDn3g7DikL imiA2+3e399/fn4226Vsco7Z1dXVyckJJiLP0N9sNvvw8KD+kG7oE/6MR+Vy ufn5eUgFroVz8nYT5nO6CdshZOijzWazWq0Y7a07LkeEPtFNwge/Ir3c3t7i EmbLpW+C1q9GGVq/UoHWr9Sh9SsVaP1KEVq/UseIELB3hnejX8mRMnJmO72A TrzLM9tfojP6JL8Ep8KidmauDAaDqVTKtFOAnDDPVA7zIb14PJ5EIlGpVKhk FUe2M/okKh9vkSOaLRaLy+Xa3t4uFAqmLbjq9frp6ams6mHCIg9fXl6+6da/ zugTr8BQBwcHvJQxstvtvJrGmGcv6kvIEk3GEUfy+XxWqzWZTBJK9/f3ih7V GX0SKU/2Na+urs7OzgYCAf7CXDnYM9t/BZrUbDZJLOFwmDGF/9BluUJOMQQ6 pk/Yk7fs7e0xNE6nc25ubnd3t5vT0fsKyRVHR0ckGVqLR+H/RCKOoZ4rRoQ+ ydGU7+ZSYK1fjTK0fqUCrV+pQ+tXKtD6lSK0fqWOESFg7wzvQ7+igsDfyGPl cpkyh9r57u7u3dOJDuiT3Fi3tLQUDAanpqao3zvYkWcw2rsdDw4OoECUhyRk 6llFWaYz+sTDcSGKUGpnm81G/odR4GNmWyfzErJm5uLigjnd6/WSijEUjqG+ ZqYz+oQLbW9vBwKBDx8+8LOFQkFurDOzU+FR9/f36XTaYrE4HA4Cgb4rnkDV GX2Syn1/fx8L88aFhYVsNtvxSfvGQE7eLhaLNJsoGB8fTyaT+XyeZqv8eGf0 iZfe3NycnZ05nU7yWzwez2Qy5s/nsrJubW2N0Jufn2d8z8/PCT1FCWt06JPk c/mtBPN1tVptNBomH9xfQetXowytX6lA61fq0PqVCrR+pQitX6ljdAjYe8I7 0K9k5RXVKDSYnE8XKHYeHx/NXD73BG+lT7I+jdj0+/0ej4c6a3l5udlsmnOd zEuImkTLfT4fLQ8Gg0xbDLfKyfwd0CdKTvynUCiQyqxWK3X0xsYGZftQHDhc qVQODg6i0Sj9jUQiu7u7V1dXX758Uamj30qfpGY/Pj5OJBI4lc1mw1A3Nzcw ih70pJ+Q06SPjo7C4TDT1sLCAl1WPG3+rfRJpLyHhweCNJVKYSXcOBaLnZ6e YkDTLucTkFXwqIuLi0AgQCzgG+vr63zC/K6ScN5KnzAITyZg0+k0rsgb+cvJ yYn5cxTWoPF7e3urq6tyOwB9LxaLtFylsB0p+oShnp6eCDdSE+miWq2+aZmo eaD1q1GG1q9UoPUrdWj9SgVav1KE1q/UMVIE7N1g2PWrn/4y15z7cXqOt9Kn u7s7as9IJDI5OelyuahJMdebTq4eIGSJXalU2t7epv2UhzCio6Oj35b/b6VP Il5ReFKEYluPx7OyssInZl4n8xIi51LyR6NRuNCnT59SqRQzl4oC8Fb6VKvV 8vk8M+PY2Bi2zeVy1KTD4lFfvxGher1OUExNTdHxcDiM3X7b/rfSJxkRDBUI BHCqmZkZzEswmudk+9chq7CwjMi5ExMTmUxGZYp/K32S1aGXl5dut/vjx48M RzKZ5BOTS3wvIev6EokEA22328kezEoqWt+o0afvlkwza2O3oVMAtH41ytD6 lQq0fqUOrV+pQOtXitD6lTpGjYC9Dwy1fiXHVpO+Wq0WLSfhv/sz219CnT7J UU6UCZSEFNFMkbFYrFqtqp8iZQa0T+766zcwy29ubjLir6eaN9EnbMjkiKF2 d3dtNpvT6VxdXd3b2zPhme2vA3/IZDJMW/SasT46OpIL618PDXX69OXLF8x+ fn6eTCbxqJmZGQyF3e7u7nrck35C1o/t7+/7/X4mL7fbvbW1Va/XX/eot9In XkGspdNph8Ph9XpxwsPDwzedGD9w4DZ4FHFE+y0Wy9LSEt6FoZ6fn19PO+r0 SVbEnZ2d7ezsuFyu2dnZVCqVy+VUVsSZB2QJeoEXEXQigB8cHJCoGe7XM+0I 0ie6TDevr6+xDwGCOw3dKiytX40ytH6lAq1fqUPrVyrQ+pUitH6ljhEkYO8A Q61fyWaTl5cZDcUOr15BnT49PDxks1ny2MTEhN1uPz4+5vvDuEqNBkOHTk5O mN/pi81mo9o9Pz9/RVxSp08YkNoTy5DHZPlEPB6/u7sz56nRr0Nuqz88PPT7 /fRldnY2nU7jBq+v+VGnT81mkyS/uLj44cMHXpHJZPDGIVp59RLwH4rolZUV osntdhMmFxcXr3xfnT7JzsFCodDBWjizAY+imwcHB2tra1ardWpqam9vj2B5 JaDU6ZPsrySHBwKB6elpPIrQq1arX758GUaPIufU6/VwODw3N+dyuYLB4M3N zet9GU369PLKYCC/fhqiEdf61ShD61cq0PqVOrR+pQKtXylC61fqGE0CNuwY Uv1KVl4xM97f38McaPaQbkDoBir0SXbDXV1dSRJbWFigqhJzDVGZ8BLkGbrD RE9fnE7n8vIyKZop7FfKjCJ9ko2o8mS73c6TY7HY1tbWcK2TaUMCpFQqpVIp yufp6Wkc4Pz8nLL6FZVJhT6JjkGxSZ7Hnlg1kUicnp7y5D72p59giKmgs9ms y+ViCgsEAru7u6/s71OnT0yCBNr29rbb7fZ6vT6fb29vb4j2on4HuR2ALtAd i8VC3/f391utFt35qUcp0ie5EbJcLuN+Ho+HJxPUGE39RkizQdb1EUT0mkxC TJGCKpXKK+M+mvSpvf2fxMsUVqvVms2m4mF9ZoDWr0YZWr9Sgdav1KH1KxVo /UoRWr9Sx2gSsGHHkOpXstlEth5AfSkNFO8Oe09QoU+Qh6Ojo1QqNT4+brPZ qDeLxeKw24rGPzw8yFIQEg4WoI+y/ufHL6vQJ8oobHJzc8MXqM2tVivZvl6v q5wPb2bILr/d3V2K6IWFhfn5+Ww2y+z/qzpahT7JOeTLy8tY3uPxMAQXFxfU m8PrUbJKqtVqlUqlRCJBpPh8vlcsoE6fiLXV1VWsNDExgcXIVMO1Y/dHyP44 pvhoNDo3NzczM5PP52u12k89SpE+tVcrEXculysUCmEoTDrUhmqvKAsGg1gJ akRYYahfhcko0yfJ581mkwC8vLy8v78flsSr9atRhtavVKD1K3Vo/UoFWr9S hNav1DHKBGx4MYz6VfvoV1l5Jb+0HcbdcF3idfrEP7FSpVLZ2NhYXFykNgwE AhcXF3zyDojW09NTsVgkOdMvh8NBH3O5HEzpxzpahT7xU5iFr1E+wx9I9Xx/ eNfJtCEnh5+dnS0tLfl8PovFsra2JqdUiWjz3fdfp09yjjeetru7G4lEsCff PDg4oEg3qD/9hJCidDq9sLCAraLR6OHh4U+pkQp9knO89/f3Q6EQ0yJ+lc1m W63WMO5F/Q5kWjnLHVfBo3AVAufh4eHHvdu/pU/in1dXV6enp1hpdnaWTEUs k8/fR47CB/AQv9/vdrvpYKFQqNfrBNGPXx5l+oTbEFMijDOpYSIS1G8P6zMD tH41ytD6lQq0fqUOrV+pQOtXitD6lTpGmYANL4ZOv5KDfW5uboTrXl9fm/8O +j7hdfpEABKGmUxmYmJCrqE/Pj4e9pVXL8GgU+fSL2pe7MC8375R8eXXfkuf +H6tVovFYpSZ09PTgUAA7xrevUs/gi4zxeMJc3NzdrudUvrg4OCnUfM6fRLa kEqlPn36xNei0ejJyclQlJkqEEGPDtKpSCQyNjYWCoX29vZ+VOdU6BOxyTyI R/3xxx+4Fp6Jr74bj2LQSS/r6+vBYND2DWdnZ4TMdzttf0ufsNvDw0M8Hrd/ g/ptfcMCuVGRTnm9XqIGO6yurv5077amTwz63d2dmIvk8/j4+NtD7wcOrV+N MrR+pQKtX6lD61cq0PqVIrR+pQ5NwIYRw6VfwWbhtHKgE4ShUqlAd99NBf1W /Io+yTnkWCaTySwtLVksFqqnV3bYDS+YyOg7c73VamXep78nJyd4yMsVDq/T J75JOUn17ff75YQiCswhPbP9VxDJ9/j4OBKJ0Eebzba+vn59fc28/903f0Wf ZJ0M4YYXxWIxLMmEiLXxOkN70n88PDyQW1KplKxXXFlZOT09xc1eetTr9Im/ k6Aw1OLiIu5EeG5ublarVfxwEB3qC2S/bS6Xw1AOhwNbbW9vn5+f0/GXougr 9Ikn8E3MUiwWRQQLh8NY+53dwSH3TZCKE4mELMEiBdFlWV/08puaPrXPcpff TDWbTVnUZ+Y5S+tXowytX6lA61fq0PqVCrR+pQitX6lDE7BhxBDpV7J96fb2 tr3y6vn5eRjP1u4VfkWfiET+ube3NzU1RWm5trZ2cHDAh+9ylRqdIi1T+UKN 6C8piL63Wq12Cfw6feKbzIDk9k+fPnk8nsvLS4rNd1M+v8TT0xMlIXaYnp52 OByhUAizfHdu1a/ok6yTSafTs7Oz/Kzf7+eb73Ldo5x7f3NzI1v/xsbGIpHI ycnJy9PpX6dPJCjCDU/jZ8Ph8NHRUbVafZceJV6BHVwuF1nI6XSWSiXCre0V r9AnWcElh4PNz8/b7XY4A097lx5Fp/CK8/NzTDQxMbG4uEgo3d/fv/yapk9f /2krYq1SqZB/mOCINTOfhKb1q1GG1q9UoPUrdWj9SgVav1KE1q/UoQnYMGJY 9CsYLA6GU11dXdFI+C3B+C4raHX8SJ+wEjZptVpU3yR2i8Xidrv39vbkhO1B t7dfYP6ij+3+0nf629brfkWfyO3Uy5eXlyR2OaAmkUjgVI+Pj6Ytl7qBnOUO QQoGg16vl1KaIvq7dUE/0ifRcxqNBtU386Dsr0ylUqVS6ZV7DIcdRFCxWFxZ WZmZmQmFQjgPHtW+ifJX9EksXCgUVldXsZLVak0mkwQpbjboDvUFchPlzs4O pnA4HDabbXd3t1wut7d9/Yo+8b94FJYJh8MLCwtYOBaL3dzcmFmp6AZ06v7+ /vb2NhqN2u126S/xhee007KmT4L2We64B5Pa3d0dCdm0qUbrV6MMrV+pQOtX 6tD6lQq0fqUIrV+pQxOwYcRQ6FdwVyKxVqtR45DVr66u8K5hP1u7e/xIn6ig yVH5fJ66e25ujnKJYHx+fn7H4tXXf+5mqlQqzPs+n080h7YS9Sv6RJWUyWTi 8fjk5KTT6SwUCnjXu78FAJtQR6+trdFrl8tFHX12dtY+y/1H+oRhYQvb29uy SAZD7ezsEH0/PYP63YBQwhPK5fLm5qbf7x8fH08kEnLu/ddf06d6vY4XEXRj Y2PBYBBDXV9fm7b07hVwhmazic/gIQ6HgwAkF4kS9VP6hEH4X+KRICVHWa3W 4+PjRqPxvj1KVhYRUyRnMjY0aXV1dW9vr72HV9OnNoZoutf61ShD61cq0PqV OrR+pQKtXylC61fq0ARsGDEU+pWsvKLuhsrCaavV6oivvBK8pE/8HZuQyV+e B0UknpycjIKhZIVDJpMhV8t5X2ITiuKHh4fv6BMGwVD8byKR4BO73R6NRq+v r1/uOnyvkD1fOzs7Ho8HK5GxmeCazaak6+/oE9bAJpeXl2tra1jV5/MtLS2d nZ2Z/FCaXoEKGs+Jx+PT09PhcDibzbaj7Dv6BKF6fn4uFovYLRQK8X3+9/z8 HMMOuhN9hyic6XSaIHI6nQ6HAy8iUWMQ/us7+oRH3d3dkcNjsZjL5cL9+C+s +nLX4XtF+yx3cs78/HwkElldXSXtSN81fXqJ9lnusouQ3I6VTCgFa/1qlKH1 KxVo/UodWr9SgdavFKH1K3VoAjaMML9+BSsYogMxjMRL+nR5edlqtc7OzuQ6 MNnkRQyOzvlgcmo9NJK+wwHm5uaoELEJzvMdfXp8fMzlckx8JCusd3h4SFH5 7ldetSEaAl1OJBLj4+NerzeZTJZKpa//N33ik8+fPx8cHLjdbrvdjj0zmcxI UQVZhcVctry8jBGItbW1tdvb22q1+h19qtVqBCOfYE8mwY2NDaw3Uh4lRgiF QjMzMx6PB7JExiYes9lsmz7xhS9fvvBJLBazWq3T09OEIfY059KafkCWyJ6f n8MqCaiJiQkC7eTkRCY1TZ9eon3cpZwYcH19jYnMNp1p/WqUofUrFWj9Sh1a v1KB1q8UofUrdWgCNowwuX4Fg6UIomC8ubmBx8Jm5df6g26XKdCmT/l8/vT0 tFAobG9vUxg6HI5kMnlwcIChRqSCFsjZv6lUCgIg98edf8Ph4eHGxgb0KRgM tlotfAn78B2bzUbFfXl5KStDBt1840ByxlAkalxF1sDkcrmHh4dsNtumT0dH R9SM2A0reb3eSCQCCx3BdY94y+7uLt3Hf7AMhmJ2IxElEgno08rKSqPR4JOd nR35zuLiIvFYq9UG3XBDwewPjST0CDFcyO12n5ycYLp0Oi30CRSLRblqwefz eTwe/uQTuWNu0M03DnI7wNnZGd23WCzxeHxzcxNvub+/xyaaPrUhOy6bzSbG +fvf/87s//j42D5azSTQ+tUoQ+tXKtD6lTq0fqUCrV8pQutX6tD61TDCzPoV fKDVahF0l5eX5XKZNIVTmYq+DhZCn+bn5/f396kT7XY7VMrlcpGvqApHM6Xj M8xuOIysGoIMQJl2dnaorKFPGOfw8JCCcWJiYm5ujtx+fn7+5cuXEWSespuJ Ohqy9OHDh0AgsL29TTUt9Am2mUgkQqGQ0+mcnp7mn+09hqMGimg85Pj4OBwO Yxy8iPhigotEItCnxcVFnI1P5DwxjIZJR2fl1UvIbtOrqyufz/fp0yc8CuMs LS0JfQoGg6urq8lk0mq1YqtsNvvypPeRgqwsyufzGxsbWIMcjjUKhcLCwoKm T9+BUPr8+fP19bUc535zc2OqdK31q1GG1q9UoPUrdWj9SgVav1KE1q/UofWr YYRp9StCD48i4iqVihBXPEqnqZcQ+mSxWNbX18lRc3NzDoeDgpGU1b59b9Qg Z7nXajVJ2nBLj8eTSqVisdjU1BT/5O+YaHZ2lmR1dnaGX40s7Xx8fKT7ssLK 7/cnEolgMCj3o2E9qKbT6fR6vXyYy+Xe/Zntv4Icbk+spdNpbAIJhwlsb2/j P9An7Ea4hcNhPudP/k6tPegmDwwQy2azuby8jNt4vgFbRaNRrMQnQqLg5y6X 6/T0tF6vj6ZHff0mYZXLZaoVTEQCX1lZ2dzctFqtmj59BwwFE5ADBGQVFuXw 8/OzSWRPrV+NMrR+pQKtX6lD61cq0PqVIrR+pQ6tXw0jzKlfwQRISq1Wq1Qq XV5eyvGtJqGs5oHQp/Hx8dnZWWqf+fl5Jr5Go0F6H3TTBgzyNj5zdnaGTSgP bTYbf05MTIx/w9zc3Pr6+uHh4ciqfAICikC7vb3N5XJMcGNjY7iT3FwJJ+cv k5OTcuLT4+PjoBs7YOBRVM0HBwder3dhYQHjMNNBn3Cnqakph8MBMz86OjLV 4hDjIR5VrVYvLi5cLhcRJ3lJ7hnEYnyyublZKBTIUSOez+XM9p2dnaWlJYyD ZQhA3EnTpx8h0SersK6uriqVCp+YwX+0fjXK0PqVCrR+pQ6tX6lA61eK0PqV OrR+NYwwoX4lB180Gg3qIBP+ytU8EPr0xx9/UEHbbLZIJJJOp8ntI7uqoQ2Y 5NPT/8/edYBVdWXdMb1308skMb3/KZNJMpM6mZRJJr1N2kx6m/QymTFRuiig YJdiQbCiNKmCAoKI0kTpvffe4e1/r3sREUvek/fu41zO+u4330QQufvts8/e 6+yzdhfXO5wJrFixwt3dnU3EtSEXiZxELV26lPOEwsJCmXAympqa2BRhYWFz 5sxh47i4uPD/8v9ftGgRG4rr65aWFl591v41rQxVyz0/P5/zJfYo1Z2G+St/ f39OOIuKimSMYguww3Do3rhxo6enp4eHB7vTbAW8DNmp0tLSVC1ua/+m1gd7 VHZ2tio6x15kb28v+atDQr1FWFNTU1lZqTZjjxMZTMlfTWRI/soYSP7KeEj+ yhhI/spISP7KeEj+SkSMQ/6KM1V2Hi4GCwoKGhsbOTrJxOCQ4A9rxowZXPXM mjXL19eXa0a2laygh8EFTn19PedOnDhxYTh9+nQuEoOCgiby/cqDofbMsC/F x8evWLHCRoGtrW1gYCCvwaamJvUCnbV/zXEBDk2cO23btm3evHlcuXD6xB7l 6em5Y8cOLqgnlA75EaAeQJSVlWVlZXFxx+7ES4/DFKeanJ+3tLTIeD4M9hne 7Lha4fSSw7iDg8PWrVtzcnJk+nQwOGjz6lPHEXJgb25utrojSf5qIkPyV8ZA 8lfGQ/JXxkDyV0ZC8lfGQ/JXIsL6/NVALw30UG8H9bRRTwt1N/e11nQ3VVYU ZJXn726rK+tsqjJ0t1BvOw104zsnIgywElugtYKaiqgmg6p2UXliTUrASqdP l9l95Ov8ZbiPXWv25p6SZGrMp9ZSNiNMOgEx2E/sLZ111JQ/ULO7PW9LccIa v1nf+Dh85jn9I/9Z3ySudd0bvWygYifV7aHWcuqspcFe/K2JBQMZBqinldqr qKmAanfXZYTvifAK97JZMv0jT5uPPW0/iV/tVpcW0lGUSA051FpGXQ3UNyFD +kAfXpz9pLmIPaqvNCkr0ifA4yffGV8umvah36yvQxb+LydmxUD5DkPtHmop oa566u+aiB5Fg1h6HVh6htrdDbsjSpPWBsz9zxKbjxdP+9jb7rOUjXML4/y7 SrZTvbL0OmoRzw0Tr+hjj+IX59dvKWVD9ZftyNi0JNbXaaXzl8vsP00Lnle6 zb+nII4qd1JDNnsddTZgqRomJN/O62hQWYC97f0dDT0tNdXF2VWFe5qqCltr SwywTDO8ju3T3039ijtpSLZL/moiQ/JXxkDyV8ZD8lfGQPJXRkLyV8ZD8lci wqr8FRfR/dReRy2VVJNJValUEk/FW6gw0lAQ0ZO7uSc3xlAcSyVbqWInVadT cym1VYPJGZxgmTy/Mr94TQbtXEhbppP/c7T0EZp7w4DbVW02Z7fYTK5znNLk cqNhyR/J9ymK/I4SZ1FpAtXtnXgljwHlTFkiZW+k6P9Q4PsGz/t65v1fjeO1 VfZTyqZfVWN/da/btf3zbqeVf6MN71KyO+1dR62V1NWsFOATBlw+c8VXkUzp y2jz/2jtqwafhwY8bmieeW25zVUVtldX2F3T6nKTYcEdBr9nKeRTSppN+Zuo MVfj8nBcoLOO6vfS3vW01ZY2/pN8Huyee0ez85Rq+6tLpl1Za39Vm/PVvfPv pGWPUtAHFO9IOYHgkLsarf17awuOyf3dVBqPBRX9Extq0POBgbm31jldW2E3 hZ9Ku6u73K4fmHsL+f2dNvLS86CsteBF+7smGIVlAB/Fe9nuVZTkSgFv0Yon e+fe2u06pd7u8urpl3Y7XzroejktvJO8/0whH1PsVMoLBZfVO/E06Ngxupux AOuykSGUbaOSuJ78mJ686MG8cH6oaDMShrIkZAhNhSBFezvBYmkFyV9NZEj+ yhhI/sp4SP7KGEj+ykhI/sp4SP5KRFiHvxpUDqB7WtDUwbUel4flyVS6jYqi qDAC6XpeaF9ueF9uBBVEUGEkeK2yRDTMNORRRw1qQ656dN+Lxdl7Xyc1F4O5 4qI4zYvCvqCAf9DiO2nu9eQ82eB4Vt/0E3tsTml3OL9rxsU0+0pacButfp6C 3qfkObTbF21abLG+Dp3bCobqQJnDhiqJpR0etGUarXmZlv+F5lw94HJZq8OF LXYXNE2f3Go7eZDt5nopLbwDHGDIpxT7C6iJohiUP+3V6CHRN5HFVXBbJdjg gkhKWUCR39PaV8nrjzTvepp1Qbfj+U02k5tszm+2vaCbPYoNtej/yPdJCv6Y 4h0oyx+UFxuKV26/rhWxBgdAGndUIzQVRlG6D/wk4C1a9gh5XNfvcnm3I/vS eQ3Tzmu3m9zrOHnA9QqadyP8beO7tNWGMlagsm7IgU8O9uuanzGACO2sh0eV xIEQjp1Ka17C4nK/1uBySbvDBc227E6TW2wn98+4gFwupsV30dKHlaU3lbID qDSOmgvRjDRofTkjC4J9gD2Bl15jHna0PWto888U/CH5/JkW3DrocungjMnt tme3TD9zwP50cjyDXC8j92vI96+07jWKt6ddi2De2t3U3ah/xs8wgIOtnjbq aqLWEmrKB31Xvp2KY6gouj8voj8v3JAbwg/lhyFDKN4C4pSDP2cIbGH2Jd4O EKAsHsklfzWRIfkrYyD5K+Mh+StjIPkrIyH5K+Mh+SsRYQX+inPvzgbkmWVJ yDw5Bc0LpdxgPDlBh37Ur4LX2oTCkJPVpmK0JOm4v4itxBl4Qy46OoI+AJPg fC7ZnUy2J9D0Y/FMm6Q8v+PHMG0SPzT9GPy57Yn4NofTyO0y8nuGIr6j2kwc TOu45OFqhSuXnEBa/RJ53U+OZ5LdKTCU7fGKTY5R7TNkJTzHkM1x+CoMdTq5 XESL76HY6ZTpS33tur75ZaCqnbRrMW14m+ZcRTMnKx51IqzBhlKMM8JWMJ1i qBPI7iSyP5XmTCGvByj2VyzAllJrv4sl0dtO7TWU6Y+l5/lHOIn9KYqhjleW 3jF0gDvtW3r8VTYULz32QO8/UejntGc9anAdc328WCp2YOmteRmcsPNkxVAn DBlq1NKbvs+jbI4filGzr8Tf4hCXtRYHEzqmjgd60UqU6oW+0IW3Y+nxgmJv GfKoSYfyqGOUIHYiTOp0Nvk8TOveoIJwXPjt77L2+1gOBjBXnCFU7ULDVX64 kh6E4DlyeoDvCQWdVRRD9dk492GbW3jXk/zVRIbkr4yB5K+Mh+SvjIHkr4yE 5K+Mh+SvRIS2/JUBpRzXhs1FyDCLY3ESnXN42uoQTzD4rsIoqs1C41ZPi6LJ o7OqR1G74gSeC8O96yjwn+jr4GLH9niVrTL24aJ78Z20+gVK9YbROuuUGyj6 stVgHwxVnwPqKc6OltxD7lOUetBoK3GRaH8yzb6C1v8DN+nKE6E5M9Cjrzuq BtRxHbXQHcpcTuFf0tIHyelMlM8medSMc2De9W/Q9tlYuW0V6L3RGVShOQ4v XDXHTFVa+K4yzUrTFfLB4zpa+TR6sdhQ7J8cpvTWX2RAv2JjHqV541Kz1x9B mNueaIqtJpHTWeRyCW7yghTdik6bft3F84E+0E11WTisQbvjy+R6CYipfacP xjnVsTTvBvL5EyXMoKzVoLC6m3XItPPqY1u1VkLmsSwBDVdHoK0O+fCyzQ8H S1+7G22BQ7phlvIoyV9NRKi9lP1d5cV5HnNcfbwW78nYWV6SP9g/8aQtfgMG 3vU6Wpv2ZOyKj41aMNc9YO2q3s5m5TqAQW9xfkyAobg4ykjdsWKpV8jGdQlb okqL8iaiWMqRAf2KgY625rqq0ohNQU4OdpsjN7U11vR2tlk0zguJgd7+no6U 7QkRoYEb1vpvXL+6rrp839KT2Ad41CCbpaejeZXfijlus5LiYwpydne1tyr5 lbTVuIam/BX7Q3MxruQURCLPNCkvPThNLYig6jTUhjoTAeZ6p70Wqbvf32jR nQpzdYJpFfRweWhzHMgZl4tp2WOUG4SsXmclT1cDPCHBmdx+jxaF6ceZRl6N ZLHsTqJZF9GKJ9Cx1lqBHgDdxC4k2924Jhn4Hig+u5PR9TGif88Ej1IN5Xgm 7ojtXIR7PTpDew0uIsVOBV0w8wKlPeaYo/Eom2OxbF0uwZXerXZopGRf1Q8M mKaR6Qfmat5NiFFDLXymxyh+2KN48a59Fb1JDTkK0aeXpccv0tUINib0c1wV nHm+QvGZuu72eRR7o/P5uFe4w4NKtoDC0hO4AGmvo+Yy7H0F4Uo/lUlnWyM7 skIoLwyaA1Vp6M61GHUs+asJBzTGd0Fms6mwJifZf+n8oNXehTs31+SmGDoU vlQ/sWvMUI4Xu+qKC1M3p20JXLtsweYg/77aXJx8QQCk19q/3/iASod2N1FL Sd7OmNDVXnEh/qmxG6tyU6ital8iKp1KQT+EW7vri5qLdiZFBfgsnL0jJrCz PKOvqRSn89KjhqAM0+moG2gs2bM9MilibWyQ75YQv+biDGopwxGhji8umYqB HmQILWV9NdkRG339fOZlxIWUZcb1NJbh6oTO6mXdQSv+yqBEnjYcQ1ftRJ// WMiroRw1FN0y1akoEHrbdRLhObB01eOO5I55kB6afcVRls8jmRmHU2n+TVCf zlypdGHpYi4hdvxmqttNCY4U+C8wKkfJ8o14HE6jBbfiJlR+OPR8DHqh3ztq qCYdnRvLHjW5m+jQLNaxuFUX9iXtWYsmHKw+8aGyfJUpULsKeEO54XXKWG3l eAYorI3/pKxVoMVwoUkHmYMBHtWYi+6ygDdp1oVj9qjfgf7yvh8UVl4IKCx9 TLrkz5qzI9yv3Ej+z9Dc68zgUfYn4yJ56Ge03R09zLqZSziodKk1FSJDKIiA G4w1PQhBg3dZInyVdwrLnHBJ/mpCQG3zbqtU0tddVBQN/8zwbYqbH7Po20Sf H6siXZriFxqyA8C78g5Sm4GD2vYapBD6WJ5GwYA9lEN3Syn6cjnPL0ug3MDe dL/KKNe8QPu4xd+k+f3Sn7ocaUNRFJQPazPxnep8VX3kWkYCbd5t4PHq91JV CmZUsUdl+VdEuiUv/SFz1dT8jXaN8Qtx/6IwgiqSYKimAmTvE6sjyzB0nRy6 kbmo9UriKC+UPao9ceGe9bbhHl/kBNh171jSv3sVBgxBBTEdhkJSqqMDaGPA joGlV6YsvRQsvZygwczVJZuc96z7Nc3vf+n+U9u3e0J4k8MXG0pdet2NYG/0 1P7x2zAMjcluyFUEk7dgiWWt7t/lk+L7v80Lvi4McqyLce/LXEV5QVS+DV7X WCB5v/EJjfgr3tfYAdhh8iNMvhFw5BxVpRrqc/VwPYdXR287RD9W/JXm30I2 JxxlN9Ehy0Ons8jnQSqKxSwnHSxDLklKFb3omRegpjOLldSOtZnn0fJHMcaR Y7seDnQMKJ8D34Nu9pDYjpk8yu5kWvMKtMp5F9AB2KO46Ij9BffgHM88yiaZ UY96l9B5Mq4TxjuiotEBe8yLYs960KHzbkTr1BgJ9uHH9nhceV77Gm3+L1JQ HaCvC4l0yCfkea/SHWqmpceeyYZyvZR2eVLlLj3MJYQkZiO6Xos2owwxW3qg aGaWbqPKVAuJ3kv+agLAgNZZzjN5uWFa6F9p1sU04zxyOMNgf1rf9JP7bU42 2J/K/185rbgYN6lXPUfxTrTbD2eR+qDijQGOFFugRJrogiZ27z8hieW453j6 oN2pA7an9tmc3G97CmIX56KzLqS5N+DAIuLbofmqumq7/S20VYLn3LkQAptL HyG3y3Fe5gBDsTsN2J4yaHfKkEfxl+bfCmXXzT9Tzgb8xR7xA76RUMfoVO6g VE+K+AZLb95N5HiWQVl68KjpJw/CoxTFUadz0Ou+8mmK+hFyIjXpeh+dMxKK mAPX19tnU/hXWHpzr4dHOZ4xgKV3Cj+89Azq0nO9jDyuRaIV9jWOinjB9omf lBoL5aJuWSLtDcD5++oXyf1a5OfsPw6n99ucAo+y42B+KkctxK6Ft9Oyh3Ev IHkuzNvbpofaWUfQgr9StAKobi+OrvLMl53u78JKAoXV2y4828CvUJOJLqkF t0IjxVyFoVrycHk4/yZcPMkLUZJ5cZehEoJaSihlPm36HHeXzGmo3+EHsqHW vYpTDP5XhN4Be1uprZySPczUeXXQ43U/sgUuOTHjUtzVpxzz1edAh3zjO0ga TdNxMsKjeIsM/QQXo5qLrf2yY0NvBxpats0EI8rFmnndyeZ48n6A1r+O88GO OrGPJDjANubjMML/WSw9O3Nx7Mpjexy8NOoHyliOQ3yBgzkbyoAr882lWID5 4Ud/Z/DQTzAWXXkiejxAHZu5QJb8lW6hDtDpakQhXBRFCU4U8TUt/TPNv1kZ enIYMVL+Ei/2xXdBtjTmZ+yMnJ1yaYmrvvolZ3jrb6+BemFRJCJS0Adg8Nyv gSDDIXMzVYWA90Sv+/CdcfYgKDjdasyDzQdEDvtHggGBmnfPuj2ogzh9DfsC AptzbwADc7jrA/wlNuOSe2ntK7R1OmWvp4rt1FoG7V8dg92AX5D9gTOBXYux 9Pz/Djp05gWHPlhUB+iwoRbeDmmL6B/RRc9Lr6lA1/cKFY9qr8LSywvFUXLI R+j09piCU3ibw8Qo9ijnc7H0/J5FWEtfhhZxzlX03YiF/rQOiBtXJmPpxf4K fR7Pe2CKw4mRslNxGT73emiWhn4KuQwk8EXQ1RQ6NdURLM5f8YrobsEOXhBl zs6rUV1YBZH7GrZFzuQ5XId/g0Btxq6Gg5Orda+i5BG3D4RDB9e2HK45sHAK ZJY+mdGB61hMRuNNc+cisYfHVe3EK7BHmbHz6oA6+kRyOIPCv8bFig5hh5uo AhQ7F5PnfeR6uQU8SlF58riGVj6Fk0FxCxmOrpx7F0YhneZgYokYxYWh01kU Ow0HZJ111n7hMaC/ixJdwfK5XmoRQ6nJ1cI7MG5P6FuE/T3IEEricBZs2jAX o7uw8sNwMNSQb/a0U/JXugXv+y2l8En/59A/6XiGoht5HGTofjN5UAfROp6J dGvF47TbH2W4jkse3vqz1kChYu51mOY8NCb7mN+Ke5NgKP5O+1PJ5VJ0zkR+ jzqay0Nxt8gjYKAHnEzWKugJLPkDaIQD588e1kpDY6BPhEfNvABNI0luIGd0 rIvFS680gaJ/pgW3k8tl+wevHzk3GzYUf//s38PIW2yoKhXlpy4xOABOJn0p Wtbn34p+IXvFUNOOOaKhJg31M2DpnYZW0jWvUOSPmHjS16lPCotfin2gfi9t +gJH+bMuUsZkHw9v+Q2POhYPDHUK7mXMvxnNpbkhSoySsD4sz18NoOW1qQBH q5Ygr9QnL4zqs6i5UFS9NfXmYEUy9qYl9yjLytz1zjTlzteMc7B+cTBRJGqk 6m1TWq8XDUUhSxiKH+fJOBmMmYq7AwMCDudVORku3CK+xVGLhaykJgzr38AV xfpsa7/z0YLL/9YKqKy7T8ECsZCtZl2Iw8EkF9Aygl4q4RKMPSplPo6MLcEb qw9n9YH/wv3EZmFjVF8XztkjvqHlj1nMoyahnHG7Aqeu4l4B4M+Xa7qOairc bP7e7P0UViiaFmoyQEqY1aMkf6VDcDLGO0JzMWVvQKPp4rtBQx0FBY106yxa cAuun+xciNNJ1NH6Ihx4QXGULovHvfuAt1ALH11K5nAaLbqd1rxIe1ZjEG13 M0KobqAKITbmo39jyzRa9gjNufJorIRS+kTQMsEfUprXvkNDfbFYXAfxMimM hsDj2lchH+pw+tHYyuksUFgb3qZdS8Aed+rrJi/S+z5qyKOyJIr+CT0JLhf/ NhtzyCzC5nhceV7zEu47c2HY1ai3m7y9HSjiiqLRGur3t6F5Q0eTbp1OLhdS 0PsU74B0oqVUyMJQX7A4fwUB2xScrlqo+WroCabizbgmICiBzNVH7W5K9aYZ 5x62Nd1cVc/sK3GVPn2ZqKuvqQAJ4ZqXyfYkCxbRnK9ylPN5CNcH2irFcypO LDkJjLMDy3d0CYDxz9wbaNULtHuVtd/5aMEfMW9tq19QihRLehQnGAFvoiGk Idfa73wUMFBfG0QDvB8gpzMt6E5sKLfLMHqV005zEw5agH/hxgIIqHJaeLjW dPM8k3DQrHJ9LWLyHpwq1+fi+r/lyKthCqtws3KLsNOMybnkr3SInlZcqEn1 wgjaWRcc/VBjdYXy37U/BVtk7K+QhhA06To0DDgy3jIdzJXjWfsmqx5duFOk Rx1OwxHS6udRRzcWihf5DwcckJWjc2PxPVAfGmMzPKRHT8Id1Y3vYqg0h1Bx m28Phqqgsv4NEFD2Jx99SqZKj0K64Tx0QecEYi6MbsB5UWc9Rf+X/J4Bc3XU Q43Vh0tOTiQW3k4rn4RCcoeO7scNXRmIJt8nkFXanTKmYK4Oy55xDoV8jO76 llJrv95Eh2X5q8F+RZA8EbqslrgaMJK/KghXRoq34D6CcOxxVwMVhGGU3tEw wyY+My/A7hDvqBzZi2YoDkc1GbTuDRTRh7vfba7H5lhafCcuKmIWoWgZQlc9 7npHfAPyyrxqTgc/LpfgTDBlAe5MibjxFcfiVHTpQxZfevxwhpDsjk5L4cCR vK2cAv5hnjl6R3om0Yyzyf1qyt6IvjjhKj7e9coTac8qdJtb2p04BnJiFvIJ euOFC+b8C/OHW51B5dvNM5L4yBlCfiROIrqazFgXS/5KV1A1r1qK0dwS+T0y JQcz5WOzLkTStWU6upU42RNvqR4EXrmtZWgo2vA2tk6zJGPqBQHOJXYuQj4P UU0B04kDoCgUNeQgjQz5FO2yTmeZx6Nm/x7K0omzEPmFvmg/DM4eu5twuhf9 E/n8Gen39DEf/dgol7+WP4ZZyQWRIBKFSycOBjwqF/crA95EhWKW42n10HDB bUjjS7big9BDjOrFJ569gba7or3T8QzziKiwRyHp+hjTCdur9auuJgAsyF/t V76KNLco62Ge/HBqKqY2AVWwOByFfQWl38OpOJrx4UzDeTI0kTrrBUsPcMuy DZIULpeYb+bgER+3y0VVwapMwT0vv6fHcCRqyt7HmWf4V2iN62q09pubDv58 F92FqVIaeJT71bTiL2j3EgvDx1iL7raIkNroZxKYfC73sjeKl5xD+coFIiez LtLCUE5n05wpVLxFyaOEyjlxvNWGA19LN1+pD/8r5lbBkvyVrgDyqoSyA5TO q/N/S0bGxC3S7iTM5eFNOT9MyFOeUWirQKm76d9K55UZs9ZJ6JHgXXLty7ge 1Sk418cxubcdhvK6H+QVNMHM51G4+fUAJHO58NEBmgpxySvofWz95j2b5qXn dBa0jyp24AhDbBiovxPtB+vfgADm0dwZPJxHqbdT76ZVz6N8GBCwD2QkoHlV RbWZELR3u9TMt3U44s04B/eYMldCKULCSrAkfzUAne2WUssqXx2QoIaBCGop E08Fq3Y3rX0de5yFlK9GhSmH02j5o7iP3yUUzc4Rld0pa7UyM8Lyhpqm9qq9 jvvOwt2dL46BeJfPg1pYSX0C30PbSWuZtd/cFGA0ajcOMedcBR5AAytBBes2 SpmrtP2Lcz+CDVWaAIUEDXqK1MfuZAr7N3RjuFASC30dEJ71fQqXwTUwlP3J uEHA259gExsN1NeNDagwSqPjrdxQpOW1WWY8MJX8la6gtsFvc1Yu3VugE971 MsS0HfNwtV/cU3vetqB3kYG5LZY4csUB67nowMkLoepU8ZL5keioBWe++b+W ktb0uA4n0WnecF3hctRhsEdxtVi+jZJc0dlidhkHdGEdD1om1Yuqd4lU8oyG AQfEzYW06VPy+ZPZevmGH7Y8Z8JL7sX0HC4P+0XWoBtug/f8g9k6r0Y+tifS utdQZNVnCylzoQtYkL/iXLqxAMyMNqerlklQtQCH7rJ4ZZreeRZvlRl+Ftym pAdpIqUHnQ1QUYuz0+KWpfrgKPBayDv0tln75U1Ehi+teAJZkzaG4of/uaTZ ylgcccB1REsJWuwsqnw18lGnU0X/BGcWSKKWN+hUT4r6HofI2rgTW2nFXyBU K9ZcAA7m3U0Yw+FykRbNtOrD8XD7bMiFsT+LAu2PtyCSGYMAZb5mWslf6QgG qknDTPlFdyrtH5aZa+x4Jgbo1GbiYoKgJU9fB3px05fhdM/uJMvEtEnKHYFn KeYXgWkZRmEkLqJ6PWDOzqtRHsW7zIa3cD+xIc/ab3u04IDc04KeItfLQDVY aJecOZnm3Ujb5+AcXLgbOiq4pi6MgiL9/FssGKM4nQh6n7bNElnfyUB97WhN 9HlQOZi2TG7vcBoGp+5ZC0MJd0NHF7Asf1WXjemlWvJXZdvwLwrEX0FyoZOK o3FVjZeDZvzVvJvR+oixzuLwV5zy8e/Mtb+dJpcHpyltoi4X0+qXUJOKteXt 8sQxiuulGhmKH5+HKHYatKQEQkcNSomQT7Szkioryhsrb3mcs4mC/i6Qk0Ef YDloYyib46B/vvZVnIAIhIFuTNPzfRIS9xo006oPJ5xxdspY5wZrv7/R0GYw 8Sj+qjCCSuOUY2XzHMFL/kon4GQVM6qiIL40+8qjmTZo1KOMrV98J+UGUtVO kVKvkeisHzpG5HzVUob6HVR9Ft+F84u2ClzBEw5o7e6iTF/MhZl3g2Xj/4q/ UpKbYEeHI9FRi10+/Guwu5YTa2V3nXUheuGaChV9J9GAvsdOSvNBBWTRY8Qh fadPMOBSyFmEBqTWLcW0/h8YnGE5sVbb49ACl+SqKG+Lc3SoI1iQvxropcpU Kt1meWlWCx6wWhxspdZKULgcujUjr/iZczXOO/jfFUjPsLkYV/k2vmNxQfKR D2doPn9GEiVQtwwjcSZmJ4ER1cpQC2/DR5O9wdpvbgrqcxA31r2unZXUJ/Cf qKMFOt7q66DIH2n54xa5BHG4dceF5JJ7qXKHtV/eeBjQU1S3lzzv0zSY251E mz6nHfMx5UoUcH3XVEB1u3HrX6P0QHkKotBMaybqQPJXOkFXIwTDExzJwQI3 TUY9XEQHvImZoYJez+Gqdv2b5P1nCw/QUSYSLr4L/HZNpmCnh4zuJvAkEd+h fLbsSPHf4brrojtx1x4X8YRjGwiC4XF26Eu07KapKMGufBoXeEVKKvaBS8XO OuiZuF9t4RP8SQiDblfgXr9gogQKhm4OrqYFt1s8mNueQCufotDPdDXdUhxY kr/qoYpdVJKAwxrtDlijoNMiEH/Fv2pzMYRlzDJFwviHa8PYqZS5QiT+ivOB mKm4dKwlf8WP1/34jHpaxTmJMFCCAzpALNXef6hn3o3QXN2zxtrvbgpqsyCn tuYlTd2Jn4C3MIaA63dR0NeBARPeFpBcONwz/Rik5fNvpYrt1n55o8HlQ1sl 1aSji0NLd+JlHvw+ZNwEYkQ5K27IwzBZzdqzh/krTKEyT04u+SudgJdtqhfG 9WrQ2u10Ni19mCK+pd5W0WgZA4ro8u2gGni712CQx9wbKdUb48uFK6JbylCJ BL2nzNGzWJfakEedhXGE8Y5oAhGo9hlGTiAMtegOLfZK7/shg1AYae13NhUG NIg25GKA9czzLD57nSss5/MpfSnIasypFwoco9ijklzJ/RqLu5PKsfs/B6U+ wYK5HmBZ/qo8CQ1R2vFXyoyholiRDrZ623DhceciTVtl+HG9lMI+xzQcgfa7 hhwK+QzCyJoJy6gPV6M1aWjBEkKwgn9JTvZip2oxeXDkM/tKpLXpy6z9/qag cgeWwMqnNXUnfta8RFmrkBuIAg5Tge9BdUEz6TkcAp6GW9VlCdZ+eaPBCUx9 Dq7QLvo/Td0JUqKvUsx/RWJEB/qgolO1k3I15q8i92kdm+EwQvJXOgG74oZ3 aOlDFi8Mp6mzCE+hZY9Ra6lgN+O4MGyrhrbzjHNAy2gQ2WZdRBvfBTMvUEqv ojRO6Sl6RJMtQBElCP4Q8/VaxRp3YsCTNBvZozZHY+xRC+9Ar5pYGOzHQd7e tTjR0yartzsZfH6aN6b4iQXe3GN/pVXPQUNPA0PZnkQzL6L8Tbi0KCksbaE7 /iqMiraItNlxYVi5i1IWWoG/Uq+cSP7qN58l94BdbxWIv+qn2F80NdE0yV+Z 8oC/8sflKVEA/upfGD6oHX/1u338Vby1X95ocPZSl01FMVbgr9a+qih7iMNf De7jr7Tuv5L8lcQoGLC/r3ySFt6qRU+RejPO6z6M8OPaUIikQgWU27NQ0mp2 WWDGuZDiif5JvCaQ/FDa9BlSR812gfVvYCRTQ66139wUcJHIqcVWO1ANltMp Gvk4nY1MFYxot0gCdAO9lB+GZNXjOo3cyfYkUMfxjtQs1pZkgEeFfUHe9yma PJY3lM2xcKqsVRgFJVA1rQtYkL/q70GncXEs5Wh4f5DXePEWkbyotx13KFKX aM5fXUbhX6HvSyBb8dYc9iX5P2OF/iuRBgYZUEfH/opU3NKN6yOfOVdDRzRj hbVf3xRw+bxrMfk9o6k78bP2FUjPcf0uCjhMBb2P+k67MDUJo4hm/16w/qvG PJy8L7pTU3eyPRHFCy/5pkJrm8BoWJO/alTEDCV/JUHKnbge3FBzuVSjClp9 RByg07FvgI69VgN0bI6nmefTqueVWSeiqDco2LUE5JXrJdp51NKHaItoA3Ta a1D+YICOVjcFhgbofA22QQ7QOdLSO448/0hrhBug00udtSC9HU7X5DBCeUQc oKMLWFj/agek+bTUvyoIx78oECfT10kN2ZSxVGv9K7crKPpHHKUJZKvGfIr6 gda8oDV/xWG8Phtjd8TIoBT+aqsNGoA1uA0x/Hhch4HXu/2s/fqmoDodDWOr ntPUnfhZ9zrilUBHpb3tFPIpEnLLjbce/UyiGeeRx/WQ4hQFhkFMvancTovv 1tSdbE/E6IR4B8j0iYLBPty1rE6zAn8FlZgeyV9JALxX9rTCCZ3P1zSvmHs9 uu7zI0RSdmqroJQFtOkL7QZAcwXqeCYUkrvqRSL6GDvmYvaZNjeY1IfT1PBv IHcvEFrLqHwbWrs1s5L6bPqM6vdADl0UcJ0Yb0/rX6OZF2i39BbehqXHe7RA 6O2AUy17RBFJ1ooUtT+Fon+mTH8MNJfQEJadP8iez9WHlvMHi2KoTLT5g+3V lL1ew8JQeTB/0AFq2wLpt1tl/iA/3n9Cn79ATsVInIU5R1o29S24jTa8TdkB 1n5zU9CQQ7ly/qAR6OugqJ/QXyfnDx4JqshqDkoJLd0J8wc/w1glweYPFuI6 kvbzB9mZzdRJK/kr4cHbOsfhrFVaHyDOuQrdMpkrRUrAeMHG/opWT00TsEnk 9QA+I2iFCXGAqCDBCbeKNCP6+Jl/M2Yb7Vlr7Tc3BXV78AuveVnTpccPZ6pF 0dRcZO33Nxq8Z4V/Qz4PajpAx/USaIVVJFv75Y0Gb+tcTddmWmOAzoeU5Abq TEJDWJa/qt2N6kMz/io3hEriqCJFpJSA0/iuJtxbn6nt8Z/HdZQyD6Jz/DGJ Aq7OuEbjSk2zsXrTjwUF5PtXnNSYaWSVRkj2wEfsfJ52HsVbRti/4VECoaWE Srfimr9mVlLb10M+wZQ6gQ5r+jppqy3mS3KY0mzpcUK+4nFI0wgDA3o5mosx X4xjlGa3d7lKiv6RMn2RvIkC3vi4Jm3I0ZS/yt1EhbGSv5LYDzTA51LGcs0H QP8eF/HSfERKVtEA/xPEGzVugPe6D/eyuxrF4a8MFG9HDqdq3dG36jna7W/t dzcFNRlQnFj1vKbuNG2fVhg7lSjobafQz8nzXk0b4GdOxpjR8iRrv7zRQAN8 KTgHjRvg7fZphbXI7VtTWJC/0v6ANTcUehp1e0TiZIBBqkjEbK9ZF2pX8iz8 PyqMhq0EmpjQ3UhFUbTNWTsRaYfTaMEtmLwm1pwgBqcx/s/SXK3EHvnx+xum ulSmWPvNTQF/rFzyR36nrDtNmo1tjgfbEDsVIpMChSkustKX0eafUXZp406c +a98GjlbQ461X94kDFJvC9TU3S7XjmbnMJUyHwc3PW3Wfn2jwalmRz2mzOdH aMVfhWCaDGfj/WZjDCR/JTx6Wqk8Gedi2g/QCf0Ex0wC9XXXZ+PkxfcJTXUJ +FlyN1XtQGUqhAApZ9S8s8dM1dRE0xRGdOlDYEQFQsV2SnYn3ye1ttXqF8Cb 1WRY+/2NBqeLG97FjVQtNfrsTyaXS6DRJwp46dVmQSJg4R2aupPtiWD1o38S aYCOLmBJ/moADTONeZQXrlGCmrcJHV+NBSJJCgAG0G7LHqF5N2g0/oaXm9cD oBpESQlUcAyv2gXNeaezNBrf7HQ2ZupFfi/e+JvcYHS0LtZQR5pr9uwNSHEF AtcO3c0UO01p9deEbXA4nWZdTAkzxBt/wx6VPIc8rtXInThGBbyFpi+BmvwB ZfxNyMdYetp0dNieAKEwzsYb8hRZckHA+053K/R78yNx8V+L460QKttGValm 7HiR/JXwwADonZB1ssIAaOXOr0D8FQZAfwq2wQoDoHehmhAiWeVfcqBP4a8m aafAM00dAP0IpS+19vubgoodWAIrn9LUnfhZ8yKUWgVSJucwtfGfNO8mbQdA n4o5X2LxV3V7MYzDCgOgX8HxruSvtIUF+StcpmijjlrQodrwV/kRIGQ66sTY 5kYCjdk/IqhqkBjwP+F6Mf6t7kaRmq9ISQz6OtGC5Xa5Rtmm2xX4XFKFUrlX UZ2GX9v/75oEcOVOXMQ3yDB7Wq395ibBgCfVi7wfgFNpYCuPa9FWlLnS2i9u Injp8dZcnkCL79LEo5SRLvEOuI7aJdpIl/4unCmHfIyzSw2WHpr8b6CyeDL0 i3O/RgHvPn3tmBecr0mHdl4YDmcbC83IG0v+Snj0dUDJOd3HCgN0or6nVE+R UovGPIr4Dre9tB+gw2UpJ/bCxDcDBuhwVWtznHZWcufU4m+CpRacpqb5aJWm jnjWvUY5G0Vq7cYAnU9oyV3kqFmYmgSxU85XxRqg01yE20yapanqwyt9w1sY QSjQAB1dwLL8VV83dTVjoisSVAufseZtosLNuA3U3SIefwVlp/m06XNF79HC RzaOZ5DnHyj0M1D6wuQDKgw42CpLIs/7IH9q6buW04+luTfieChPKJUwFU0F VBCBIxu7kyyeRHHmz6U6R++hwfSiYW8AWn0W3KrFTuepqoSFWfudTYVBEcbc DS0453Mtfn/E/mRcpt65GK0RveLciVMx0ANN2nhHmjPF4sF8v0rYLtGCuZJt 9nfhQl9xjOWHFAfjeKuxgForzXhqI/kr4cEe2FxIu1dqrn91JcX8D7pbAulf cVIR819a96rm+u3345/ubhYnxBlw+MKZtp2Ghpp3Iy4xZa229rubAs4odvvT 6hc1XXr8BLyJQY2N4nTL9LZT2Jc4ZnU8UysrHYMcbP4tVL7d2i9vNDijaKug mlR0bGrpTlxkBb1H22aKNJJJF7Akf6WAC/+qNCqz9BTCfdIWXD4LR16RkkQ1 FVKmH26CWPpsS8TJgyPRXExxDrTBwlMIMbj5DFr+GM5o0AEiSuK0D4N9OFne 5gxVBEuPLNk/edAgnqFIkfXI3oD7jxrsdBv/SSVbhZR55Lja24Y7/r5P0Ixz Lbn0jiH3q1Gz8K4B3lg0j2JDtVZAdp6zTUsPcRZx8uBIDPajiOAqxtIimXmb MJuYS2CIX5nNoyR/JTzYA9kr8oK1HgDtcS1td6OcQJGOxrg6S3SFLoFmyn4c PG2Ox7W4rnqRrtszts/WIvUa+Sy+C02/ucHWfnNTwFVPUTSyRy2XHj/BH0GK pL3K2u9vNPo6KeYXSNpqNpIJx/fXQVGNDSUMFPUGzq45e9TSnexOxvWTNG+R PEoXsDh/xZuOqoKVH2HBM1bOTmt2U0MBkgGDaPUOKTcpelqgzuH/dwinWKhh xvYEmnUBOJmizYIpt48EZ5slW3FDx/VSJee0THnofC75PU3R/8HkQYE6/IfB q2BwAMlM6KeQP7VQ3GZH5dC99lXMPhNIDHMUuhpxgXfLNNQUlmNmZl1IC28H o9haIZLO9khwdM3eSIkuOOoFzW6JpTeJbE+i9f+gLdOR3IoYzEmdC1BJ4V8i /XM62zIepbT3u1+NK0iVKeJ1qakwDFJnI1ZEYYwFKazcUFyCqE7D8dagOXc9 yV8JDzQBdlPJFoxvQ22olWDRgtsoNxDCpwLRMpwLcfznyKyZiDRnF3Om0Po3 kR6LdTCdsYyWP4prAtoYip8Vj1PiLBQRAqGzAbKNYV+BpdRmdpXt8fDe6J+w QQu0aXKM2rmIwr+CIJU27qTyxoH/ovq91n55U8A5akct7ji7XKTdNWf7UynJ BZsIekQltIPF+SvcIuxEB0tRDFgmi2SnwZDYaqukribB9rhRaCnBrZON71rq eGtomt6/qLtJpKxpFFQVrJKtGKHoeqmldj33KZTgiHtAAh2PHoyGXMoNovWv Wypuc3o541zcg2irEE35aiSUtrHMlRihaDl98oW3Y2nnbLT2y44B6Cwqx1W1 pQ8imFhi2AT6Hs+k7XOoNAE1i7gY6IGyB+fJFprYyIbiGOX9AEY4YV6JmEQf KXNeejuodBvm4VpIZCAvDG20LaVm3/Ukf6UTVHFMexgUlhYDdJRnyR+oNgPn 9QKlrL3tVJ1OuxZrp3WPATqPYECweAN0gijoA1qk8QCdAMHYBt65+rtwdGh/ ikZDLe1PJefJlOCkHN+Ls2lyGcIb2fbZ2g7QeRMyboIN0FHCVPBHkHDX7Eo4 /0NZ/qjfRWx1EBmW56+UpcceVZuFY2Kz3yIcOlpNhewVtHfEiUgHo7uZSuNo 50JacDvUpM3IzPCP4g1i3o3YKbhO7+8UtfkKUFSwGvKVW4TvksMZZqbZeYOb fwuteVnsLjUVXQ0o3BJnoQ/EzHW0otm+5B70d2WvB3llvqn0mkPhrziGpHnT utdxCm/e82XH08nlYgp8D3Qoe5TAMIBTai1HVrP+dZp5gZnTAJvjINW7+mUq jATbILBHKY3H7FH5oeT3DLlfowgbms9Qdieh+SrkY0pyo5ZisWMUxnX1Dt0i NH+TtiIsULaN2mstIYwp+SudQMvJevxPuFyETnsOcWJpkPI6bavGncfZV1iw 9X3kgymNn9KOueIVhhXbcQrDHmVpE01TzjJsT4R8bm0GddRY+81NAe9cg32U Mp8W3YEGdQ1sxWmwz4OYbYS9QJylx4aqTqe8UDCi2nDsnAPH/Bdyah211n55 E9HfRfFO6OHXwqMmoV6Yex0VRqB+F7ctRExowV+Rsvo66rBf54db4Gg1l1rK dOI5HFRrMmjdGzhYN+MtQpvjscp8n0CN0F4tUtw+HPo6kXNm+kI53MGsI2W5 KufaPN5BOfIT31CMki3o+V/6sDmtNP0Y+Gfgv/DD9TF0gzPknhZwfUMXScxn K5eLoU2RMg8bqw7CFL9CaTwUj+ffYuYSxu5kiNvvXIh2Ph1AVQzb/F9MnDTn vdRJkFWZcxXtXY8tVVAZwwNgQAtWZ735m7RzQ3FqVptloTZayV/pBK3llOxB oZ9ooezkcDqq9eAPcaorHgxgZiza+j7y8biOdnhQXoh4bfDNRdopO9meiKbl rTZIWYUzFGPPWgwEnH+zFrZSVcLYowSDQZG5yEPru6WkG0aFqdMgTVCTKV6Y 4owoaxXmSc252uJWElIlTD/QiL9SE9SuJpAzFckgncZ6zBpMuZvQecVu01En 5MzBQ8JgoPYakN4JjhhV5j5lrGQ7mKtzUWmGfwOZX/7hCEfi0zKDfah3qnZS 9I/Y+2aer/S0jy2qO54Bqag1L6NFrWQLDQqoHX1INJdgSD2nN8v/YoaQrmpe Lb4bUuRpS0Fe6ePS92A/KKziWPSWr3kRDZBjH/Uy4xwYfN0bOIot3wan1UGY 4ldoKYWaUNRPGHjkeslYEyqug/gneN2HpcehT1w1p1HgYM7VRGEkpXqhC2vB LWO+dzOJnM4EHbrhHYr9Fbklb3w6YERJae3g4qs+B35VEGUGLSxOMIq3Ylht SxmaryzToib5K52gpwWcfLIHjuztzdoqefDjehmU8XYuFK+nCDDg3DDsK0hJ W7RXjdOMGWejMOTUhcOCcC2m3U3Qb+QorYGKu/s1OCLBLMte8+r7aQTe8VMW 0Mq/WZaWUY9c1fS+NtPa72w6+rrQe7DpU/K+z7JTCNlQblcgwy8IR8EoHCOK XrU0UJRef7SUzMXwY3siuh1if8Fil9AcWvFXCoa7sAqjx3rMytlpQSRUl1sr xNvafgMGEMgcYM3S0G5/Cm5Mc5lZk4ZDRh2UzyPBJU9dFu32o3k3IfMc44Gg 66U4gU12B8UnYhpwJBioLAEDXlc8riQJY8gTVM2rkE9A9eij82okejvQL80+ 4P0AKKwx7m7uU2jFX2nXEmXwmYjVyuGBLqwELL0ld481SeCs0uEMivye0pfq cIALp389zRA2HHtDOxLLy3GDIDsAA+X7u6z9buaFAXMNOEMojafCqLFqYXGC UZ1O9bkWPYaQ/JVOAFHNDirerIzwOMeydfSCW7F+K3aIyjx31qHzk0s2i6q4 Q7n9KtSG3Y2iGorDTrqPBaQbDnqWPYr9hcOmoGCPqt+L+49I4C229HDwehJF /YA6SFCxVs4hU+ZT2BeWVXG3OR5XLAPeQm0lKDg14kzS/+8YWGZRmt3+VFRV RZvBV0toDk35K7ULi0s5ziqrM6kkDvcFTEtTg5HZcuFcnUZ1eyHvw4FI0BlV h4UB+3V7LTi6nYtQ9Sx/DP1FJggOTFJOr85BS8Oq53EXOHMlljP0kPVlq8E+ sA3sDIkuEPn0+TOILJsTTKim1ZuVnLWufYXCv4a2Q0WyMsVSX0Qff+6tpdCB SZ4LISbOqYamN5oSq2degKlJa17GlSjOYHXTeTUSEOvroPIkqNSyP6z4K3lc j1za+Mu8vF3ypsZ+uPJpJEsZy9EjyomHqEn4YaB2YdVkYumFf0nef6a5N5hW 0XCy6nKxsvRew4F+XjAWsnDN6r+JwQF8+qVxlLWGQj9HQJ5zNVoLjPcoG0Xe hJfeskcwETXZA4lll7hl3eHR34PVx4GlIYdKE9FABcFM4zOEYMoPw8FWRQpV Z2CkS2e9RXV6JX+lFyitkrW7sUI5bluo5FH3hXWv4xoCDl7FzDFwaLiHdq+C TKulbhFOQoWOw0QP/HOinlCrh4bOEO5gj7KxQB+Iw2k4Ewn9TOzDRP6IO+ug csYJvNsVFll6/HCmwUs71Rs1o6CX7nnH5xw+axUqO5NSCOMfzjSczsFhYpqP wIeJA32odhNmgIWzkAoWx705V9GSezGStalIb8fTgkBb/kqFgXo7qat56C6h SYruuSFYv1wPcl7aI5T05dHAQJ21WB0Jzrgb7nqJsakCfxtX3O7XgNLZuQhb Q1+ntd/FkuCo3t2MBrPN/8EQFofTMSTXyCjEFTfX3X7P0t61uI6qv5JwFFrL sIJif4VQtmmHOJOQr254h9KXCXyAZSQGB1DOcPKZ5EarngMVbLz+NqeUMy+A H6YsQHwTa8yNqeASrK0CdEr0T7T6eUz2Md6jOPtafCdmPu5dL6TMgmkwoBGo eAtlrKAVf0HmY7xHQd7kLCy9rbYwdW+bsAWdccBQqk7lkCsdgpkmSA0ED6m1 t5Shj8vy/IDkr3QFzipzgynO1lKdRTPPxzHZVjvx8zEDpMLXvIIswhJF9PRj IbGVGwiBCKFTMs696/bgIIyzAkuQolybL7wDh5KQ1uyz9tuODUWbcTDq/SeL LD1+fJ9AOleeZO33HBs4TWorh2TKnKvNPBQGzyTcTHS7gvauUwboiMzJ8HIo jYNC8rybLOJONsdDjyXoA8EmfuoL1uCvlCYHNPjVoBCu3YOj0vJk5bw1Fh1Z BVHKE4mH/5P/sDQBYac6DQdkvKzaqpAACD2dyigYFEmQbEWZYS6Kl8B/4iYg 5wyL7oSypfvVOP9yvYzcfk9zpuCA3vNe8n+G1r+BY/pts6ggAgmAoKKOxoPr FHanjmoIZmb5Y2rGpi/Q57Dicdhk3o00+0rEZNjqctB682+CoZY9SgFv06Z/ U/IciFFzmsFVj6BHosajpxWvWRRLuzwh6h78Ia1+Acdei++EbjmX1UMedQUM xZHf6z5cYg14CzlY0mzK3kA16YJPGzQC7AaDAzAUhx3eAbfaoLmdS4/lj2Es L5uFkwe3y9E+BENNAb3MS5K3M84rwr9CM7+q4zR0yK5f/gp3vlrRA1kYhbuE Mf/DWfCq52Aozqvn3kBzroSh2KncLiP3axWPuh9Lb8PbMGmyB1pDOQEQUWbB NBhAOjUXY/mkL0NwDv0U86mXPowhnhyjeLm5qoa6AvGKPYoruBV/xfLkpRf7 K1LKkq0wNc6OdexR+4ZSddbj/Lc+FxlC5Q6sxOKtCFyF0XiG0oPNIKxK4pAh VKfiRKypEBR9dwv6uCxvJclf6QrqUJjcQEjxcDw3IzNjeyLOH5c+CMVIzmlF pxp4ZfHa5J2RAz7HK/PeuJxxLlIODnp1WWihFDolY4/qqMMet/FdbIjTj6Hp ZjIUDjXOwOlPgjMCIKQ1BT/U4NDNCXzkt9gNTToI+83H5SKkbTFTsYlwySk0 OE3qbkb2HvUjEiq7k83WAGlzLHh7rpsivsFmKrq0Ji4IlFD1Lgr+GNchzalB N4lmTiaPayjOgfYGgKOWsBKsw1/th9KL1d2qXBnIo8pdkAUo3YZ0tDQe/8uJ a9UucDhNBZB/13/P1SFhGMrnuX7hujjiOwr+CPqZKx6HxrvXH2npI+T7FG14 Cxd5OLLlhmCKrv5uCxoD3Cisw93SjGWUOBNNC2teRr3MEYxt5f0AUlNcFVRI Bi5/uOQZ7BV+6z8acApaCXoTWehUeJRKO3jeC4aBzbXyaTQRRX4PMajCCOwF +rstaAxwo7Ad5PnedZB2D3of/rPir7gxt/huUH++T4C2YgNum4X7p3W7dTJn 0FRwwtBaATIq3Qed24H/AvfCS8/7T+R5H5yKk+21L+NOpbr0ancrIz4nHjAp oAtLryAcpxKbvsCpBC83jk4co3weomWP0bpXKeg9MMZc+zTm6vO2oDHo70EN yH7VVIR2rMqdkGQvS0TJVhKPUWicIdRmQeGZ00gMsB7QUk9A8lc6REspJbri ZMeMswgxGuYeVFLspYLeXRoFTpl6WtFQyoHd/Wpz3iJ0v4aSXChno35ONBrz oCYd8CZ6NsylJs0e5XopzjVay5RUX3xw3OYUggscztjn3Wg+tuF34KKDPgAv rY+CSJVuqExBnjDjHLPR7LYngjZM9kDdrQ+PImUWIWdQGKNwpdnciZfwvJtw Wl2RhEROaIJdcFibv1J7sbqReXY1UGsVtVZibQ4/bZXotuqsB3nV16Xzro8j gLMFrvWaS5CrF0Sgyz3TD9I6qUsw5DRtKWX4QhQ0Pwx0X0MuKm6hmz+PGphZ 3wEKi4udsm3oF9qzBrrQad6KrbwoYyXG9XLlWBqPE5/2ath2IoYgA/W2Yn3V ZCoydEG02x+dIbsUK6X5IOyzoQojUSQ25utQgdxIqHcJ2U/q9sBncgLhURkr IKSwczEMxf+fDcVLkv2N6+iOWp3MGTQZSi8Wx+rqNBgqeyM0Utij2FC7POFU 8Kg18KjSOIwXb6+FTMEEhNpfxPtaYwGOJPI3IWnnAM4m4hjFHsVG27sWnlae hOXZ2QAOZyJ6lML1QcuiFfqobTUwWksZNZeBTsdTgYSBVxx7XW8n9XXDSpK/ khgLuHYrS0R+tfZVUMpc1o2Fc4Dm1Y0479g+B5wMZ3H6IKJ5lXE23lwIQenI 79Ev6nbZGFisSeB2Zl2ES+iR32L7qM/Wz6kilzack6f6UOD7tOQP6HIxXuDi cMyV/7Pon+dagN1VH4kZ+CsDChyc0X+Nzv+xq5TPvhIdztH/QX7bkKMT/orf orsFZzo75lHEt7TwNrQD2YzBoxxOhQjw6pco/DssPdwc1EuhzTGEM6jCKHT7 r3yanM8fUzCH5tWVoENjpiKt5VQEnbT6cCohYX3+SkJCQkJCQkJCKEj+Srdo rwbfzpWv41ljUi6aeT66K2OngcTQR+fVSKhdWOXbwfV5jUELiwtD+1NQGGYs p5It4t+vPBRaSkEOcB3tPBms5lF7lOulaJJPcsMVzt42a7+VuYEDiAGcL0d+ Bwrr6MkrZcr2soch/sAepT+SgQ3FDlC3G83/824cg0dNQoyaez2WXk2GPmVI B3rR2pEyn+bfgvOIo77sbHM8LhQEvgexZdl5NQ4g+SsJCQkJCQkJCZMg+Svd guu4mkz0t2z+D65CL7oLOodG3v+yPQHSiPNuHpJDzPSHiGt/p046r0ZC7cJq Lcdw1WR3SFj4PgGCxShFLKXnyuksWng7hAsivwUnU50GLRFdFoY9LaCwcoJo iw10LbwfUKZ4nGQU6cff5nQ2Buj4PwtiZ9diDBvqbdMjI6rcImzMx9Jjfwj6 AL4x52q8vpGclfO5GDW44gkK+RQdSiVbqKVYh/wVLlC0ofc4azXu2wa8CT0Z t8vQnmdMjGKP4u9UVVujfsCNaV567dX6ubQ7EoYBeFTFDgwDDf8SLXnzbyEH 4wwFj5oMlT+/pyFAnbKQsgNhKNl5NQ4g+SsJCQkJCQkJCZMg+SudY6AHnENx LIV8DFFxe+OmyDmcTovvAnmVE4iiaUJUOgbqqofK0/bZ6A5yn/LbtaHaczX7 Str4TygiNhXiFrDuDaWKpbChon8CM+N0llEya05ng+za+A7umFftmiiSF20V kDdMcAIzM2fKbzOikMc/ljyuJ79naLsb1WaC3tE9BvpAtmevp8RZEKNzvVRp MfotTsbpTEyUCP4Y+v/12fgJurmue1gYwDvxy26ZhhlnrpcZ1YsFtasbIQic shCqrd2N+o9R4kDyVxISEhISEhISJkHyVzrHYD+kZpqLIW+YsQKFT8S3tO51 Wv0iLX+clj2CK07quJNlj6I3Zt2rFPYFbf6Zdi6iPesgh4gRtAMToORRhmU3 5EKyb9cSSnTBAMHgD8n/7+D9lj6EQSeefySvB9Dv4fc0yL3A9yAjw0V3zkZc rOtqgoSd7oHhHd0wVGEUdFnj7Cn8G0zjXf0CRNKWPgyP8n4A5lr+GBR7uNAO +QSG2q5Mf65KHZrJqKHKn9XQ0woKqyweV9u2zYLMWtB7MNTKp7D0MBfmXhiK lx4bipdk0AdgBZPcIF7HfthRMyEGxKgCrXV7ob+a6gVDhX+FMUP+z6EJjZce G4rXHTsVPOopWv8PLL3YX7D0OKyVJoA3RtuV7j3KAJqOX7Y4FsE5YQZt/i/I 87WvIEYtewxW4tXHHsVG4xi16nlMZYr4BtKFaT4QzG8qwrVB/RtKGEj+SkJC QkJCQkLCJEj+agLBMIDyp6UM8513+6H6i7PDrbeoHyBvFe+ESTrZAbio0l49 0Wuc/m5M6KhKwdycZA8ojcf8D9Rf9M+0bSalzFM4qwR93oAzHoZB9OY15aOv I9MPxMtWG7A00f+Bubi+TpmPTi30EdUoHjWBnQpTmWpxazLTF7cC4+xBPoR/ TTH/pXgHSlmA8UMVOzCoFwzDhIWBDP1Qqq9IpvTlIF62TAOdHv0jmL0EZ9ox F/Niyrahj0j/DVdHRF8XhsKUbKG96xGjeLmxiTiex0zFDV9eehnLcaeypXRC x6jxDclfSUhISEhISEiYBMlfTSQY0KXQ3YLmmdrdqAFLt1JhBBVE4kC/NJ4q U9EFwRU0NJAnMNVASpdRVxM6hapTMb+Yi8TiGCgaFUaDtqrcgRJbLQz1pwlm AgzgENhhmoogtlaehHG07E6FUWC0VEM15oK8GvKoCexUvPR6O6ilHBrjFclU EgcT5YcrhopDbwwvydYKkBIT3KNoEC1GreWYwF7OSy+WiqIVj4qisgRQfA15 yoTBCS8/PtAHmr2lhOr3IkZxAGcrcYziSMXxqiIFBmyvRsCf0B41riH5KwkJ CQkJCQkJkyD5KwkJCQkJCQkJjSH5KwkJCQkJCQkJkyD5KwkJCQkJCQkJjSH5 KwkJCQkJCQkJkyD5KwkJCQkJCQkJjSH5KwkJCQkJCQkJkyD5KwkJCQkJCQkJ jSH5KwkJCQkJCQkJkyD5KwkJCQkJCQkJjSH5KwkJCQkJCQkJkyD5KwkJCQkJ CQkJjSH5KwkJCQkJCQkJkyD5KwkJCQkJCQkJjSH5KwkJCQkJCQkJkyD5KwkJ CQkJCQkJjSH5KwkJCQkJCQkJkyD5KwkJCQkJCQkJjSH5K2HQUUO7V1HWaioI p4odNNhv7V9IQkJigsNAHXXUVEg7F1K8IzXm00Av/lBCYgJA8lcSEhISEhIS EhpD8lfCoCaDVr1Aq1+g8K9ox3zq77b2LyQhITGxYRiguj1UGEUL7yDHMyk/ jHrb8YcSEhMAkr+SkJCQkJCQkNAYkr/SFN1NVBxLe9fTVlvaMs3YJ9WTOuuo LFHyV5ZFRy3tWUd71lJeCJUnyQ63Q6N2N+32p+wAygulhjxr/zYSFsZgHxVt RudnTiAVRFBX4wFflfyVdaHavzSB0pfRriXUVmntX2hiQfJXEhISEhISEhIa Q/JXmqKpkGJ/pfVvkO2JNO13xj6e9ypFYjStel7yVxZE7W5a9watfYVCPqGk 2dTfZe1faFxitz/5/50C3qLQz0BhSegbfR20+b+IPIHvU8R31HggYyn5K+ti oAd8e8IMWvYYLbmXKndY+xeaWJD8lYSEhISEhISExpD8laborKfcEEr1Qu0f /NH+Z9VzqP5mnEeL/o+WPoxSceRXuTxpr6aybZK/siwkf2UMJH81oSD5q/EM yV9ZFZK/mkCQAqQSEhLjC1KAVGLiQvJXmoJjS3s1tZRAzKo6bf+TvpTcLqe5 N1DQh7TVjiqSD/gq14wD3VS1S94ftCwkf2UMJH81oSD5q/EMyV9ZFZK/mkCQ AqQSEhLjCjIBk5jAkPyVpuDA0ttG3c3UXgMia/jJDaY5V9H8Wyj8a0r2oOaS A77a1UiGfqpO3Zc+fUlx9jgEzNlA22aheNk2k1IW4E8qd0Kv5mD6vacVJ4aF kbRjLiW6gCKLd6IdHpSxHH/ekCtIxDMoBGANXmTvOkpyo20zoCQW70jJ7pSx AhRfYwEZBmGEznq8V9ZqSvOmRMVKbLRtzrRrESo+/lJbJb5TRVcDqBj+gbwL zLuJltxNK59E2a7qj+1aTK3l1NUkxrkG59XdTVSZTJm+tGMejBPvgHfnz53/ kzc4fvfOBnwnDNWP7S97A6X5KPZ0pjg7uBM7YXYAVWzHiw+jIQciSJu+oLnX 04LbaMk9FPCP/SpteSEQEOvtsNZ7mwe83GrSKTcIFmBrbLWB52yfDX9ja7RV jP5+LmrY8fir+ZvgQmyEdB/8RbYhP/wlTi2aCsTwnFFg3yhNoKxVtPIpUOuL 7yTv+/Hpqx83O1V12pC51PTJ4XTKXEn12VhfvGS2KSuOXY4NUpcF30BoGgF2 v9YK5fvXYFXi++3gpdvdaW8ARLd4CaPJYbyajl+nIJIyfOEtHHv5yQ9FgE1w ggMkuSoBOYV6msEysfPkBcOpeD2yY2yfgxesTke2ORIch9mjimNxoqH+KNhk Fr6frVoQgeinHrCy9TjW8Y8N/ZSW/4Xcrya3yyjog6FPh/8Wr3TeDni7kbAY JH8lGKQA6XiGFCA1BlKAdEJBCpCOZ0gBUqtC8lfjAmXbyP0aXB7c/D9KX456 52AMH/+F/ZsivqXon2jD2zTrItw6nHk+LbgVadXOhdTXuZ+WGUZrOQrJyG9R h7pcSnYnk9NZ5HEdCp8dHmDPUBONe4D9a6eaTIr8nta9io41fnfbkxC33afQ ir+gauYIzwGfjcB1MdeVq18i7z/TrAtpxjlkfxo5nw8jr30Frwyub1+C1JBD oZ/TyqfJ9oRD6I8tvovKE6mp6BCGHYfgFL2pAISe75M072YYx+F0sj+FXC4G NRf2Jd69IRffya/f3wX2gB3J58+wp/O5ZHcSzbyAPK4FN8VVc/n2/T+ZN9DA 92DAQ6q0hXyCzIpTUKGhNkMGvQcLsLfYHE8zzqXZV9K618BiVWwf/f0ZK2jF X2nd67TpczSksRF8HiTnyTRzMlbliscp6gewHPCc8crDHA7sGwnOaLTj5XPw x81OleZDtZno+Rnir06D1/ECXP86yC5em7zi5t0A18paje/sO5DbZPerSAZ3 uuZlOBXbmeOSy0VYy/yPbv6Z6vfgdxi3i45fhwOR71NgjTj2hn8NB+AA63Q2 1pHrpfiTlPnUXEw9LbTdjYI/RMh1PItmXkhzrgY9zjlPe9UBP5MjP3tU7K+4 RT73evwou1Ng/zlX0dpXEfZ5F1ATVJU927kI33bwp8OW5JXOXz2YcZUwHyR/ JRikAOl4hmyANwayAX5CQTbAj2fIBnirQvJX4wLG81dcEXNp4/csyKio7yni Gwr7gjb+E9XNsodQd3NJOJJp4dKpNIF2+9HGtyngTQr5FB0U/LfCv6HQL3Bd cd2r+DmVKQLwM1zrpS9F44ffM9jBuWwM/ZQivsbDKWXMVLSTFUWjslNproJw sHwc8/F8i2fTv2nDu7ThHViMf05H9VD/g9p/lehKC24HybP4LpASvGtwrsvP rsXUUiZM/1V5Eno/wv4NEwX+iyK/w8MfMTsMF9SceBdGKh1BRIO91NdOuYHo 6GOXiNxnJf67IR/DW3jTTPYYap6hff1XoZ+jDF9wKy2+G9mmaiJ+RO+/6mqE WVIW4K3XvATCKuh9kAC8RgLeBsey8ml4CC/Djpr9f0vlr3weIs8/wq/i7OFX 8U4wSNQPSDKXP0qJM4Vk9kb2X3lcjwSJ35Gjh/px82tWpR7Qf8UlIfvM5p8p +geK+g6m42jDRlv2KP4WW6a5eOgnc5zhv8juFDsNKzH4Qwr+GGszXF3LX6LA 5Li0cwG41s56axrhCFD5K/70OSDzO/KS4f8M5+XzFX7/1c/DbhyotzlD8HCL DUX9iKgb9uXQV1c9B9qTV2t3y34inSN/5kqKc0DgitwXuNgsbMC1L9OaF0GI qTYZ2X/FudPsK8nlEoRE9dOJs6O8TbL/ytKQ/JVgkAKk4xmSvzIGkr+aUJD8 1XiG5K+sCslfjQsYz1953YfD/bk3gGTgurgyBRUQryCuWZzOpoW347pc0eb9 V3VaSrC4uMZRu632rMZXuU6vzQQXkeBIrpfQkj9Q6oF/a3yCK+WVfyPP+/Au vk/i0lZxLBQL67IQwHkr38N/skXhr9pQu/EbpXmhAK9OQ5Dn7yxLxI1LrjTZ Vj4Pwgiclw6zUvpIn9KX0tKHUDuveQnVbkMOPm52A3YV3t2KoqlkyxCTwG7W 04qGmS3TKdMP38BW4i+xN7JHbXgbTXoBb8KAbJlh6DV9aiyAKTa8RdOPASmx bSblBKFRrTACBCbnD7YnwqQZy7ESh6HyV+yQ/NXVL6GjpjoVdERHDdXvBfkw bEP2UhFhfPo0/VjYwet+XHxjT6vYAXZl0xdYqhzZOMhUJJNBWWtsH16SeSH4 wxnngqhJ9YT78Q9HN1c02Pgld4PnCftqqFdwHELlrzhvcb0MvXYb3kEQ5vXF DsBukOCERjKb43DNduXT8IR4R8Qr/iq705ZpCPiOZ1GWP9bmcC0M/soPMSo3 iEq2oom0uQg9kPnh8L3Fd8Im4V/v/xRk+mRVSP5KMEgB0vEMfSRgloZeEzCJ Q0LyV+MZMgGzKiR/NS5gPH/l8yCKxJVPUc5GKt1GrZXUWobSOGsNls+yR1EZ ZfoO/QSoqaQi41r1HFICLqBqM9BC39WI+rohG5IsXBZxwhDxDTpPxm1Kxq/T UgbCbf4tqHm5bNzuTrVZ1FRMnY1DAzga8oboF8Mg0lS2TFMRqsW63UhZ+Xv4 rdlWbOpdS/BzVjyhSDwl729+0Ef6tHMhaEze77hkzt6AE2d+8e5mXM1uzIeh 2ET8n4zBAervAW9TvAVUHn8DW4m/BI/ag16jFY/TxnexexZt3v/z9Zo+FcVQ zC/os5p/M0X/B37SkAOhMLZP1U40tHjdj3UU/CEaYIah8lfsk3Ovp5j/QdGo rRoe2NcOyRT2osV3gxOLd6DSeOu92xhgfPpkeyLYlZipVLwVntZWgTWb6Ydu tMV3QqCpIBJ8Kf8VXs7pyyh2Kuzp+wTIPa4N+fuxQiupIZ9S5lH0j/gSB7TK lHHaF6ryV8v/gl7EBbehpC2Oxfpqr1J60jajK2/mZPL6I/k/i/ctjcPS41jE L8t5pvf9uP2d7I4/V9cjqRE7E4G9IRdVNhZvM25/839un4Mo7fsU/I3DmgqZ PlkVkr8SDFKAdEyQAqRGQAqQjgVSgHQYUoD0yJACpBMekr8aFzCev1r+GIJY 6Gej1x0HqIA30bcQ+B4WL7gXA34OL1KXS7DTcYw6WBKzpRRrOfJbMBWcufWN 172vpwUm4hzJ4TRadCeoAC6Qjzrl47DPpTHXzgnOCErD1tYHf8V7PfSa3oC5 DtZrMh7FMSBkOCD7/x2c1TD0yl+lLwNb4vcMFhF7xSjwZhT2Jb7K2TWnBMNQ +St+2KM4Xx0FTqvWv4HOnKD3scmKCOP5K/tTsdGz23D2PvTVQfA55dtwIdfx DNq7FjUL7/78MyO+BYHMCy1mKr5n1M+sVS7/ctRyngxecXwmUUP81eMQ+OJP n6PKSHDesv4fOGvw/hP5P3dAzx4p9SwHc16nsb8g50QX6BHBFqjaiau+XveT 6+X7uVDJX1kVkr/SCaQAqTGQAqTGQAqQjgVSgHQYUoD0yJACpBMekr8aFzA+ fcK9sBdQ9XCGMBKcAHDo5n2QY9eW6fgqZ0RtVRi143weLglyzsDLnKvFkYN1 eNWvUOqvRXdgZxzFiY0fdNbhNGrrdMRYn4ew+2MMxxH2I4W746qQgznXd5wG JM7EUQ4//NbBH5HHNeT5B7TZ8BagM/5q7zrs5pxm+z2N8J7mjWPiki3KEUzN IT5iTq05OcxerxwNz8LBgWqlDe/AMZY+iNuIbKVh6JW/4p2OaxC2Gyc8nA+M QlkiTqb4q7yXsS8N+56aPq16Huwxu+gocMIZ9D5WJWdf/LmICOP5K06o0n2Q 8+xPkwxYg5w4ed6Lr2Ysx1/nZdXbBhm6ebeAcmHv4npwZFDi4Bb6KezJUWvG 2SiCuprG4yyqIf2rJ2AcXixsh5HghRb4AS26i1a/CHUvruYO+GobrX8TFuAq mIu+4bE1OBItx83TvWspdQlcLt4R8ZzjdshHiO24q3gBbikiahkkf2VdSP5K J5ACpMZACpAaAylAenSQAqSjIAVIjwwpQDrhIfmrcQHj0yeO7aGfgZAZxa40 5mN5ciDiTTP6P1ja/PDGl+aNyvHI43U47rldjgSsp1Wb1zUZbRXY1ziMuFyM 3Ok3f08uqzlvrE5F8OH3mnMVij6b48nm2P1v7XopIk/y3P23JvXBX3FEZffg bcvmOJr9e/J+APzJlmnYB9mF2mtGf395ImgrDulzrwc5Y3cyDDVsJf4T3hlT Fu7/fr3yV9tm4TCdP/0dcw/BA9TtAVGw9jUYh42J3j8ll1bTp/X/QALP6foo qF1bnH2x03IJIyJMkF844yCuyYDtu6UEXUMOp6EMURu2ef2yQWacDRc9cmji 2MV+21I6Hq82Dx3/PYk0m2MvR+CR4NfkMmTZI0irYn8Z3WPGX+VUx/MPCFAc alrLhv6cTcfrce86eNqSexDuHM+EsNhIm/Cf8Lobvokp+SvrQfJXOoEUIDUG UoDUGEgB0qODFCA9JKQA6eEgBUgnPCR/NS5gAn/1AiJSssfogm6Yv/J7lqJ+ 2sdf7Ub+wDUgF+Yb/6lQ97/sP68Zfrba4lxj5E268QZT+avOOjSeJc8BZ77u dcQutknM//Dw+3L65HE9Lb4HxxPb3fenSfpIn1orEG95P0I/7ef7hih9iLfm hGfnIuTbqqgCb/GckSa5IR3a+C4YmKgfcIoaozgJO8yyR3HiM/9WHBoOQ6/p 05j4qzdh7cPyV28h7c/UPX91JjIiHGYNX+w9kL9i36tK28df/Y1mnodi0Pcp mO7goDR8yMircuQB2fjByPb16J+GjtSHsZ+/+gQOw1nQqK8O8VffUKIbugvY VmwWXrxcy3DlwtYO+gghndcjgvZUBCv1KgF4wmDoYrFNZPpkVUj+SieQAqRH hhQgNR5SgPToIAVIDwkpQHo4SAHSCQ/JX40LmCS/cMjxN+CvflL4q2f281e8 J6Z6gr8au2aUdQH+aj5e3Ej+CrcpP6U1r+Di0vp/4NSP409v61BGBEs+T6v+ PjpN0kf6pIIDMpfMBeGIz+GKcNPcG3AY4fskCKvyJHwPB3mO6vzKM86h4A/w zZyHs237Ffdjh+FNf91roxWf9Jo+DfFXr+JlTeWv2BqcrBZEjP5b+/mrp9GW LCLGNP7GgP9sKR3BX6WC0WIfY4PMumAolI2UqBUIw/xV8Mcw0SH4qy+R/oV+ jgOCQ/BXHyI6gb9yhYlUMVXOizitYkuGKVo6vB45gxroxk0TNDlEDvW5cSHJ if1gn0yfrAvJX+kEUoD0yJACpMZDCpAeHaQA6SEhBUgPBylAOuEh+atxAUvw V/xDuCDasw4NogtuHd3WLhbQTxWAzgTnc9Gb/Zv6V2wrv7+j6YVjfpwDWPeu BhhkoA8hnYOM9wMIelx4jkyT1BR03WtK+uQmavqkQh2C05SPDIq3/kw/XP32 eRCZz4a3h5SaSuIwCIlfmXe3eEe4UEcdyCs1x87fBGUPv7+NThiyVsMJ+YeE fop+G92A19T8W6FWFPXj4fWvXgTHxRXHsByo/vmrTmQ4a15EBsjvaBb+ip+A t3EEv+51uOWonykKzMxfDeDsmOsRro5dLqadi3EBp7MW/4raZ1W6lTJXYDEe wF/1otE0cSbK6iV/kOmTxpD8lU4gBUiPDClAajykAOnRQQqQHhJSgPRwkAKk Ex6SvxoXsAR/xSuRv4cXGlfcc69XUoWl4/eG4JFh6vFfRTICFxeP22djC9s/ 3MegiCLGKwOIb0UJeQB/tQdhcP0byK84cAmaPh0ORZuxD655CbyBeh8wNxhv yq98yAOvPWvQy83J6ij+Sv1z/lshHx/QyC06ho7/nkXyeXCqM3z8N/9mJFrD mAj8VeyvqCm4vBp58V/F0fFXHJ34Ry17BA+btCZd87cyB8zLXw32YYWmesNW vBdwuBsJjmDZG1DdQLThQP6K/zzJFf/Q4rtl+qQxJH+lE0gB0iNDCpAaDylA enSQAqSHhBQgPRykAOmEh+SvxgUswV+pXA0XjBz6VjyOa8K8YItj8XM666ir HodBHJc4x+BFV56EMDhubxdCfqEUldr8mxFVOGqpY1nQWNWEKzYtxagf6/ci VhsG8VLoKFakdTjvUkda89PVAEIvyQVxyeNaGGRkmtSQS5u+xMEEpwf8T7RV IkaxJWFqw3hsoD0keIPj6FqxA8bhj5hfQVXbSF+GXZ5fbekjYDJpH6MV8CZS LHXw94DSnsf2ZEtutUE/0rybR/NXucHK1KTX4Wlcbnc3K4yE+nfFsdLB4KUR +yvunPJbs1l4RTTkKTNxCnFffkh+4XkwfgfLL+iYv+LVwYkl506rX0Qloo5r 4d0fa2qADP1Hw1/xck7zwaAcr/vgXdvdqCAMEay9GqGJ3Y/XMq9u9k82KQ76 xyXMzF/1o+EqZyPkGriyyw4YavVnG/IS429QWcSZ5x/AX/FTGE27vGBhj2sg p8M2RNRS1uP4bPvXESR/pRNIAdIjQwqQGg8pQHp0kAKkh4QUID0cpADphIfk r8YFLMJfKcD4G2csVaezae6NaIbMWI7asD4bP7A0HvvsjnnYQ/esGb/pk4rq NPAAvKGrAqrq+BuVtiqOwebFKVOJMv5GTYRWPYcxqZwy9bQotV4/GAn+z3Wv 4aq422Wj+aumQkSq4A9AkXGkYiu1liqKmi2o08ctuTcKnNKEfoocgG3C2VFr GTXm4rOOc1B2qyeQIfBnTeqgZzdYY/aV+CrnnKrboHt2PQxoeyIyilH8Fe+V nGWteQnumuCEfYGT2856peVvYJwqPRoDdqTCSPSVTT8WVkpyRTLA2QK/764l vzH+Rs/8VTe4Td7lOZf2fgAMSUsxDpE7G/bpMpnOXw2Pv1n4fzhu9n8WmUb+ JhwO1u2F+/FazlqNYHhwwjZ+YPb7g/ympXEIPk5noVqsTIGteEHx361MRmB3 u5zsTjqQv+rHcf/etWg3dZ6MYM6G5RyMv6oS7+KuRxEg+SudQAqQHhlSgNRU SAFSUyEFSA8JKUB6OEgB0gkPyV+NC1iOv+ppodIEyliJfWH1i4jkG/9Fm/6N HxL+DYV9jf+/+Wd0uXPwH+fqWO1V+D3jbHF2wDs4x3POcyK+RghSTxB4h1I1 voa+0w7ftv4N/B+OMEmu4FuifqDAf+EOHceu4I8OSJM4UnFc2u4GzQHeE9lE kd/h+zkDaS75LcGHcYPcIORCmz5F1sSW4VeIUJ6N70CQgV8nY9nQjS0MA0pA bF/6CA77OCuId0RyxVbaMh1/svRhHJKOSp/Abq3DT2YDcqrJP583Ef6x2QHo /Ri3Ch6/CbRaFdDOBVhi/F7sNkEfwIYhH+I4Zs3L8Dp2ITZdx4grALpPn3g1 lWzBL88W8H4ArqUuitipIFjaKkGYmJo+carArsI1Tuw0tDtyaFrzEmIX4tI3 sCQvPc4fon+mOPuhUePjEOZNn9hWXQ34IRzKOIaHfAxV5wTHoajFP4H/ZP3r uIYzMn1C0lVGNWnwsSX34BA/TIlaMf/F95RvH783knQByV/pBFKA9MiQAqSm QgqQmgopQHpISAHSw0EKkE54SP5qXMBy/JUKLjNTvdB6tPD/aPZVOMS3PxmH Pm5XYEvd8A4UJnmfHYc9oiMBUYVW1Mtc4vFGBnXBc8j2JMRtth7vYqmeYMhV VpwjUkE4rLHkXoQjftx+DzaG49WmLzCQgkvmUcd8SDjLsQmufBJ6qg6nw1DT fofsojwRw6CFaGbI2YgylpPM2VfSjPPQNcT2mXk+7iXxPs6bOGeSqvYsrm/3 Ii8NfA9DOvh7YKXL8Z3gEz7CpsB2HpU+ocm2DMJinH+6TyH7U8hWkWtgY9bu po5aa723ecDelb6Ugt7D9VLn83HYxz42+/dIqvmVDx4npPv0iVMg/kzrsyFf yTv7zAvwiU8/BqsjzQcNBrxlm5o+qUAXViqydE7Fh38yf5vjWTiqhrrvj3C8 cXs4aGb+SgGvRy54E2YM3SLkh9cjB22O0nF2aKr0uO6A9ImUVcwmjbNHLjpn Cnq3bI5DVxv/62zwtgpNbDFBIfkrnUAKkB4ZUoB07JACpEeGFCA9JKQA6eEg BUgnPCR/NS4AQsAd4YVXUHX6oRODjhravQqHLwXhUDcaxTV1NSKI8XaW6Yf/ M+qYj0MWB66iGLREckbBy3moJckNx4LZAdSQjTpxnPMzBgMOp9qrEKXZDpzh qJ0J8Y7IrHgjq9pFjQV4CzZOdwsCGu9Zu5QpEttc8LK7FmOPy9tEmf6UtQan VyNHWquio5yYZfrCUFwSqoNg+G+1lEFoS4j+q/psRFROlTkz5I9Ytc82Z5B7 ewOQhA/rXLE9BwfQT8XRmOM2fw/HZw7FqUtgXvYl9kb+P5w7VYwIy5yBs5XY buw5yXPgSFumY4dlY3bUUu94nQBuJNqrQWHlbIRHJTjTFpuhZcLp5bBsxUiw Pdnx2LDsV6M2UFKKGvZV/ir7YW2mNm9gfvBe31mHm4NpXnAS9ij+xOMdEFL4 E2+rxKLj9cJu1pA3JNk0DP7P7maQ5/z96Neq2l/LoAurCvkYxx/1J7MvxTlQ nCPkZdgJi6JQB43bDiKOsQWRlOE7tFKGh1ar4Nfkj55z8rxQKtkKI4z6ak4Q liS7TVkiFpQKNR0tTaBdnohaeFwh3pvmg8P68mRYhu3Ma1zV9CPFjLwk+at7 1kE3mK3HTsuW5ChXuXM/WyhhAUj+SieQAqRHhhQgNR5SgPToIAVIDwkpQHo4 SAHSCQ/JX0lISEhISEhImATJX+kEUoDUGEgBUmMgBUiPDlKA9JCQAqSHgxQg nfCQ/JWEhISEhISEhEmQ/JVOIAVIjYEUIDUGUoD06CAFSA8JKUB6OEgB0gkP yV9JSEhISEhISJgEyV/pBFKA1BhIAVJjIAVIxwIpQDoaUoD0MJACpBMekr+S kJCQkJCQkDAJkr/SCaQAqTGQAqTGQAqQjgVSgPRgSAHSQ0IKkE54SP5KQkJC QkJCQsIkSP5KQkJCQkJCQkJjSP5KQkJCQkJCQsIkSP5KQkJCQkJCQkJjSP5K QkJC4v/Ze/PgKrPz/tNTqfw8Tn4zydRUTZzETpytykucpeMlXR7HjuPUePsj TipVsR2nJk5C2924m25D0+xbs4odWoCEEGJrAWKVhJDEJhYJJAECxCIk0UEg BAhJiEWi6QbmU3pGp15fSZdzda90r3S/nz9UV+8973nPec55z/O9z3vec4QQ IiIUvxJCCCGEGGQUvxJCCCGEiAjFr4QQQgghBhnFr4QQQgghIkLxKyGEEEKI QcbFrz75yU+++OKL164l6l6ZQgghhBCJAXrpxS4aGhqePu3/5miKXwkhhBBC eOLiV1/5ylf+6Z/+qampKd4lEkIIIYRIaNBL/9RFY2PjkydP+p2P4ldCCCGE EJ64+NULL7zw7W9/u7q6uqWlJRolJoQQQggxXEEj3blz59y5c9/5zne+973v NTY2av6VEEIIIcQg4OJXn/nMZ/7mb/6mpKSktrb28ePH8S6XEEIIIUTCgUaq qak5ePDg17/+9b//+79vbGyMJjfFr4QQQgghPHHxqx/+8If/8R//sWbNmt27 dz948CDe5RJCCCGESDjQSDt37ly9evWIESNef/315ubmaHJT/EoIIYQQwhMX v5o3b96cOXMmTJgwd+7ctra2eJdLCCGEECLhaG1tRS+NHz8+JSVl9erV7e3t 0eSm+JUQQgghhCcufnX06NHi4uIZM2ZMnz49Pz//yJEj169fv3PnjtbCEkII IUQyY2teoYtKSkpyc3MRS2+//fb+/ftPnjz56NGjfmf79OnThoYG28dQG0AL IYQQQoTHNoD+5Cc/yee2trZ58+aN72LmzJnl5eWXL1/WWlhCCCGESGZszavj x49Pnz597NixkydPXrRoUfST1Z88edLY2Gj7GGoDaCGEEEKI8NgG0F/5ylf4 3NnZefTo0fz8/Hnz5s2YMWPq1Kn8XdWF/SuEEEIIkTygf9LS0pwQmjZt2ty5 c/fu3VtWVoZqilKDPX36tLGx8bvf/a42gBZCCCGECENwA+i/+qu/csfb29uR aoi0L33pS3/6p3/65S9/+YUXXvjoRz/6ESGEEEKIZAL984UvfOHFF1/8/Oc/ jyKaPn169GteBWlsbPy7v/s7bQAthBBCCBGG4AbQn/vc59zxR48eVVZWcjwt Le2dd97h74oVK2bMmDFFCCGEECKZQP+sXLkyPT09NTWVv4cOHYpyzasQmpub X331VW0ALYQQQggRhuAG0D/+8Y/7Svaki8EsmBBCCCFEgjCgQqi9vX3lypXa AFoIIYQQIgzBDaAXLlzYV7KnXQxmwYQQQgghEoQBFUKPHj2qqKjQBtBCCCGE ECH0tQF0aWlpvIsmhBBCCJGMaANoIYQQQogQBmgDaCGEEEII0T+0AbQQQggh hGNAN4AWQgghhBDRoA2ghRBCCCE+MvAbQAshhBBCiH6jDaCFEEIIIaYM/AbQ QgghhBAiVmgDaCGEEEIkLRJCQgghhBBDAm0ALYQQQoikRUJICCGEEEIIIYQQ QgghhBBCCCGEEEIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBBCCCGEEEIIIYQQ QgghhBBCCCGEEEIIIYQQQgghhBBCCCGEEEIMPk+7iHcphBBCCCHiQLyEkASY EEIIIZKW/gmhJ10MRHmEEEIIIRKceAkhCTAhhBBCJC2eQujRo0fl5eX79+9f tWrV8uXL09LSVqxYMWPGjClCCCGEEMkE+mflypXp6envvPMOiujAgQMVFRUo pYHQaRJgQgghhBBTIhFg7e3tqampEydO/MIXvvDZz372y1/+8gsvvPDRj370 I0IIIYQQyQT6Bzn04osvfu5zn/viF784efJk1BRKaSDiVxJgQgghhBAf8RNg nZ2dR44cycvLmzdv3owZM6ZOncrfVV3Yv0IIIYQQyQP6Jy0tzQmhadOmzZ07 Nz8//+jRo6imWEWuJMCEEEIIIRw+Aqy1tXXOnDlvvfXWpEmT+LaysrKuru7x 48exkmdCCCGEEEMOtFBtbW1FRcXMmTPHdzFv3ry2trZY5S8BJoQQQggRQhgB duTIkaKiorfffnvGjBkFBQXHjh1rbGxsaWnR2qFCCCGESGbQQigidBFiKT8/ H6WEXtq3b19lZWWUa2HZmlcSYEIIIYQQIfQlwBBLc+bMmTVr1rRp0xYtWhTD R4pCCCGEEMMGNFJKSsqkSZPmzZuXlpYW5VpYtuaVBJgQQgghRBiCAmz+/Pn/ 8i//8pOf/CQrKys/P//hw4fxLp0QQgghRMKBRtq9e/eaNWv+67/+67XXXmtu bo4mN04fOXKkBJgQQgghRBiCAuxHP/rRpz/96a9+9auHDx/WkgtCCCGEEL2C Rrp8+fKhQ4e+9rWvffOb32xsbIwmN07/xje+IQEmhBBCCBGGoAD77Gc/+5d/ +Zff+ta3qqurteSCEEIIIUSvoJHu3Llz7ty5b3/729/97ncbGxufPn3av6w4 kdO/853vSIAJIYQQQoQhKMBeeOGFF1988fvf/35TU1O8yyWEEEIIkdCgl77f RWNjY7+DTpzI6ZaPBJgQQgghRHhMgL344ouf+MQn/vqv//ratWvxLpEQQggh REKDXvrrLhoaGqKZf8Xplo8EmBBCCCFEeEyAfeITn/jIRz7y+7//+//93/8d 7xIJIYQQQiQ06KXf7yJK4RSrfIQQQgghhj0mnD7SheSTEEIIIcRzUfxKCCGE EGKQUfxKCCGEECIiFL8SQgghhBhkFL8SQgghhIgIxa+EEEIIIQYZxa+EEEII ISJC8SshhBBCiEFG8SshhBBCiIhQ/EoIIYQQYpBR/EoIIYQQIiIUvxJCCCGE GGQUvxJCCCGEiAjFr4QQQgghBhnFr4QQQgghIkLxKyGEEEKIQUbxKyGEEEKI iFD8SgghhBBikFH8SgghhBAiIhS/EkIIIYQYZBS/EiJWPH369PHjx/fu3aut rT137tzhw4cPHDiwt4vCwsJ9+/YdPXq0vLz80qVLDQ0NH3zwwZMnT2Jy3Zs3 b1ZVVZWUlOzevbuoqOjIkSNXr14NKdiNGzdOnTpFkfbs2ZPfxZkzZ1paWjo6 OmJShp4wmFAkyrO7m4qKisbGRuwT5iyK+qSbp92ESWPJBqgKEeFK22vB2tvb r127Rus7a9ANEqfwQohIUfxKiIHjww8/RC3cunXr7NmzlZWVhw4d2r9//54u iouL+RdJgxtta2t7+PBhvAsbN65cuYJZEJmIChTmhQsXbt++3e/c7t+/39TU hLWdUEG7Hjx48L333gtzVnjx458mXoSXlGbhgoICZ+Hm5uZ4FVUIYSh+JUSs wPF1dHQ0Njbi6bKzs2fPnj158uTXX3991KhRb7zxxrhx4+bNm5eamrpr165j x451dnYiz2Jy3TNnzqxbt+7tt98eMWLE2LFj58yZc+TIEfetueaTJ09mZGTM mjXr1VdfHTly5CuvvLJ+/fra2tqWlpaYlKEnJSUlM2bMGDNmzIhuVq1ahS7C PmHOoqgfdIN9ehU5IWkSRAWZnSmPK1swPnnt2rXjx4+vWLHCWYNuEMMYphBi kFH8SoiB4/33329qajp//vyGDRtWrlyJwpk4ceLPf/5zNMybb745bdo0JA1u 9OrVq3fu3EkQGTD4HDhwALOgMxEVKMzt27dfuHCh37nZw9D09HQnVH7xi19M mTLl4MGDYc5ygalexY9LE0YgxZfwknL//v0hFr548WK8iiqEMBS/EiJ68Hc4 vubmZrz8zp07Fy1aNGvWLCTWG2+88UoXI0eOfO2118aOHTtp0qT58+evWrWq sLCwoqLi4cOHjx8/jvLq4eNX9+7du379+u7du9F7FOnll19G/qFJsrOz6+vr W1tbo7x6T5Cd9+/fz8vLGz16NGrTCSEqnpubW1NTE+Zc96gLMx4+fLi6uvrW rVvuW6QFmSNod3ZTWVmJ4nrw4EHMaxEpFOPs2bMlJSWUau/evYcOHQoOp4pf CTHMUPxKiIHg0aNHeEy8/7Zt2zIzM2fOnDl16tQxY8a8/vrrP+vCNAwHERWb Nm1CMDQ0NCRnFCu28ava2to9e/YsXLjQCRW0KzoWAYmSRH31ehZf3bhxA0Hr hBllCJE34QVSvDBJSU8LIykVvxIiAVH8SojoQTWhuGpqambPnj1u3LiXXnpp RFiQBJMmTVq9ejWKK/o3+MLHrxobG9EVq1atclcfNWrUhAkTUIbvvfdeW1tb lFfvCa4fAfDuu++G1Bq1mZaWhjwIcy6qhmRIU9LPnTs3Jyfn/Pnz7tvHjx+T OSV3eWJDRFEwxhUvzp07R5VnzZpFqd58800k9+HDh923il8JMcxQ/EqIgaC9 vR13uX379tdff/3ll18OI6XQWj/72c+mT59O+suXL8dqTvsQIrbxq5MnT6an p6PBQuy8adOmGzdu3L9/v9ez+IoTgyJzx44dSOJgc4QXSPHi/ffff66kVPxK iARE8SshoqejowPthJB46623Ro0ahaB64403Fi5cuGLFClx2dnb25s2b161b t3jx4rfffptvkWR4w7lz55aXl9fW1kYpusLHr2wdquXLl/PtT3/6Uy69bNmy ffv2nT9/vq2trbOzM+rah9LU1HTq1Km0tLQQCTRmzBgKeejQIXsxsNdzh2v8 ClPTEMeOHdvcDekTagkIIUREKH4lRGxBG9y/f7+urg7PnpKS8sorr6CXRo4c iXhAPiGoMjMzMzIyUlNT+RatRYKXXnqJb9PT0/Pz88PMERquxCp+hRTB+Agw RFpw2QcDOXfy5MkbN270eu7wjl+h7ffs2bNz505k2/79+xNEcAqR5Ch+JUT0 tLa27tu3b/369aamUFyTJ08uKCgoKyvDs9++ffvOnTtXrlwpLCxct24dYswm aI0dO3bTpk0HDx6M8hXC8PErpODevXsXLVrEtxTs5z//Ob6413XRYwXuPi8v b8GCBSESyK7OV9S3r5DdcI1fffDBB8g51PWdbgZu5XwhxCCg+JUQsQUXjzc/ fvz4qFGjTCYhG15//XX8aXFxcWlp6aVLl6qrq48ePbpz585x48ZZsp/+9KdI i4ULFybIYgKDSaziV0+ePMH4ubm5WBKbh4g35BzKDWnX67nDO37V2dnZ1tbW 0tKCbGtvb4/Joh9CiChR/EqI6Ll9+zYecMWKFS+//PIrr7yCrFq2bBlC6/r1 6/fu3cM/dnR04AHr6+vRXThuEqC4xowZg8dHFbgnhrbEJbmhQCoqKg4dOlRU VJSfn2+77Zw+fdre+AuJPvUVvyLllStXDh48uHbt2hkzZpgUHDlyZGZmJlqi oaHh7t27AzH/CiWTlpZmU9DfeOONKVOmWDzKIntoGMyCDAg5C8FAqTZt2mSi lPScuHLlyr1792JJBBJmpPoh70Iiq1BcHLx27VowT6pGBckQU+zbt2/Pnj22 BeSJEyfOnz/f1NSEAgnOAcNWZI5s3r17Ny1CepJxRQzIQYxfUFBgOVRWVoas e9/a2mrByeXLl0+aNIlSISaRlNnZ2RSAktO+lI0KIrwPdcPlQuZf2VIMjY2N NCgXpeFofUrCRVF6p06dqqmpaW5uDjnLwmIoLrdhEP3E4qVVVVXHjh0jE8pG FyovLycZlqHumvclRJQofiVEbME94bI3b97885///KddoAf4FxdWV1eHl8e1 IZCuXr2Kb0V0oRAQXSYtkEC4PL4yz46PQ3dxT128eLGkpAQPiDPFj6MoDhw4 gBLDxeOdgxsik578caDOmXKVkHiFSQWEhJMKwaiOrQROCREtyBIuRMmRcG4H atQRRbp//37PeWLoDUQCpcVr4/0pJ6Ulfzw4/3KiCY8Q+Rer+JUtlIoAw5KY HZMi3tAzlvO0adPS09MpQ69lRhTt2LGDqwfna6FYECGoI4z8XIFkkS5MhyJF uXEuRsBinIUFqCP/YhkM23OvScyCrTAyLVJWVlbXBQcprRNRfOArFLizvEWu aAuaiV4URlJiFpqbtjMR3nPZWJqDbkMyrkgyehrl4S/lp9hkRR3p2NQueBZV ppOgykjMFen2pKQ8FgxE8lF3OsD+/fv5TJej7oqbCeFQ/EqI6MHjrFmzJiUl Bdf/2muvzZ07F9f86NGjvlLOnz//lVdewYlzCtrMpTTxg6xChKxYsQLNMHbs 2JEjR9puO2vXrrWNjD3jV7YW+vr16znC8RFd7w9yXYr37rvv4lgthBVzawTn UE2ePJmKTJkyJUTY9JyIbo/nsAxmocqkHD169IwZM1BNyBIcOjKJMmdlZVlN DeTQsmXL0LHHjx9HP7jcqBqJEWPUfcKECYhhyoMSXrp0aU5ODhZDqQblBFbF tnxLnuhh0m/duhXBgHBasmQJ1Rk1atSYMWPGjx+/atUq9AYCyZ1rds7MzKSN rNZ2OYQQNaLkyKSmpqaamhp01PJuKHDITjcmqJCCNBkXxWi0O1lR7JkzZ65e vRqRg4QLWTUL2dnY2Lhx40ZnE/qJvc2KrehgZIIl6QCpqakYBHlP3bXulhBR oviVELEFX5aRkWFSyqJSyBXuC4vbBFN2dnZyHLHh9oiZOHEiP/lRF+bZSd/a 2lpSUoKaQjPgAUmJPENgIBtsdneIFLlz5w5uGgfqnCnyjIPBmIlJBVRHcEK7 +xafjpxDwiFaUD5cCOeL6LIdqKkXfvzw4cM3b97suZaUPcLbsWMHyfD+nEJp yZ+So0AQQgUFBXj2kAdYsYpfWdjEln2wKW2INxdxohbIG6zda5mpLIoFoeXs hmJBxpAedYSRnyuQLDJDm6JIy8rKrA9gMZK9+uqrtu3Rrl27MCyqLKQM6DFs ZaJx8eLF9IG9XWBqE1HkwIdFixYVFxe7GXoWeTNJaY93+5KUyEXqgv2xBsdL S0uvXbsWLACNTrc5ceKEvfT61ltv0Rxcl/Lbbk3UEd0V8osAvUeN3nnnHepI R0Kprly5EulurTBr1ixsTgegPLNnz6bL2RPVfrSsEMMSxa+EiB5cP05w4cKF NsEJP56amorOQfnYhrwuJaLl9OnTeMD8/Hw8LN6zurraEqBJuAHxX/h0dAgO d8yYMbg2XCHemQ842aVLlyIVzp49ixd2efYVvyLZhg0bEAkcxA86WcK/OEdk 0qVLl5qbm2NoBzzyvXv3KCGlNVWJZkBO4JdNRdjjLerecyI6ZaZUuH5b8sJE Dh4cIYQxER4IAHLGCMHFGexBHvm7ye3UiHqh9BAGliHpMSApR40aZcIVNYKg qqurcxGkYPyKq1MGyrlz504uTWtSHY5QHnJASVqlbHKd2R87I1RIZiLKugHq hQJQciQZLVtVVUXLTupm37597ur2duHFixexDBKIpuSiJl+5roXObK8lckCQ WwzT7NYzfkUH2LJlC+IZW1F9zG66HTmKECoqKqIkfS3EKoTwRPErIWILtwAO GvdnrwTaY7uWlpaer9vjNHGsNTU16enpuOnly5ejgioqKmwVd9LX1taWlZUh A8gQB4oHtOnxuGbEACqI45x79OhRF6GKPn5lc3W2bt2Ks8bhciGLQeGF+YA7 pkbIle3btyPPOjs7LdRm874oicVtSGYBGdu9mhORFpyIwszNzQ2RbbGKX1Fx MrdlH5ys4ggVcREn/g2ZOYYqQ7llZmbSZDZt3rDFTrED1UQCPVcgPXjwAGsg htF4aWlpVHb8+PGmfk3/WPRvzZo16B8nvYxg/AqZtGrVKjoDIo3PHMfytDsf yBBRhFCkpsgnLuokJTozjKSkF9mW4sg2OgwKqr6+3l29oaHh/Pnz9BN74kzb 0V624Tglp9i2b1F2djYCD93urBeMX9HbyZxLcF0UoO0VTia2Ui6fqRE2pO/R FfXwUYhnil8JEQuamppwLosXL7ap13hMPNHx48dxlE6iPBdcKgLGnjqN6F5r HQ/oIj/uwdbGjRtRBe7EvuJXpaWlaAbLrSf2VKivBTn7B9qGDDdt2uSukpWV VVdXt3PnTopNRUxg9DoRHWnUazmRAfYGQYiwDIIC4auqqiryQd2hCjCjxeuA HCyC5/aFNHFlK4+ZGAjGr7C8zaMjmXu2GwIyo7y8vLGxkXOPHTsWxs4mgBFC YfYffPToUXt7OxYIuaL1Af66I659w8SvTDVZxDIEckMIIaWC8U8hRD9Q/EqI 2FJbW4tOeOWVV0Z0zfnhh39+fn6YJdnxmydOnMAFl5SU4GERG3hDvGpLS8ve vXtRBZbViO6Hdy914w6SBmmBj34Wi/hVcPEE53PtOZQ7MnLkyF/84hcIALeB jrtusGwjupdcCAoAlAY6gcTuirGKX1VWVqIJreQTJkxYtGgRsg1tiVpzV3/3 3XdDVhizqJfFuHqSmpqKCsWGzxVIra2td+/etUefFonCbuhexAxVM+uZcYLS ywjGr8KIHyDzKVOmYDF6SHNzM4XPzMzsNWVQUobff5AK7tixY/z48cFWs1df g92MItFP0O3OesH4lb2tifyjFjbzMKQ8tidURUUFttJbhEI8U/xKiFiA88Un bty40dZvx/OiwVasWJGVlYVHxr3i43A958+fR57dunUL3RKyB58tfIQftMc3 ZDJ37twtW7ZwenFxMS4eJTB58uQRXbEXW8vUrV7VV/zq2rVrXJdMkCI2tdse e/EvksOCV8HHWNFDhqaCnPzYtm0bShJ5mZGRMX369BF970JoQghZ6OZu2RPS 9evX79u3r7q6GsdN5mQYXGZh2rRpq1evRhigNKhvR0fH4cOHFyxYYGIJi2E3 rIe5EGP2OqFFBSlGdna2W8kqGL8ymURDcHW06NatW5FtGzZsmD17tosmocdo GpORDQ0N2Jk0JLAYHRXnA8qEGlFySmWLa/UVv7p+/TqSjIpQcXvqOmvWLDoP bZeXl0e/cq1vsbicnJyrV6/au58941foHBJTF+pLHSmYTVqzwtMKfEWZY9ju QiQhil8JEStwhTYDB/9lThYXjO8uKSkJ84P90aNHeE9cP/cOf1FiOMSnT5/i E9PT0/HgaB7csS1EgM9FYiEhFi5c6GJKy5cvRzPYrRdN/MoWf0AHoklsPs+k SZPIHNdcUFBgu9chXdwcMK7i3oarr69HoixZsmRE10wnnDWfKSdnITNWrlzp FhE1wRN8/Bd9/MpKTqXQBlZylAzKwbYf4oOTPdjw9OnTTU1N7lyzWGFhYUjU LiUlBVOfOHGCRqFNnyuQsD8SiPyxDPKMymJhvs3Pz8d6qEe3KyJNmZmZiQXc 9PVg/Mom7dPua9asQf9gQ+QogtnOJQ0pMWl7eztdhcKTD2lI36uktMd8fcWv 6JZ0DJoVwUmZMRHJUG5oRetmmM6kl82jo74oOqxnqi8Yv7KycQnyp89QcjQe xcAIFNhFwIqKirh0zxXAhEhCFL8SInpwKLgVnHjIdB2bA4+nxsGtWLECj4ar PXv27JUrV9ySlQaiy61Mbo9v8H23bt2yhzUWiUpNTXU5r1+/3kUw/PcfNPfN vwNkB5QM2sBNOMdx43DRGLYBMaLRxbX62oUwmv0HUUEtLS1bt251cguL2TJf z7rjPIgNm0/Ot0hERKNNBQ/Grwy0HBIICdfW1oaMwdoIMLc1j80JD8rI8Nvr 0IJh4le2wIK1oL0wiICpra1FX5GAS1MLdI47l880Mcef9Ra/sgghwo/jWIbC o/fQUVScrkXmEyZMCK7fJYToB4pfCREr7H1AXKqbbT5x4kS0Ckc8Z7AHwa1P njwZtYPLwyHiAU+dOoXo6uzsREKgRuxB4Yiu1+3RDCYzoolf2drjO3bscHOo SFNSUmI3tb0hiHRx8mPx4sVOflA294DPlv3kq/v379vs/fLycgthjeh+fBmU FtHHr2zNLpSbTf6xmB4lR6ZydaSaKzMSjsqiTEJyiH7/QVuEgYMjumdJYQFy oAAIyKqqKrdKlb3JSDHc5Plg/MrkE2oKI6B/sCG62uVsIK4osC2h4LP/YF/x K3pFc3OzxSRtphySEiVmS10hzMjE1ms16UXb0cRYzyoVEr+ywiP7MYut906t 6Wn24NLUO70R4TcQi9YKMeRQ/EqI6MGT4shwanirpUuXjh07FmdnU4hN4Ywf Px6PPHfu3EWLFuEf8af432PHjuHgcEa2fYmL/Nj6ooWFhThWmznf3t7e0NCA 93cCI7gKejTxK/wsBWhpabkdCZQnJP5mVFZWOqE1bty4BQsWlJaWPut6v/L0 6dM2Ed3eiaMuqIues7+iiV9ZE6BU7RKIAbQEY5q5e1MUiCK33IE9ZKyurn7W 2/wrxBjKBymChkTqkAkt4hYgRc+gVY4fP+7KFk386ujRo6ggFLXLmcLQKAgk W6iB1kc10YtsiQlb4cos0zN+hd0Q5HQPviIH/tqyWvYIEsvwAZWuVRSEiAbF r4SIFSiKmzdvlpWVufgVv9wPHDiAf++pNMKD00QSoBM2bdq0du1afOWJEyfw oVyira0Nd4mucNO8gwsyRBO/ssU/ydmdazkj87gi4gTVhAqiVJSHgu3fv9/N v7I8d+7cyaXx3QUFBZcuXUKxIDxsbr+bQm8bOpPYlSf6+JWtBEVR3c6D9k6l bXZcUlLi5i9Nnz49IyMDaRGSQ/Txq+AiDLbXIerLzbBiYCTx8uXLR3S/RIAZ sYy9gBCMX6GRFi5ciG0pPM1Bo1y5cgU9bOcaqGUO0hOeRRe/ou1oQfSkW+sj NzcX49jrFZYzl6ZLUC+sakvxk4CSo/16vj/IVbADbUHJsZ5t+oOcs9c6+Gsv I4TsfihEcqL4lRCxAm+Fj0ZO4HHwRL2+xm5YDAFNUl1dfe3aNUQCAsBNwEao hMRGnnVJMlSN25uPBDgye3gXTfzKImOkuRAJ6DEXWwtiT4tswjNCBZFmspDE SFPUi7MA+qSqqio4Ed2IJn7F8IXWMmFpq57m5+eH7NaHKFqxYoU90uIv4qG8 vPxZb/ErBFjI2mWkJL1tx2N2Pnr0qPs2mvhVUIDZPjvI1xDLlJaWIsxsmQV6 F0VFkz/rLX4VIsC4SmVlJRrbbQ+EZWhEewjYe1cWQjwPxa+EiBV4W36548Hd JPYpU6bwcx5X2I/41cOHD2tra2u64LbCS5I/+gEdgu9DL7k5RbbS0enTp59F F78i/9bW1i1btrhzbQEEjtiDsI6ODuQWuVEGi2i5JSAsgoSc4+roMb4KPrrK zc11y4facqBoP1ee6ONXFMameFmxET/oLopq++WdPHkStWbSFNWKuWijkByi j18VFhaOHj3abfoTssUzhUFBbdiwweXPZyzD8We/LJ9MNAYtYDYMnrtmzRoX OYwmfmU9ga9McFIpsqVBneDkw/Xr1xH5bn9DKoj17Kloz/Xb0X42qd51Y/oM PY1Gp6fRXTMyMvLy8mK76ZIQQxTFr4SIFagjvN6VK1eOHj2Ky8PN4a/xhitW rMAx4a9tMxR7PdA2BcarktKU2+LFiy3CYG94cdbuX2bhwoVub76UlBQX5Ygm fmXfbt26dU0k4IJDZk/ZCuQU0gmJpUuX4txx8c+6407IMPfQk7oUFBT0nIge Tfwq+JaiPRGzPQQxlLMhIs1F2N566y3UCAr5WY/12zmX0obEvlBotjVMDONX j7vgWiFT93sOxQjszMxMe8WAq8+ePdtauWf8Krgg7bOutwMwEd3MIm+KXwkR ExS/EiJW8HO+qamptLQ0OP8Kp4xjjTR+9axbjOHWy8rKcJS4P/QGAiA7OxsB g1hyiwzEKn5l7w8WFRXhZ01gUBHTCagOlB5ShJSUpLy8HP9LZe39uGddsa+2 tjZyPnXqFKKiuLi4sLCQxEgd27lvzJgxZpaBiF8hIymbbXlj+aMJXfQpuEK7 rS6Vl5eHDAsuShZN/MpW3yJPtziDrR5PKzjZhkANeQcQLXTx4kWL5ATjV0i+ kPXtKert27dpd3duRkZGdXV19PErs4zNrcIsKEP3bqCdaBFItB99wJ5lU0dK SwPRo0LiV0g7OhWJg01Dx8CqmOvlLigbV1T8Sohnil8JMTDguO/du4eLxBWi oHDB69atQ4/hidykLPwRXi81NZWU9fX17hGbD8HZTdHEryibm9XjDwoQF2wC wLAHZMEoCgIjZKNnVF9wGdJeJ6JHE78iJemDS3GGx5ZZsKn4wfgV7RKyqZAR XKIhVvErNB5KJrx8MpBMaEJbu962eEa4PustfrV+/Xr31qR1RRri2LFj9l7n CMWvhIgFil8JESvwR21tbZWVlcH1r4qKik6fPt2P9a9wbfjWGzdu4MdRSvhl nHLIXn5GrOJXxokTJ/jKPSpy2PpI1Ai5tXXrVsRASKSCAl+9ehVFkZ2dPX/+ fFKSs80QCzIQ8auKigq3Qcxbb72VkpJiD/UM1IXblMcILs1qRBO/MhVE+ufI tV+Ga1EqW0AjGL9avHgxfcZeTDDs0So2d+fGKn5FjTAFUrDX2VOG/RDgK5P9 FHLnzp2U3N5tdPEr+gZ6Mi0tLbiv4rOu+BVim8w5UfErIYIofiVE9JhYcnzQ Be4J2YMfRKhcunTpzJkzyB4c1pIlS95++21bHQuJsmDBArwwrtB2KhnR/ZAL jzmzb2wisT1mSoT41fXr11Fubgr6iK6lEjIzM5F/7iHa0qVLncywZUhLSkqe PHkSjKJEE79Ck2zZsoXqj+heJAGlF8aGiIrly5f3fH8wvvErWpasnhu/mjBh ArLqmeJXQsQJxa+EiBUogY6ODnyf238Q940AwMmGiV+hB1AIaICqqioEwNWr V5ElNu+FE9FIixcvxim/9dZbkyZNQmLh9N95553Zs2e79dtjG78yIYTaSU9P 51pc1GbX2KIEOG60FsoBJVBYWOi2P7alO8mHsvEtaZB/+Gs+U/6UlBSLv42I dfyq556J5I9lKIaTbZs2bVq1apUTDyO6F3+wvfmMKONXnZ2dFNuda4/nKFIY 8bZ161Yn/IZW/ArLVFRUKH4lRJQofiVE9JiTAhylrbTQc20oA7mCbsnKyrK9 m0d0LfJ5+fLl8vJyt3kxf20u8Ya+QVe4t8wSIX5l0RWbgu6D24Uw5B29aOJX JieQpq6mlCeMDVFTbhX0IRG/cpP8R40aRZMpfiVEHFH8SogYYlsVjxkzxuYd 2b6BRUVFwVfVQkCEBNdpLy0txaHj8hoaGpYsWTJt2jSLHSG38MjoFtwuXhtP 7TZMiW38yrh582ZVVRXXIhlu172rGMRWebIlFIJv8I3omqxl27hQL6rPVzaF bESs41cWO6KcbmEKH3quTxVN/IrGDZFAVve0tLQw4o1aYzRbJCGO8StqRF1o 3MmTJy9YsKDX+NX9+/f5Khi/Qu3bFkiKXwnRbxS/EiJ6WltbkTQFBQU5OTm7 d+8+cuRIX4s24LZOnDiBBzRvO6LrTUC8LfIJH2dxJPTJ/PnzcXNnfxkU0alT p052gUjjLNuIJJr41bVr18rKynDQuZFAaW2JUZePrW3utvp9LrbGVHZ2Nson uI5WNPGr4EIN1BS9kZWVFWJDbEXhzYx84F97jBiv+FWv619xYk8hRGnR1bYQ KPlzob7Wv1L8SohBQPErIWILt8Ds2bORQLZGKG4Ud4Yv6/lAEL/GceTBwoUL 8YZ4RlwnPhEv39bWVl1dbULipZdeGjNmDDoBeYYPtQXS8/LyuApOfMTz4leb Nm0KkShBQRWMX9lcJsrZ2dlJPnhwyobYQ1zh61EOeH+uhfO1Z5c2Qd2uSzK0 h60Ebm/wbdmyBcFAXZB5VMo9mItt/MoW3aJsFuXzFG/UwpXciCZ+RZoQCWSL Y1DHngL4ZDe0gpOg8YpfIatQ+7b+FX2MBu25/hWXRp7ZSwEjute/omPY/keK XwnRbxS/EiJ6uHHwa/g4ZACedMmSJfi4XqdgdXR0XL58ubi42G2yg7NGQuDc kTcWYZg8eXJqaqq91xYER48vxqORCR9w+iYPoolfxYqgiPInPT3dRZCMaOJX V65coRgWg0Iijho1qmdNTSxhPWyIJZ3Eilf8yu0/iOlM1s6fPx9RFFyA1Cgr K3Mr/Ifff1DxKyEGAcWvhIgt+DJ+xeNbba0qFMuCBQuampru378fkhLfjVdF 6oSs937y5Mnr168H14GfPn366dOng9PFcc3oJVuuIXz8CmWFNmhra3PnkgAH bY/Jeq7fTjlJjPN1+/fZcW5tBEBmZibKwd5btCUUTCQUFha6rW1Qkjt27LAI iYEEcgsjxDZ+haFCZIkPPeVN9PsPBiWQba6NnAspLcqtoxs+u6Un4hW/MsEZ fv9B+jPNYeJ8RN/7Dyp+JUSkKH4lRPTgdPCJKSkptuwSHm358uU4QZQPesa8 rW0xg8vDAZHYppSbPMNX4tHc2uMIG/wdHtye0Tzrinq1tLQgsfB3O3fuxIGe OHHC7QAY3/iV1YtS4d9NQkyZMmXlypWIwM29QX1HjRplQiU4hd4Ixq+QOgge ZI/7tmf8Co+PcLJlPG/evInNsa2tlcolbH6XSV97WQ/xibzJy8vDhlxrgOZf UX5qQf7u2/DxK9qL3OxxsE2epzCtra20uz3C43QEoRNRJHZvPip+JURcUPxK iNiCluAHPp4ULYFAsjXPURf2q7+pqQm/hme0sMDu3bv5Re+k1OzZs6urq+vr 60kZjGvhi0tLSy0ygBggE7y/29EvGL8ic87FyztnumzZMi6NbMNdoj2QYbh1 t0J7MH5lc5nws7jmrVu3IudQJhSVsziXD+SM03crTdk0e1spvaCgwM3LstlH iA20gUVXKABKyR5dxTZ+dfHiRdLboppY48033+RCvco2ZJWbIUZKrJefn//w 4UN7tbNn/AqRhqmDz3DDC6TDhw+7Nb6QsghaWgTpYkEqzI6OxVaU1naUpqFR nrYwWmzjV0FJafQVv0K3kw/9xzoqTWPvJtCLKLnljEQkc5ujRVdB4506dYrO YFP1FL8Sot8ofiVE9OBQ8ER4H7dfjC09evDgwZs3b+KtcPQ4NdvNOSUlZfLk yeby8In4r46ODsQPHnz58uX2KIfjeD23yQvuu7a2Fp/uNtCxJTRRYs/iHb+i Xowb69evdwJgxYoVJ06coMB3egP/O378+FGjRo345Sn0RjB+hTDIysqimu7b nvErrnX8+PHr168/64rkoDrQSG4yPC2CWjArmYxBdNE09gAUQyHMUCDPYh2/ Cq6vboSPXwXfDbSZY2SFZqbnIN7sHU/OdTP8g+8YKn4lRFxQ/EqI2MKPepx4 aWkpTtD8HQ4Rl71w4UJcGGIAr1dfX19RUYErR0eNHj3aHleRfsmSJUgmNAb3 EULCTXHn5/++ffvs5rKVqZBPzl0G41eoC3LIzs523yKr8KcnT57s7Oy8ffs2 aopvuahlHoxfBecyUSRECCeaE3/WNRUH9YLXRv5RHSsVftwmUVMXtwIVomjN mjVc0VZVpTw7duxwV4xt/Kq8vJwyWCzOFiUoLCzsVbadP3/eTW430E7uDb6e 8astW7ZQcYzmrhVeIHE6jYL2c+KqpKSEzmAqBbNTMPfOJraimm6GW2zjV0FJ afQVv7LewhEUIy3OX+yzadMmzjWlin1Q5ijGV199lQQ04rRp09w7hopfCREN il8JET24KhxicXGxRWZwVciMuXPnLlu2DHeGo8fl4UCzsrJSU1ORH7bDDoLB LU+Kk+Xuw+mPGTPGpifhFlFKODiy5XSLb5Azzm7KlCl4MTepPr7xq2vXrqE2 qdeI7lWtqHJDQwOap6M3KJt7Dw4LzJ49G1HnpoIfPXoUu9nCqjYPbePGjWVl ZbbZtD3Vwia4exO36AFkD9oGRYHrxyCIDTeXCTmEJMDy2BCL5eTkIINpGiyM HTCpe9AWffwquD+gxaAQQhTm8uXLiBDyDxO/sugWFeG6tBGnI/NoUwqMssUC bntr2x6I4y5CpfiVEHFB8SshYosFbfiNjzfEmeIKccc4LJQVTpCDeDcc98qV KxcsWIATf/XVVxEDuGN+5iNsEACIMbRHZWXl6NGj7XmiLemA78aZoqnIBCnl 4kUmEkj/rCuaQQ5IBfc2n4mQVatWocHeffddJNz8+fPRLTZ3K2T+1ZUrVxA/ NpMKfYJUwIlv27aNguXl5VmwwvY95FvKz4VsoYADBw5MmjSJco7o3r6HQnIK 53JRUrry2ObUCJ6nXTzrb/zK4mn79u1DFdjjQoublZeX9yrbUJuoCCzvHqIF 37O7ffv2+fPnMzIynAixZRDQV06mhhdItswXTUyTURcaFFthH2xLZWk7FKwt D4uh0NXoGfd+YjTxq/CS0iJFfcWvLPZFj6KyVAf1SzKbPkeGJEPP22uq1Jfi UTuOYBBTfYpfCRENil8JET0ICbwkjtj2Pu6510xP8JWIKPw7msfcGX8PHTrE QVuWwR7h4RbdRs/G5MmT8fsVFRVOwMQ3foVYcm8+mpzruatgkHPnzrnNdEzG IFFsojjfHj9+HKlp09TNAhMmTKDwZWVlztTkz4lmZIs1LVy4cOfOneirZ7/8 mM+Fa7ChCQyTKBgBSVBQUGC7GD+LRfwKyU2GlMRd17aBJh+ULd+GiV+hUu7e vYvec49ZDXvUGOxOVART074u9KT4lRBxQfErIQYCnBe+devWrXjt8Pviub0F KysrTUqZGMPBjRs3zua9WErcqItZmfCw9/WC86A4lxxQAm4SeE/I061hFYxf 2ZQbPocINptOT7LgGul8yMjIOHv2rC1fgLxxbwj2Wlq3OIOTTO6RX//iV7br H6rJ7Y24YMGC/Px8RGyv6TEpSmPbtm2uCosXLy4sLLRdCFtbW+vr67OysoJW QseiS6uqqqyO4QVSS0tLc3MzSpITLepoMozK0hBut6MRva2OFU38ylS0zX/r S1L2Fb8yUFb2hqDLn1Yzqwb7HhWh+qhTe6XimeJXQkSH4ldCxIQPP/zwzp07 JSUlu3fvXrp0qU0iGj16NO7Y9lXBP+JkUT7Tpk1LSUnZsGED7uzGjRv4btMh /MXt4itx+rNnz8Zn/aKLN7rgw/Tp0/Hd+Eq0VnB6c3zjVxTGLQdBIakd8s+J q54gPJAfVMSJtC1btqBw7CGdldZ8uj1aHT9+POayZSIMlOrKlStnzJhBApuw hPQKLkqASLBWsIdfzoaml7DP2rVrMT7iBAkRq/Xb7RFkdnY2KpRr2RsN9IG8 vLznxq9sbS5kJ4mRKBTbMkEG20aK1ALDUs133303ZHdCxa+EiAuKXwkxEOAN 8XE4XLRBRkYGvh6BgZrCFeKdTUrhWzmOT9y8eXNxcTFO0F63f9bl8lBWOTk5 6enp6AdOtBDQuHHj0BJ42DVr1qABTCEsXryYIxUVFbYhDjnU1NTgiHHWeGHE g+kBTseBzp8/f9WqVXjh1NRUzp01axZKz2SAzWiyVbm4LjqB0zkLGWBCBcjN lQE/7haIMPmRmZnJJWy6DmfZlnYoNyywdetWNI+7IgrKlsd81t/4lUWckAou zEL+GCEkfuKweUqoJjf3LLgLISUhQ1rB9o6k/KShjbKyskhgCziEF0jImIcP H9Li6Ftsix346vVuMAXZUjsMa8Gr4Ir60cSvDCoeRlKGj181NDRQLzohHQaR Zo9KyYSs7GXPqVOnUmwScBYtbiuGPVP8SojoUPxKiFhhr7TjoVAUeGFkhq3V iW+1B0mTJ0/Gv+ODdu7ciTy7c+dOSADBdhjE4yOQEFe4LRw3ngvNgBPEkeG7 6+rq3GwlI77xK4qEj7b1rCgw7rjnzolBbt68SYExgpMTfD537pxNRLdFULEP 9aXuuGwyD6kRguHYsWOIEMxie9YgCBGrQdmGwkF1oFExC4qCiiMSkLJYFSFx 8uRJN/PKiD5+ZZPJEaWYGjtQKtQLrY+afW78yuVw//59rIfqs0ywACUxoY4E QroggUxju7MUvxIiLih+JcTAYTsM4nZxiMuXL0dNTZgwwUJJSCn8b2ZmZl5e HooCZxfiy+yd/bKysvnz53OivYGI+kInnOmiqqqqsLBwXRd4zMrKSrcaOX6c PIuLi/HCOE2bSjRx4kTcN94cp48XtueM6IFNmzadPXvWXbejowNdh6+3V94s AmZiAOlFsdEq27dvD9l22TbBQeRwCXw9iREYXBpphFo7deoULp7rruvm0KFD blfE/sWvTETZsg8GBeagkw29grxB/9gLArZeVlCY2dR3KjiqC2qBZHULkIYX SLZSFk3Q3NxcUlKCHWw5d6xHMtKTLaLazeYKEn38KrykDB+/etYtsWh02osE ZGLhR5SbBTyR9PRkW63LofiVENGg+JUQMQRHho9GWuD40Bs4Ypwp7nXPnj38 PXjwIO4eh46zQyfY47MgaDBywN2jBE6cOIEy4XTORUqhWHCCbk3voFqziBDC ANe2b98+rhucn0P6uro6hJxbTMCmfMcKm09VUFBA/lSwvLw8ODesJyYOT548 mdsNBnEPIk3IYR/qS93z8/MRDyE1Qo2gN7AGZsGqpEF7YPCgW6fWFAy1g1ko HhUnN7JCYyBdsLCbeWVY3AybUx4kMemDuyIatr8hLUgaLk2pKIb7ltxQv5QT U9NwlAqbcNGamhoq1dLSglkQya7W1DFklpo94iQZxbZMrPUpPx0Ji5EVdQwR 6jaxnw7jcqYz0LucWCI9Z1FUTGoJKBjira8XPIUQPih+JcTAgUPE15u8QVfg v/DgpmFQGvhfNAA+EeUQEhl41jWlGZ9+7do1tAEn4vJwo/hudMLNbhBCZ7q5 ceOGzb961v0mYH19PV6YC3Eu7h53XFFRweXwpPhT1AVnoQfcozd3XXQdTpyv OJ0CU1oTA1ZsynPx4kU34dydhUhAPHAJfL1dkcRIIwqJXMHF24Q0g4HCxXzQ LZQNIYRnt8x94hsmorCqkw0U2K2I3hdcF0Flq4nyF3sGhZlNfacMe7ugFjb5 ymoaXiDZToL2ViNXwQ5kziUwGslIbxoPu7n37xy2ujsdg1Ih8Gg4WzPf2dYm twcFJ/rHVp430EthJCVtTQG4RF8WRmLduXOHRkek0S0tE/7S5UpLS20RMGwb FJzPuqNe1NR2xOZEe64aTEPhKRU9gdpRKkqOuYIlFyJpUfxKCCGEECIiFL8S QgghhBhkFL8SQgghhIgIxa+EEEIIIQYZxa+EEEIIISJC8SshhBBCiEFG8Ssh hBBCiIhQ/EoIIYQQYpBR/EoIIYQQIiIUvxJCCCGEGGQUvxJCCCGEiAjFr4QQ QgghBhnFr4QQQgghIkLxKyGEEEKIQUbxKyGEEEKIiFD8SgghhBBikFH8Sggh hBAiIhS/EkIIIYQYZBS/EkIIIYSICMWvhBBCCCEGGcWvhBBCCCEiQvErIYQQ QohBRvErIYQQQoiIUPxKCCGEEGKQUfxKCCGEECIiFL8SQgghhBhkFL8SQggh hIgIxa+EEEIIIQYZxa+EEEIIISJC8SshhBBCiEFG8SshhBBCiIhQ/EoIIYQQ YpBR/EoIIYQQIiIUvxJCCCGEGGQUvxJCCCGEiAjFr4QQQgghBhnFr4QQQggh IkLxKyGEEEKIQUbxKyGEEEKIiFD8SgghhBBikFH8SgghhBAiIhS/EkIIIYQY ZBS/EkIIIYSICMWvhBBCCCEGGcWvhBBCCCEiQvErIYQQQohBRvErIYQQQoiI UPxKCCGEEGKQUfxKCCGEECIiFL8SQgghhBhkFL8SQgghhIgIxa+EEEIIIQYZ xa+EEEIIISJC8SshhBBCiEFG8SshhBBCiIhQ/EoIIYQQYpBR/EoIIYQQIiIU vxJCCCGEGGQUvxJCCCGEiAjFr4QQQgghBhnFrxKBDz/88H6/uBsJpI93RWPA QNtqeFjp0aNHEfUN4969ew8i4f333493RWPABx988ChyOOtpJMS7lrGBWj+O HMzV6ceTJ0/iXcUY0NHRMdCDOTASxruiyY7iV8MDCTB/JMB8kADzRwLMHwkw HyTAkgTFrxKB2trasWPHvhohP/nJT77xjW98zZvXXnttGGiDy5cvv/HGGz+L kH/7t3/ztNXwsFJmZqZ/x3D86Ec/ejMSioqK4l3RGHD27NmNkVNWVnY7Eh4+ fBjvikYL2qm8vLwwQgoKCjZt2pTpAVZtamqKdy2jBe20dOnSSAfzV1555bvf /a7/rUri06dPx7uuyY7iV8MDCTB/JMB8kADzRwLMEwkwHyTAkgfFrxIBboTv fOc7fxMhf/7nf/4rv/IrH/GGm+7u3bvxrmu0nDp16pvf/OZfR8hnP/tZT1sN DytNnTrVv2M4/viP//jbkbBhw4Z4VzQGHDlyZE7k7Nmz53ok3Lt3L94VjZbH jx8jh9ZFCLoIc03xYPbs2e+99168axkt/PhCDkU6mH/lK1/5jd/4Df9blcSH Dh2Kd12THcWvhgcSYP5IgPkgAeaPBJgnEmA+SIAlD4pfJQKST/5IPvkg+eSP 5JMnkk8+SD4lD4pfDQ8kwPyRAPNBAswfCTBPJMB8kABLHhS/SgQkn/yRfPJB 8skfySdPJJ98kHxKHhS/Gh5IgPkjAeaDBJg/EmCeSID5IAGWPCh+lQhIPvkj +eSD5JM/kk+eSD75IPmUPCh+NTyQAPNHAswHCTB/JMA8kQDzQQIseVD8KhGQ fPJH8skHySd/JJ88kXzyQfIpeVD8anggAeaPBJgPEmD+SIB5IgHmgwRY8qD4 VSIg+eSP5JMPkk/+SD55Ivnkg+RT8qD41fBAAswfCTAfJMD8kQDzRALMBwmw 5EHxq0RA8skfyScfJJ/8kXzyRPLJB8mn5EHxq+GBBJg/EmA+SID5IwHmiQSY DxJgyYPiV4mA5JM/kk8+SD75I/nkieSTD5JPyYPiV8MDCTB/JMB8kADzRwLM EwkwHyTAkgfFrxIBySd/JJ98kHzyR/LJE8knHySfkgfFr4YHEmD+SID5IAHm jwSYJxJgPkiAJQ+KXyUCkk/+SD75IPnkj+STJ5JPPkg+JQ+KXw0PJMD8kQDz QQLMHwkwTyTAfJAASx4Uv0oEJJ/8kXzyQfLJH8knTySffJB8Sh4UvxoeSID5 IwHmgwSYPxJgnkiA+SABljwofpUISD75I/nkg+STP5JPnkg++SD5lDwofjU8 kADzRwLMBwkwfyTAPJEA80ECLHlQ/CoRkHzyR/LJB8knfySfPJF88kHyKXlQ /Gp4IAHmjwSYDxJg/kiAeSIB5oMEWPKg+FUiIPnkj+STD5JP/kg+eSL55IPk U/Kg+NXwQALMHwkwHyTA/JEA80QCzAcJsORB8atE4MyZM//4j//4/0TIl770 pV/91V/9X7z5+te/PgyEQf+k5l/8xV987GMf+x8eoM2GgZXefvvtj0XOZz7z mW9Fwvr16+Nd0Rhw8ODBqZGTk5NzMRKam5vjXdFoef/993ft2rUyQlJTU2fN mpU88unBgwejRo362wj56le/+vGPf/zXvPnt3/7tw4cPx7uuyY7iV8MDCTB/ JMB8kADzRwLMEwkwHyTAkgfFrxKB27dvb9++fXOEzJgx43d+53f+D2++973v tbe3x7uu0XLlypXp06ePjxAGtH/+53/+Jw8mT5788OHDeNcyWoqKisZEzg9+ 8IP/OxIyMjLiXdEYsHjx4j+MnC9+8YsRSc1h8Kj03r17P/7xjz8dIZ///Of/ /d//fbQHEydOrK+vj3cto6WjowMd+P9GCCYaO3bsZG9QpMNAag51FL8aHkiA +SMB5oMEmD8SYJ5IgPkgAZY8KH6VCNy/f7+qqupkhGRmZv7BH/zB/+XNP/zD PwwD+XT9+vXU1NTFEcKA9vrrr7/qwdKlSxkA413LaKF7pEfOyy+/HNFLAZwS 74rGgClTpvRjqv+v/dqv/Z+RMHfu3HhXNFru3r37ta99LVJDffSjH+WXy+se jBs3rq6uLt61jBZGDwackRHC77tVq1a9601OTs6tW7fiXddkR/Gr4YEEmD8S YD5IgPkjAeaJBJgPEmDJg+JXiYDkkz+STz5IPvkj+eSJ5JMPkk/Jg+JXwwMJ MH8kwHyQAPNHAswTCTAfJMCSB8WvEgHJJ38kn3yQfPJH8skTyScfJJ+SB8Wv hgcSYP5IgPkgAeaPBJgnEmA+SIAlD4pfJQKST/5IPvkg+eSP5JMnkk8+SD4l D4pfDQ8kwPyRAPNBAswfCTBPJMB8kABLHhS/SgQkn/yRfPJB8skfySdPJJ98 kHxKHhS/Gh5IgPkjAeaDBJg/EmCeSID5IAGWPCh+lQhIPvkj+eSD5JM/kk+e SD75IPmUPCh+NTyQAPNHAswHCTB/JMA8kQDzQQIseVD8KhGQfPJH8skHySd/ JJ88kXzyQfIpeVD8anggAeaPBJgPEmD+SIB5IgHmgwRY8qD4VSIg+eSP5JMP kk/+SD55Ivnkg+RT8qD41fBAAswfCTAfJMD8kQDzRALMBwmw5EHxq0RA8skf yScfJJ/8kXzyRPLJB8mn5EHxq+GBBJg/EmA+SID5IwHmiQSYDxJgyYPiV4mA 5JM/kk8+SD75I/nkieSTD5JPyYPiV8MDCTB/JMB8kADzRwLMEwkwHyTAkgfF rxIBySd/JJ98kHzyR/LJE8knHySfkgfFr4YHEmD+SID5IAHmjwSYJxJgPkiA JQ+KXyUCkk/+SD75IPnkj+STJ5JPPkg+JQ+KXw0PJMD8kQDzQQLMHwkwTyTA fJAASx4Uv0oEJJ/8kXzyQfLJH8knTySffJB8Sh4UvxoeSID5IwHmgwSYPxJg nkiA+SABljwofpUISD75I/nkg+STP5JPnkg++SD5lDwofjU8kADzRwLMBwkw fyTAPJEA80ECLHlQ/CoRGFD59FvdfP/732cAfNpNvCvdTySffJB88kfyyRPJ Jx8kn5IHxa+GB4pf+SMB5oMEmD8SYJ5IgPkgAZY8KH6VCMRcPv3Wb/3Wxz/+ 8d/93d/9xCc+QbN+6lOf+sM//EMGsStXrty+fbu5udkCWfGud3+QfPJB8skf ySdPJJ98YPRgDGEkea2L4Njy8y4kn4YNil8NDxS/8kcCzAcJMH8kwDyRAPNB Aix5UPwqEXjw4MHZs2erImTDhg2f//znUVC/93u/98lfhiOf+tSn/uiP/uhP /uRPPv3pT3/mM58h5Q9+8INz58699957tPLNmzeHaPyqsbFx9erVqZGwfPny lJSU0aNHv/HGG6PCwhi+bNkyROb7XTx+/HiIzlWje2R5s3bt2jVr1mRkZDDa f/3rX8dF/k0A/v16N3/7t3/7jQCcOETtEwRp/Rvd/O/d/G/d/M9ufv3Xf/3X uvnYxz6GIgrebhYl/oO+QcPHu6LRgnz61re+hR0+1gf/aw84+Ju/+Zs/+tGP 3nzzzbfeemtcWKZNm3blypV41zJaOjs7V6xYMWbMGEYbxhM3tnBz2d9ef7WR 2OTTli1bcjzYtWtXc3NzvOua7Ch+NTzotwD7sz/7sz/25oc//OG9e/fiXddo 6Z8Amz9/PuM/XmDs82AYZAiNdy2jJRoBhrL6ph+cG++KxoB+CzD7mRNGdEmA mQD713/91+eqLwkwCbAhh+JXicCHH354P3JoLO6jlStXcvf97Gc/+68evPTS Sz/thgTc0WgJfGVubm5ZWdkHH3wQ73r3h/fffx8Fdd2b+vr68vLy/fv3Z2Zm vvPOOwzRU7uZMWPG293M6gJPumTJkq1bt27btm3Hjh0HDx68efPmUJyrRve4 8TzMjJcvX0ZrHTlyZM+ePYzedKeFCxfOnDkT42AiDLJo0SK6DeIKsbRx48bN mzdjnN27d5Mezd/W1sa1cBlDtDvBe++9d+jQIXpIYWFhXl7ezp07ua0wxbp1 61CVaWlpdBuMMHfuXAyCDDDPOH36dLrKggUL5s2bh7vktwwnFhcX02eO9kZD Q0O8KxoVT5486ejoWL9+PTfL+PHj+SUyevToX/ziFyYJTBXYkyyGGkae//zP /2QI4jgpMU5BQQHdrKampra2tq4P0E7D4JcLg/np06fpCdxHkyZNmjhx4oQ+ mNgNyehajMx0Ie7HWx7cvn378ePH8a5rsqP41fCgfwIM77m3C4b3Yx6cO3eO C8W7rtHSPwFW1kVlZSXjW21YOAVfE+9aRks0AgxxdfLkyVMekEm8KxoD+ifA +BdJv2vXrl7llgSYE2Bjx45ljKKPXbp0qS/pJQEmATYUUfxq6PLw4cOLFy8e PnyYH9GzZs2a6gfDPn6hqKjo0aNHw0AnhOHp06eMMO3t7fxqRg/s2LGDis+Z M8eFqubNm5eSkjK/i4Vd4CVtuvvSpUuXL1+enZ199erVO3fuDLn4Va9QC1qc 4f2DDz7AMnQAHCJDMZ4LYblv3z70Q0FBAeKBkTw9PT01NRWFsHnzZgQV7gCN gQhHi+Idzp8/zziP8ECAMZJjZHKLd/36iZkFU7S1tWENakS96DP83KDboLq5 xah+bm4u/QFBhWVWrFjBZ26i/Px8XB6WQR5wVktLC8LVZu5hZEz9pIth0H+o BU2MzN6+fTvdY/Xq1atWrUJsoy25U7hf7MbhDuK2YpCxu4yvSHnw4EHENv2E IQuzDONhh4am0ZHK1dXVGGf69Ol9jcPTApAMUYo0pY/pod4QQvGrZIZx/saN G9yww2B4HwiCAgy1gJTCvSazrTwFGLYaHprBk/4JsF27dmG04RHB86HfAgyL 8TsRaTp0HzH7IwGWVCh+NXRhRGJUZxjPysriVuUGtDkz4ZkzZw4DGj+66+vr Gfz5RTl0Iw99Yd7w5s2bjPYHDhwoKipCFWzbtm3dunU2sLtRC2Z083aAmTNn WqCvqakJATaktYTJg/v376MNaPQzZ84cO3YMpYQfZMSmM9B5UlNT8YMbN27M z8/fvXv3jh078I9YYF4X8+fPxy0uWbIEL7ls2TJSkh4JgQNNS0sjH+yMOo13 RfsJ7Xvt2rWSkpLMzExqTY2oF7UzYUB9qTjVX7BgQUpKCtZAJPAVibds2UIP sfTYEFmVkZGBDTEI5j169GhVVVVtbS132b179/ixE++KRgW9iCpQL7TTwoUL Z8+ebQJpZhfuxuE+stuKz6TZsGEDfYmzEOeMOXy+cOECBhnq1ugJqomfJHfu 3GlsbDxx4gT1pdsEZ3uGAQMyLDOMYx+MY68tx7tC4vkofpXMtLa2Hjx4sKKi ArGRDL8N/ekpwFCq+/fvP3nyZBIaKlIBhmy4ffv2gwcP4l3wQaJ/Agzphbu8 ePFivIs/SPRbgPELEUFSXV09vHuUBFgSovjV0AUfx+3Gj+jNmzcz8rt3vsLD mIaP4KxTp07V1NQ0Nze3tbV1dnY+evTITRqBobuuEVKBcR5tWVdXl5OTQ03t h/PWrVtxeThBm4ccHsY97Im0wLG2tLSQobNMvOsXAZiCYlN495gPAYmexBo4 waVLl7o+g7Oz+dh5XSCiMjIyPHsUYmP37t2XLl2Kd3X7CS7v8uXLiEmcfnh/ ZzFPuhCCk76B+ORWQkrNnz8fA9qzHsyIYTEvGaLY+XXDXYaSt0XVhlz/cc+L GR/u3buXm5ubnp7OGPLcXmERYO6+PV3QqWxGX2lpKdYmqyEqEmxgDFrG1sp7 +PAhTYwD5UbgFuOOQGR6yicMRaeizzBeMeAwGg/jKWrDCcWvkhkTYAzyw+Dx RAzpVYABDrGkpCSpXrrpnwArLy+/evXq3bt34138QSIaAXbmzJl4F38AiYkA KygoOHLkCJ1qGGwe8UwCTARQ/GqIwi3MjbZixYqMjIyDBw9yt86aNcvG/zD3 rI3/jH4pKSn87sYL8AMcp/nuu+9u27aNu95WabBXgNva2ixoE++6RgA6ob6+ /tSpU9QIB7do0aK0tDTss3fvXpshg5fcsGGDhRqeaytTFNnZ2ViG390XL158 7733GCSHyv44169fP3HixJ49e9xjvuXLly9evHj+/Plz5sxBMjkLYChqum7d uo0bN+7atWvfvn2cQidxc9XCWAkL0/0wjr0uF+9KR0xzczMuD4HNfdFXvC5o BNwcPYpeRPfASra0mpvRZxP5MC9GxtQY3Doe6el7yC17wTDelX4Opgqampro 8NxNDAs0MVpoyZIlWMlHFdB5SMyognyi4ps3b87KyqIT2pw9DpInPwBRZUMr ivXgwQMGRn5fMEhWVFQcPXo0Pz/fnqTTDezBMRVHO/nEfqd2rcK3adMmup+9 C8APnIaGhuE3J3ZYovhV0uIEGB6zrq5u6G6IE1vCCDDGRka5xPd9MaR/Agzn ePz4cXpXvIs/SEQjwPjw8OHD4RcUja0A49y8vLzh8WacBJhwKH41RHny5AmN hdfjpisrKysqKmIwDzrE53oB95n7F3+6bNky7mJ7RlZeXo4IwfnyWxtBYvHt 4Ho+8a59L9hiC62trYz2xcXFCxYssKWuGM0Y/LEPYxqDm73HZCNbeFtZApwF esPGN+x85swZfEpLSwvjW+Lb5MqVK9Q3KysLO4QfzFNTU3NyctBOqCxcHmdR ZSxGp3puj6IH7tix4+zZs3gWW8jdVm9IZD0ffHZz48aNc+fO0fPnzZs3c+bM MDeOgTLHx2EfjvO3srKSIy5+1ZetuMXoikeOHLl7926iOUdrLJteyE2ETWhK yllbW0uH594x585owwgTphcFzWXL/tMxUO/Z2dmoTWxlmhMr0SfJE23GVbhc At4+7gFfcLkSurdNUKe3M0iiDKlgeno6g6dFen2MEwIm3bVr1/79+zEXoxZ6 rKamBk0ebwOI56P4VdLiBBi3P6MBv6cScBAbTHwE2ObNmxlCE1kYxJZ+CzDc wTDYCS4MsRJge/futfdH4l2hqBgEAcatd+vWrXhXNAIkwMRzUfxqKGLv+TKy oRBQBYWFhbZ3SUZGRpgF68DejJ7a/X4c2AvU+FY+2DLmtsDRqlWrGBbwvGSb m5vLJY4fP44yqa+vv3btWkLNQbLNgxjTDhw4wEDEKMTQbUE5fNzWrVsxzpYt W6gglaqurmaASklJ6ctROkPxQ9uedJDSlnO3FZ9Wr169du1aNEZ+fn5paSk2 sYVJLZwVb2P8Evfu3TOzYBOqHKa+WOnixYtYj2Q4SjcLCzUV5izrM4z59ph1 w4YNeXl5iFWyunnzJo0SbwP0As1EwRoaGrh9Dh48iGpat24d9qHbL168mBsq TH1t1iJaorKy0uQTwuD06dPYitP7ei/VFiIgPb2FrmIv58bbDP8/NhMbpXT7 9m1ajbrQW7hf6ACZmZm0KX3eluZAW9LctsCCEcZQJMNKZGULri7r4vDhw9ws 3EdTuyds8xUJOJhQG9+YUsIg169fP3/+PMUrKSlBQzKG0MNt/VjuESxji6bS Z9wCFLO7MB1lo2t4TWV30JEjR9BjZMgp2I1/0eTxNoN4PopfJSchAgxHyciZ aN5/0PAXYAyeCANkSeJ4wAGlfwIMkYnd8AjxLv6AEFsBlpOTc+nSJW7GeFer nwyaACOHoRJjlwATnih+NRTh7ubWrqio4O7jZyBeLy8vj0Fv8+bN4Ye16d3Y 6/aGe5na3n6yocCO8C8JGEVxMbm5ucXFxQwmFy5csDlIibBqFhelGDb4M+bj +l0Eb+HChbb2zo4dO9avX89xBj2bdMq4F34VLBLjK0ljkT1bOtJZhr/ksHr1 6t27d6Ndz507V1dXhw968OBBgqwhZk9zKA+endGYAR9r9FpTmzVE4yK06EXU F49A1fLz8+0tgDDzr+g5tt2JOQ4+ZGRkMP4fO3astraWRnE78cV3RlbwYV97 e/utW7doMpw7DhG3TstiAQrPL5Hw8gnLrF27FsucOHECT8oRexRoy4X1ZWG6 DZnTCbkoN07cN+ALeaT18OFDmr6+vr60tJTObO992Cwpa3oKT/+nfWd2E36r CE7hlsG89B9EJurLtgBAc3IVBhN3h2JtjMaogmxDMFCewdx3KfjQM7iQAuqa wly5coWB7ujRo9iEfpKVlUVRKbAV3t4StQVU7e6wTR+spva03TY5DS+fSM9I Ul5eXlNTg3ziLOy/d+9eRtfBMYKIBsWvkpOeAozBMznjVxEJML7lRzSD27CP X0UvwNAV8a5EzBhoATa0diGMiwAjByRNQq0GIwEmokTxq6EINzi/+AoLCy1u gHPctm1bXV0do7qFXHod0+xG5qZeuXIlP7cZE9Ab3LyMbHxlz8ts/pXFtElp AwUHOcIwuLILm4PkVs3av38/DvrMmTOMAPw8x18zLA/OOMnIf+nSJfNlGMGe UFh9KTZHbLcXKks1qSNVI3FZWRkekML35QJsbLTgjL33BPgUi/PzLR9s9TAO YgrGPX6VYxDsiVrDcVRVVdna+Hfv3h3krZDtgY4tKUBJKBLVX758OcYJIwkO HjyIv0BnTu1eSrSoqIhM8B3mOnt2JxftpNtgXqqfk5ODVd1SD+vWreMI3YO+ imq9d+/e4L83Rz/EIaKXLl++TNMXFBRs3rx5dTcYhw5sk/bRTqYKwvg7e1sQ W/GDhQpyhHZHcx46dIjfMuTTl4u051/0k+3bt6NSMHVctqziorRCQ0MD1qAY dA+rvu3FzP3CB/7ymTY11w+UnOPUHdPZs3UaPYyVkKP0t+rqalQZvcLW5SO3 48eP19bW0sfokNajrAvZRHdMSnk4hcFtoEcPtyFUa2urbdWNgAkupGDrdNFe NtxRHbqHVcQGB4bEpV244/x1y+VhUjq/reofRmqaWOKix44dozlsjLKdlRhI B9QCIiYofpWc9BRgeIGkWtnJ6IcAw1HiBRLqd3RsiZUAKykpGXIbvvRkcAQY mce7ol7EV4CdOnXq5s2biTBMSYCJmKD41VDk7t273OyM4XZ7cgtv2bIFj3nh woUFCxb0+mac3fUWbWBw4AZn2GfwZHzgtuXet5nwtty0eVsbPzmR8dBeE+Og m7nKX46Q0lbNwqeUlZUxLF+9erW9vR1h03PVrJjPwMEXIIfwiSEvztu6VYxC t2/fRkUw5lN+jlPBQ4cOYTrGK6rf67vkdi7WsLADJkJU8BeFRiYctKW5V3TB WIf8cFPaMJS9EoVj4tc6ggQPgs9ys9QGdKUscwpcqLOzE6dAc2zdutXG+TDT 1y2ogk1oO6o5tTs2VVxc/ODBA47bA7KeVrLgla0Phouh1vaaAB7ENtax7sHx /fv3nzt3Dtdpix25LjGgdrDHWxif3oiKxj9SvKysLNqUslEjComjzM3NpdZ4 MUzkHmz11AYmJOgJtOmRI0dITz4cxzhUGStxCb7tVVTQW2ztAu4ym7NXVVXV 0tKCeQf0XUKzg3uwRSdELdAKtAVKkoojh6gFprAq24LzFqbmg8UtwaZf7ty5 k1NMiqMT+lJQHOQuo5p1dXV37tzhgx0nve08uHfvXiwW0qPob8g2rMpPG9s2 IoZ2CG5V87gLTGHbu9tbDLbELncBtaNBgyMJNbWoPnDcySfqiKFsbj91scmH pKEKNlxw69lSq8FHhCGQj71ve/jwYe4RbE4yMsH+169fj6EFxACh+FVy0lOA ccsnwg/DQaZ/Auzs2bNDPSzTK7EVYPv27cNJDcWVyQdfgJ0/fz7ele6FkD3y EkGA0S0Hf6EGCTAxQCh+NRSxjZvXrl1rU0y5DTMzM2/dumVPcOy4u0ntHUDu dHwHPpGBwrZp4G7lIIM/P6jxCPYq+oYNGzhuI6qdZYMn5/IDk5GT4xauYUyw 7dXcwxR7K9lmIpEP6g5hw5h58uRJxlJGKlRfrJbFYzwkq6amJkpO+W2nPFdl hjJcA5Vtbm6urq52c9JIRn05bk5k0aJFITOLqBfDFzbEU9i+aQyn1JFBkkrx 07u4uPjAgQN8hUGw4Zw5c8iB/DmLsZeD2dnZNlnLHqnYc5NgfA87PHjwYCBm ImHeq1evHj9+nMtREkqOPMDrUWZagRaxtgtxdiSmVKdPn0ZB0dDuK4yAs6uq qiopKSG34ClkYm9+bdy4kWEf+1NH2p0jVNmtgsVXfMAg1jeAz/n5+dgBTWtP gmKrY91iAngfRDJSmU5I4SmAzQfDqdF21NSWxOROoQ9jKAqPy7NbgLazmGTQ 5VnEEjPidjndTXS3ICdHOE69SBOUqeZtsYmtoolBKADX4hIUprCwEPPeuHEj 5m8Umnpsa2vDDhcvXqRgJg+oL6ZI64K+YQuqUxILw9KTGUlsWMAanELTc39x K9kSCnxFYu4gGpFvbV46pwQ7lS2GiR64cOECrUxW7rg1/fkuOB7cSAIrcQmM w91KmWN1d9DBUGsMjNx0tpACN68F0Kg4Nyx2WNUFnykApQqKQ0BD2txC290G 6zlVSTJ7gkyG9BzuMroWmbzTBVnRwfjKxDm9IkRBMV5xv9grpfRGbg17QG/P CzDdIMxDE1Gi+FVy0qsA49di4iwKOtBEI8Dwy0MxLPNcYivAOJHRYGitwxMv AXbixIl4V/2XsGXYTXig8RJHgGGo9vb2QbODBJgYUBS/Gorw25/bFo/gBq70 9HRuQ5rPpqQG99ez+5GRE8V15swZxlLOJb1NuGJQ5eZFYzC28IOafxknua8Z H2yBZRsw7aVsxgRb752/7q0o23V00aJF7pe7xb4YlLgoYyn5V1ZWcgl+qqPx eq6a1Y9ZSZxla2MywjNM2WxbNzRRfgYuhAQjJ1V2yooqMwYyyiG93nvvPfdY cGr3ymCUnPGTPHGU5eXlCDDzGhiKE48dO8b4j6FwoPgUaocRqCl1xwIMyFiJ 44y09lamvaBNYbgQjjsvL4/TsQPqF1kSqx0M0QzBbVy4inlDmgP/SF1OnjxZ U1ND69vjLScMrG9gDUQmYoPWx4k4G9pTLUpLnnx275+6V00xVG5uLt9yCdsE hMpyaXoXnfDixYsVFRW4FSpOt7HZR+aJKCFNYyuG2d7HboGs/lXfPe+zxzrY Ad9NqXJycigVDYQpaA4uTW9ED1Av7gLqSwvSJSgVyegwnIXiorK0l/UH5/Lo ANwISBEuRA621+fU7rWt6A+UgZ5G05PS3XrWo7hxuAW4IrZCQmANWoez6DD7 9+83R8lN0e+5WM4C7gkXuZEnGpIa0ZMxOHWnUrZnN38xCG1NaemxFmC0yBs3 kUWksYa9FEyZGU+oAgak1haBpBvTuORpzeqmZ9vCCzT6uS7cau02JpAPHYP7 jtuWq7uzbO4iuZGAs/qxeXFwIQWzgD3uZIhobW2lp9GNqQ63s922FNt+UlmB +cxtntoFHdgmmtrNYrc2XZpewd2ExUhpCTAXXYIy07J0DLo0XQ7TcQp9gLPo S1gAa3CLcdD6gxuxbSnaK1euPHjwAC1qryc7ZcX9SMmTc0WdIYTiV8lJXwKM 34bDfmUnIxoBxlfDKX41oAIMrRjv+j2HxBFg8bVAUH4ga23WGcIDzZM4Aoyv 0BWDYAEJMDEIKH415GCI4DdgyArk3ImMdZcvX7527VplZaUtRGCrjltUgbuY e5xB9e7du6gvm4XLfW0xcHscUFxcbJuH4l/4ln8ZLe0RCYKE3Bg6SMYQlJub u3v37u3bt9u8XxK43R+4Ikds1tb6LrKysizMjpNilGAUYswhc0ab06dP47NQ QXjwSPfvww7UhRxss2Zbi94cBIMYFaQzUxEq7uaj2qrslJkhyxZ3YlS0hf7Q olYRiscP/zt37pA5f8kHu1FZam2z0RhLbQMUjlN4WweDI1jSfBCFsTfWqSbH cV6cbquNman5CptgPYbN8vJyhkpajfL04606HCUKh5LQlNYWODWKytB98eJF sm1ra6PR79+/f+zYMZtwS00takdlKQYjPO4GicVni70YtBT2IRPOPXz4sIVc zPlyOpVFj1FyLMklKIMtjE+N7BkTV6E8/GurHuFMLR5or7fbAxeLqdJAdN2G hgby6UcUC73U0tLCJcifhrAeSzWtp9FRzYuRgJaiV9MibgcTGsUWD8dQtDXG vHr1KqbAgLYvibvLqIttvkkyTB2UoHzmiHUVsiKlOUdbqQBzcUXMSCe3Xscl 6DzWXpTTXtjnXsCAZNIPGWYWoLdj5NLSUoxAx7OHnuSPnTds2GBTCumHFIYb k69oUJuvzk1NARCW9BDbnYHS7t+/nyPWpW06Jaajm9mtQY/CuVMFmtJi17YE ARlSaypCUyKxKAlfuVvPJuCRA3bmK0qIcTAgRqYtaCzKRkNQHf+7wKJ2lIdS 0TqUnOtiTHsrgUtQBTqbLVXHvzYiWQegt9vb1rbZqIXrrRicTke1yXIWz7eX QDEFNqTPYArbLYjLcSHyJCUJLEzNfcEQjYnQb/yqpWPYagy2h6kJTkyBEbhx 6BX2fNApK4pEFS5duqRNnBMcxa+SkDACjOYbEjt8xYR+CzCGPka24TG3YaAF GE4t3lV8DokjwOI1+9HkB7LHdlTcu3cvpcUIJgn4kFACjF9/May7BJiII4pf DS34efv+++/X1dXZK37O2THOl5WVMfQxktTU1NhsKO5WjnPbMlbww5b7/YMP PrAfyAz1jHLHjx9nILXVrhj3uJ05wkB069Ytbl78CEPoxo0bbd0ny5CxgvGB UfrAgQP5+fk5OTkMCPaiMdKFIYjPbqRi0GD4sucONsDaskiMSAzInEsmVjBG G9u/L7hqVvhd/Bg5qSzjGEM9F2WIo9b8ZSTk0rhUakEV8CZunSub6k/ZGNZI YFPO7AkO5eE4FWQANE9ql7Dpr4cPH8YUlJz8qS/D3blz5xiuyaSxsRHHTRnM a1sEj4GU3Lg0CgRbcRV77uCeLPCXxBzZs2cP/p2GwwJUn8u5mWlhVgyzhx22 izeDM5cgK6yKkTE47VJdXe3WSyclHxjn6QbmTG2hQtyKhVZIU1lZ6WSVwbd4 CroB16qoqOBfm7ONBbCSrVRPT3MxN3saiwsmf1wqJeFy5I+RKSRfUUFbK4PW oaW4FrmRLRqG8ttqk7ZW2HP3K7Tq20NPmqChoYGGtpxNFVjO2JwOZlN6KB53 DcXjIFWgLpiLNqX/4ODc+my4MyQ3YsMeCWFMikq/JSVjI/mgHzDL1F+GIxxH g6G+SEl6LEznp4kpDPcLl7bHXubube8qbqXF3fCZXsTpwer3VXe3uIQ95DIL 4LLxuZTEJI292Wpz3mh09EBRUdH+Lujh1gTWmmYHDEVrUgWsQUOgEGgXe8EW zWC7e3Mj2HNz23OKVsOYmIieYwFzsuLe53SSHT16FAXiepQt3ktW9r4kDYGR uTr3I/3EHkSSnm85t6+691xIgWLYayy0Dj2NbG0JUHvZmQraKxv8pcXtjrNx iduTfymSTVa3xRP4ikEJQ1E1/pKVTUe3Ve/oUZxoz9MpJw1q733Q4bmWbXLB TcGAZtF468CMURSP2tkcTi5hI4DFrrk16JkUBju4Z818ZsSgOnz1fJcg4ofi V8lGeAGGuxweYRkfohRgQ3oK1mAKsHjXtReCyzoljgAbHHfZU4DdvHmTstG+ 9rqB2ynPXnOzEFZCCbB+110CTCQUil8NLWyWMs4uZGNQhilcIXeubfbBT2bu fVvVCl/AzcsQx8jAyNbW1mYjMMPO3bt3GTBxkYyWnJ6WlsawY/OLGCIuXLjA tfALtbW1/Dret28f2VrwHP/LyEN6bnx8Bzc74s22OmVYsKcPFl23hfVsWUJ7 4zunG3tSY48YGJrsKYOtFMQwyA98Ww2eYer+/fsh0QyqSV3wfRSYHBj/Octe gmbwZIjDreAK7aFYUGdiNIqN5KBqmAKXwb+ksW107E18e8GNHCw4w7VIiZPC sXIct0JF7FkARuZa9fX1lBND2YJRmILykIysqDW2ooKcwhGc16FDh2zBQM6l slQZI9i6YTYpy941Q/+QJ9ftudwi5WFcpeRkhRkxnb3miVBkMKdqjMm2NrhJ AkpFIbE2IzxulBNJaeFKFAVWIkOysrfvnaFoO4pET+OKdBK6FgWjeJxL4S03 DEXfsK1mbaMZehTOAkORxrwYZaOCZlJcMN+SmxmKi1Ie6y1mT9u7loag6fFB PR9/4Dc7OjpoepuQbPFVymmzubCGvbbAhRjKuBG4EIUxQ2Feao3B6bFcgq9I QNeydXdpbjwd52Z2QW7U1974o2wcpyFs6hR9OEQ+2bY49uIbPpFaY1t6FDms Xr3aJCsJaBSTHzazGlNgfwzL7Ub3sBuEz+TQqxiz1VBpL/ob1qP6lI2GsPmN XAgjYA1rI1s5gZ5m07Eokq3kBlwFOURX5FsKiaG4FzApJbR+QpOZPbmb6Ipm KNuRwYYO+hjOnTTc1+RvVbDuTXrMSEpalp4ZXBTF3qWlw1gL0sQ0ClfhdPqt rexKblS/5xINts005aQwZ86cYXDgWtzylNaeZtrYZeshIOrImVrzgabhuAUS bVMGfnjavWZDDVaizPQQm9xu65aY1uJE8rHXfhkAaX0Lw1IF1Ca9iH5C37BN lLAexsRQ3AIWgqOvUmZ3I2CN/4+9e3+yszjy/P+vDxLgZYU8XO21/bUxYQ8Y DEiIiwAJAyHEVTIihIUCBI4Bcx3wzHgduxF7sb6v6Heo4uzp7tOnr+qDnvqh 4/Q5z1OVmZVP5uepysrEGs1sgZFaenDogA7RM2vMK4vTnvIe+Iyp7Vub1q9u tbYlALtFjpzsHoD5dUXPWh48ADs8LRByOAHYblZmlud9PQAz+wEwH6AgdgCW MONeiFxTnaxDBcCIeruP3gTApnYI27R+tVqtyibsVSvbI8MVZ+eJ81B7ciuO wyZ4VItoZSjYDeaO62SCZjPttInG7TJHTGvRIOynHviOIoIYgU7o+7Jir0a/ 66677rvvPld26pDpuHLlCvvAAletr42YMmVxW0bvPF1htD6zbL5nqTrp3CJ8 WbPKvDfO1jFKo4pf4UnhB6bPLZ10xi9mL168WLiyazgahouJ0/OQ1ZEjR1g8 pP7lL3/RFXuIDG4Igzrs4Bum2ExinM3DU7S8B6QFuuJdGVvs8CZlI8+7+RcN eBxZGXkNRh4xOCLq5sU1FWQpD3wiZbSJTp+sLvkj8q9//WuZsv7XjWa+DMek c1iux52/LL/Oi4YdNFcNhwy5y3ayaAibTFZsOEjAgFcsku+osuSRGw0xpsY1 VKVi0BwEmo0CunArpowcODJEzsULGYIwOdaittqQ7TSiiTMcjjgO/9KWgoRH nRECMR1Goef6mWVcg3Z86ca8kv5HZi0MkhuxQBdGMRdhbKI2F+m2IThcamOO 9DZ71kPPpAoydTpS51hDpGeNzpPh9TVraVy8JKiR4apdoYwn4TQX6PSXiLhO Q5P83IocFfVw8cWeUJNOk/PatIJ+plFpezt9KDRfuKPkevNQ0+3ybIRMSsKW 8iCYgyaotuEqImkI3FH4aiAmKJ37i1poyr33338/Vazcg+cIU7MrqGlUe+7F bZKPucZ+BwEKukatB3yUNh7J5VzfiVRaRxq97LgFMiFDLFTFEu9jxtvi7DBv IfqUkGoRbPkf2r+r+qd/fTlC0KFrH0yfL9sSDUF5zE1iq3yeMvTjlKpTRVdm tRDPFFBFwkRexxNQ0sob4ZeSQm8eQ2rDTBHjbPSF6z28CKiOgxHNtSe0vLWt dnp2sE//Z405dpBBMoc/+ckt3qb1q1utLQPAbjaN+974ejBj9wBs5c5aFnjD zh8wALu5C32zEUeHH4DtLdezCZ02A2BmCsgPvAHGwVcEU3IYrG3fQwXAqOhm j95mmawmADa1Q9im9avVap7QlhFa9y59OvtQGM/p06db6/Dced7Z2xLoeWxd 7/ktEsabo59Gn6XiZJn5lHwNX6xbj7xHmGO9du1aASHlzvLgszklO2Ik+ZoO s7veKAwLo8SLMQust8uYZVcyGhUxLJtWZYWZoHE2nElk0NgrRk8/vi9rVps7 /mWmXMNuV77kl2uNqfQZR2jrVDgK8cIIl2TVoK7pMFdpvrgYJKGtuG62kZcM lqDZT/wCOks7P3uaqahdgiUNniJPytC5GHnBNkP/+7//OzLKdMTSwiqcnUHL +JotRVWZ7blCMudTKteSSFsk1G37FKajQC/9cFK6infX6KQDmN98802ucNaA o9NEuJ5bMZXGatWuoHd/cY1ggPyVV15p02fEC/kXqTrXLW3hE/FlFH+5P9JD KjrJCkkVlxzjlsFS/74nE2ymUUVYGat8ULolKH0yO77hvyhAMTlVb0QGsmk1 yVRehA5E5EhEpjdDkHNZ1FoY+fLLL31JaK75yU9+QoY8Y0faSQCp5nFMa9tq pa4qa4ErqYTLPEq4IJ/KAPGzHh8ktUdWfLu7dI6SQv0RwJPiS7fu8tkTFA1Y 8+CMqosjDNLFVKVUn3ou1WfblLgG9gjQRBuobHVFpJczwU/mqDyf8JLpNpwH qqO7dD7HjWzf0zGmgHASlOZf95aToeRsnkQiTZ4llh9Qx4cKGpZF0+jmxVOG BX9dHxSnHpRKnxS4jVR/q8jpMyXsqGwFenRoFPJHvwfKNXBLO5ivrLVCGQvU rFQNSaYY5ENhSIBimGs65vuOM2dhmMdOMeiQ3SBJQjaJ5ppdKm3dyFPhys4w Zr5IhvaafZOIHRJrz7FMC1kVhoj0MO6CwjWHoejACKoIwWz6XKZQXGPZZyx7 +kwZ+2C+Mg4INhFZLX/N1IqGKNwibVq/utXaMgDsh/3MsmCMIcOVYScHTmrH AOxmc7Nsy6R3TIyLPGAAth/lqpdpLVtFLbXnagOfQdPDCcD2hOveidBgRIyH QzYDYIABGFC1IxjG41AcUVd2Zu2wAbD1xzpasCJzL4+owg7dmwDYD9uYr3qb 1q9WqxVh6wFnMSr65vFnRnpjZUwYfFauVax/3Giefdd38NxlzJdv5h5M/xby XUHkUhl4nFmJdgrKzOOy8WIOhBiR2bnvvvt+9KMf3X333QXQ+p5VYXn0Zqw2 axgrxo0J7SxzR79ZCdf3rspaurGoVKZDV2XxuvPOO8sZzipmtbI2vnfBk08+ qfN2HKrW2uswOnkfLoZJLNV8Md4+E0UBKgV4u6UAZtzhC7UVwjAET6Tn9ScC XM9fuBLxusU7Snh2xHcuOxjT6gQJGKuDclhuF4b3QTnLX+6vAlldVrxxKaRc XPSazvkOPloPea4jR44cP36cvSXVTvDNbdK1h8KhM+DVScQ1O9+pNOSVasxn 9/KVxi3ct3VCbqWDaZRtnGEsXihplPULPbfffjvGeSX4bT0NJBCS0SElueee exKUbvOh7cGhBD0lyMJ+cdF4RHaLnNXbHYWV77rrLh+MTuCXL18uCn2kpKDb 165d81MIpwQCVbcZJ/hmWwlI//CHP+iz87MUviRUzfUIced5zSOR6o18yN9f n9FMJ0tkMSulzlTCPwjgc80pOXuEZ+FoQosG0IKiVvK4fJWm27zQW5+J7o47 7uhZ8BzRB/rTQvFXa41OkgbGUYXxtrGq3sLR0y5DRFsmAl+ABMY9wtUpNkoY I3nOSqmNP2Nx9/o0BWUta7sZ/WNTz70ZnM7MpldI6jNpEKNrXOyuEFpRDYyG 5+7o0aNlxHJXsXkIwzXeS7LnGjM7G3yOu1KGlo6vzBIujiNXekzItrcMjyf0 gnIU6r8MsWVSRaEvIbd//dd/DWCPJCcwT0WWq83aUV+Pnp/Wa5RJN8v0xLwX Bk8hRznmuG6L03y1Gk+FPKeuJCKMPLLWzOkEnw5zm9avbrW2DAD7AT+zuSpa SgjlQyYHXjt/13JHYRLX16o0bgnAbjZDW7e8QPt3aC4nDxdQrqEq6ra6sv6u vQVgB8bviDsKTpfaKABfatNy2KLwEAKwHfA7F2rVWTnO17yUlN68wCHI2xCA kU8AzPyaL28NpVipnEFH6g4nABtJzDo84kssQEeeaGQEqCYAtgOlmtrBtGn9 arVaK+Q8msfWo8dQMC8d6aoyxUh8PfvcecC//fZb7gN+KM0Ue6KTuYIdrU3p n5Fhahgx5qgVdfaKNfCkwyRuzMgXU/r5559/8sknLoZJWCoWpqPQxZEyR+xz 1SJQ+Nlnn7E57VOcPXu2U9st7LNj4FDJBIxbSkOXcRAsZFnQXYyFzkyVCzQQ 5Xo2hwnlPlg/xoqnQFvntcmHgeJJ0eMb3xe2nYjKSXj9Rh1kTHE0hmtzBwGo nQNaJXHl4zrgX9G3NjjKtYhm/AbkspYGLd+R3qroyqmV92nErVVhhDA9hqRa VK0rXcZlVE2yYN2RHsFd58+fN7M6L3o2XhCGCzIkK/SgEJ1zkT+xX6xU2y7m F/1Vh3z//fdLEbbhXSUr4Hyr/WHWCN+gs8HztZJFIA9fZgd8pVHY6UClb0bl QQ5Ln4GlCtq2Llr95ZY98/LEle82olvav/OBzOkMftv2agUSI60QYnPunGPr S54jlKAHVXSytGOzB2zHeVvckZKpMY9kC8jxeujn9D0X8T4rpR5D/hrvUFbl OylqlRSaKVwTvkHNF8gH7QMnR48epdtVFq4Kj2cQX+118v6uR0YPKZYRU56B UhCQD8Kokz4NjTZDBALJyrR6BtHjaSpy0uND4MGG0s2tV/gOxuq5IHlqiYBx 5XiIkuqAHB2kBZ+MVXaCGqGVw8Gj7dkkk56F0jWUNK+jxG53Ac2kKtTYB0Oz G7TIFJMV4fjbwVujYMQkAmmm1bwj0kD4dTvpFdxYMoSSxntyqS7eOwWcnrSk 5kO7kLhuwx095qhM+wRC6+ayzWdDOo5daYxyXMxGtgecupgAM+YmhQZSbzcy AkHiH3wsx6q3af3qVmtLArCbTea+tFIo8H1BF/YKJsy5zLI8fEHbfFR6MQA7 tK3XZ3bYRAMSpX7lOEpeVOAHdzm2q2bbfgCwg2E5tFYJcn4Z1wVym3HuFWzg 0QikYwvYb0nq8ACw7bYAmD5Lb4tgoHpw3VkJjBOFSXcBaucAWEvZnS4p42u1 xf1U7vpDC8AgDa8qrd7AQrHQCRFcGKvjkBMAm9rhbNP61cq1sTZlsrzlsX6l thuLMBs+cX5lHHIT9957LyDB5s8Bj9GqiMrEtc0xapuy0hyHJ33EYtWYmow5 W8Tilbn9v/yX/8J3u50B5MLcyL+wSC3mFESqfyRhYaxUAIQsYafGmCmWhOmr GCJ/2vnoTltX3cNnd5VfK6/B3HkvJhD8trNTiWc9I8Y3//dG21C8gS7j4oJZ Lr4Lg5tBCD+1YdFSP2LYTGR4MTd0aTyHtSQ37gnW9Ws2eWQS83kOtLjSFOB6 bHtpFZXDNfLaA3JXue7Z2yy/ez/55BOei4NwMUfv3+rfLdaockISI0fD9Vz/ f4HW+lbcu1FQxUu2bLihBgZUCIow6d6xY8fQb9aq1JNG5bP4VrwgnmurnmMt rk1idXmwrLeRZqG88Um1Crkuo4rc6ILCyuVR5+A62oB3Kjcbh7yhoMxpJRL+ 6Z/+icJ7Uorx3kxKRZdRy+oUmDWuvCOZxjLX2OemAYPi7dOiUUCwkp1kBUHR ZCrRimivErpCCd2jQmXBciWlpTaemtD1oLxFV48e6VWtu0pAnrVycWxY+y8Y X8p9jyrCejbXZ1mflVKL4bg2WabSvKAcMUYpdrGdx1hmWypcFde3rzUclQmf ZCgJrXBXy78ljkNM8ZklXiim3WzquXfJJNwBVYrnLrPsyv+61mChlpor2zp4 b1MyM1Xh0VJSdKCVEMqcsCHvpQwtBM5EeGzRbIo3y+o8HhYdssxGMaIJKk/a ZsZ8aoekTetXt2DbGQD7ATTug5NlXQc04mUWlIutMaTkw9QHwLKcCwDYYWjB nhw3v+P9mgsrRATjHG5BRxvOde6yJKgDGu0TANsrZkcczgBgpc3suMSxtcZH c4UgVnW3B1UF3VXl7XACsLkeZuOONgNguC6rVTX7vPiYTa6/oyuFpQ0A1ikM 6lHEEcZb5fNONFz/CgGwXtw6ekMHXNNp3wmATe2wtWn9anXbhvBpQat0jotb yi6cY4TmzrXZXE/V2uhQcBkM2AeOiQHPMbUy45HPNrK0LBKnX5a/Tj95Fe3o H/tWWsIvvvhC/4xbBekQxqpXwbA4avbqv621zhPxI0w908qa+RWEYEtZP7QV QsyPzGaOwuNvbrRyzvuGx+FnW/TYUEp5GYx0QDv/xVmw3ht61Y5BMb8Exe6x 9s8888yo5Mh3oBOznShsd6D81THOL5BkOzhFcJVNi+snrnwEg1+iePAAGZU4 NAW+wbKfKqFbMjHfMPg5R/+WDLx9ii01aj18WtxoCFRmCnBqlhHAXfpmQ7jy j7WjB508LXN7x+rLVEkb/Vv2J98gowCtsToX18XgdRdZucYtLiOrtBoNxAIM mIsRnbX+0QjDuIZ8SImsfBjrb4u5Xg+fFqenCDJ1PvTkyZPHjx83xbj4/Vqj nL+40VDipaBCTrgmpUoS+5Iv/ulaSyxliihwsTOGBBUSoIfffvstOXf0NVyH rzbj6EYbi3pAjP7LaelZbkV6PfG68qgaglb3CNMNXG8GMoMf1MBd0FH7qj2z nmvzW3KzEinQnNLf0WfIwSyTUmnQyvzfLJfxLAmUUMJdxZj1bFbEp73FAc/M jstKsNDmHfarjOyCCkNgpEXmf9wIVofP3duBoOal7HNlrphLSTHHeDC4cxMm ES+bCXaubQifpnbI27R+dSu37QKw1W1lQsBjAfllRvUWuYy73HD96jC33CUv wJIPsFGVH07ZpJekfTMA6V6eghPnv/CLce5pnwDY7lvHKD799NOLFy/C3gAJ pxwAM9ElogRQccRdluWsxEHXb6RKQCe4NXKCHUIAVmsLtWxaFLLkJ1sCsCqV m3GEkYCpJKhKYg0A1ukJjtst1AbQcnHln8Yu/woBsBJ5Va6r2P7CBiYANrXD 1qb1q9Vt24VPLcizomygB5xNg0ZYztlc7utbZw890S5muKqmxwx6Xy5v3lzK ypZoyoXOnpe9p+0MN6L21KlTzDgbVdRNtxeYpEOWrVQAjK3r3ViE1QMPPMDH dfacCXVBCX/0U6gzPxu4uu+++zoirR07dqwsXkWnsNgMI2Nb+Pqo64fgoqRm V+OJlGV2u6F5Ok58wcbQ9TWPjOvSsKPQiAAbT8Hwvv/++4Ts9rltxwLyOzXP qldk9rbbbiu/k88jIoWXYVcZ8GpBmnqMsM+gDr9Q9FqhXG6sH86de6rUyyyb CzaqdgafyLMQIJPF9fNKm20M1XgfHjkPRR/wi/LyrbWHdeTIkbztSy+9RHQ0 tqXUTiIYi/a2O0Y3TDH2De0uPZhimtkW4QKfVcS73qp9TEN4vc3WcufaMvAp RRpVe1oRBfg54nT4n9faqB5YQonyfBZuVGJzz4hH7/Tp09WvwaO/rvT36NGj aThUQOz0U/+zko+GHi4/XblypXRSRndXO2v6L6p/Q067nX56lmkymbuLepfu Y/A4l0hBhx4EI7rLI1nJBgzeu9Z0Ms5FBpLJBHYCC+O6kkNYJh/8mlyWCsGj RqcpZlJInm7TPfrWSl1yRht7RZPR2Rq4ewmqpA2sBKPEbsyVqomLFqJD8m5E IVGTmBkpd8cCjUrOVbKuADQoWBHJLTXq+gSfVrNN61e3crt11q9KMw5oPbjW oBEWckFQ+mxblfWr4a9zl7zASNoJVzz++ONvvvmmWR4H/zfrgQvjYnA6mxp0 SRoOZv1qLBQUuA4DcFVQ06OPPsrf8dQc9EgAGwCbg755/ErPuCBIVrQSNALx EtRsQZ/17eABmLsqII5ZFD799NMc7kgTuhkA63BoaUAAoaKPKENFw3u/MMvI aEFmfWrToVGrAsDAJyO6HaQkKD0QS2UWJwA2tUPVpvWr1W07gE/FwLiYiSiu iRk/f/78AijSqpe7yvXEiXNqbT0wR4899phv5iJ8/nEj4yUbxZRVopQxqYza ww8/XBqBzom/8MILrSlxhfxLVQ7dVeoqToHtajOLGbxjreVuCjeqBAYWWEVm 59q1a7grPSATWp7VopjACba3M1yIrx5ci2mx8Mknn7TBNHJXls2+mrau56nJ fIGc3VhNvcrJ8QuGaDgfsF/01zDCebdx/I2/ePLJJwuvveNGQ/bRo0d1cuLE CXSSSdt/1foxKWVcLEl+w41zZx3TIyh+k7th2wsG++677zZMqrAz+GSiR+oG LsCkmIsNoUguxswSDmkUJAaEtBRTlW2OEg3818cff0wmY92PdpWpgEsiKIpE e1MG97pRP0U+Y5Y+0KgNE5/m9EmMy37kkUdcb37NPkmONAuL22L4FKJAtmtM N92mhC4GaBFsck1oMwv50MZOBVJdMAC/NLAKj8EArwwdpaTw2CznW0uaeiAo j4Yny1jt983OaeLCWqIOdSCDGrAYFRqo3MOGbP5jLTkAGjwjzax7TYq72n4N bJT+rvLHpIGdyiW3W4dfIATx5EyBmQ46zPjg1xTQMX8RU4mfTh+7HfwoBWiT 63EwUwkqvIT4tjj/+1pjYYiaBpZjtgJJ8F6xiEYs8j+W0Tyr/BmrgicrctHu pCeRYSklRTOy4MH3E8lQe4aRrUAGZN5TtmT8+QSfVrFN61e3crsV1q8KR2c5 O4XEpLOQjDa/sKRlW5X1K3Tyv3zTSNHJaXZGkkGuuFv5cDZjvJwbfNwDDzzA i5XRaEGk1vq23+tXATBsgotgQFmP8pXGNbNj9cYUu4yTDYANllsqKZh8NicY x124SznBtjwDePAADHdILd9UuXkBEi4evwsAGLzx9ttvu4ygOhpQEW00A2Zu nM3aNJcuqRNwJZhaIQDWWYaOkFCMqge6kRi9PkwAbGqHp03rV6vbdgafWuVm TKq2wFYwj0uWOHFN6bVbEbrrrrvYCgbz/PnzbMWo6bB+xDxOkbfd22tpebzL yMcLZM/b4fq/N4rYtjJTPJLmrmKNfLj33ns7POUCFphdZe0RaTj08KTsLeOZ jy6pTpsI1fUDM6o3kddGHmPurpFFiljYTzS7xcUZ8A3DoWdbK37Mu6FbeeNE 2twxBDNb3b1xSr3FGZxyDUXaFD9W3vICjaq0QlDmuv2OgpMLpiqNANRhOL6j JQ4+C484bRfD6Ca6YLB2x0JWs0FZyNsBfIoFo//+9793O+LPnj3bItv4tf2+ UAEJ8ODFURcX12ZQJU7a3iW6ClKP7c5/zKSkuHr1qiFw5MbyMySlilS6vRBl PrFMYuPw5vUbyS0JgVZQftdjduSJWqbNwicetoj62eI1LfZy6ORpyigYTg1U Jc2RZMA33LTRC4q+fgMctqRZvBmP/Nxzz3lkqlzZ7f/1RoM0aL6BWj2eS0lR aJ9nk6Dc7t7We8vFsVlCgNlpLWUoQ0GwxsIygFfWiBIplOi1/TK/dlC3pbZa BBudBrrGlagtQo8O+6uTjknSfIoHNY0tQoNWglM/bTuW4W2EmpfrQCfwhp5L u9qz44nzwNIx4sWv6Vh/fCMj034lGuDMFlQ7oUyFCL/MCYshUC8FCABTaS+y dVUBnSU16voEn1azTetXt3K7FdavSoUKmTDsvfjDGLzVMqfhaod8/WoEI7VY wYbng5hi7sBrO7DBGm/Jr354GSjoiSeeAMCgC5gWAtkWMfuxfrUMADM1Jte4 YDAhtBe2YT++LycYCgEtmIS7BOPLU7HMquZNB2CVhjG6e3lq3hbZOdxlABj1 AHQB6arpbQgPhkYxC25fRQDm3maW9PzkgvKpJuoJgE3tMLRp/Wp1247hUwtK vFiVWH/xi1+w5GzFgr2AcWOpC70X58KeeeaZTih79k+ePMkrsbdzJwqv36j9 WsgQk4tyV3ZqnrvnXP6/tcb6VVtZ55999hkbYiA2kKNk1gwBHlxda27nQaAL JhcoGjmvfvWrX7Xmj7bCjdp/4TUK2fpwrYEW3utZSDafM+Ia6oFf85cV5d34 nUpjGOWxxx4r2BglaPvyyy+3XMJqASFQdOHChXI0VckR13jsWD37jFPoDoX9 WnA+ytsiuXTpkosRM3JDtXlUivic79dff603dJJS9Wv8RMg5Jrfnm9qXKe6u 2Dnd+pWIuJiCzXYMn/72t7+ZLDe2zUSqKO+0I+WsIIieO32PBnPt38KJIZ8i 1pBdYFXRfWbHTGGktKI0FrXIriCjOSqPJSkRMj9uTomldJplTCIogoXK/Io8 ztfFFAN5PCyHizDfLzgUsL7NwqcQBVl1WhaPqSsCqiyJVAT70uwgj94StQeW 8vseTjCt5UTVLVzhJyqHsEKv68Ez4gl1iwktyNAts9vELqjCUTVZ4jFJVkDH vHTAsOyjW0KCMpUZtOwQKMEpLcWgKaNgbUe2m9mMEzUL0GWVPjShBZ4hwAUI 6Fkoj1kB/GWQKDNnZRMJypwmKM2E6rndQHqCftpuyvRTwOGg0GX+RSHeSYna tJdasrs5fv1b7VSSQXbzhQBUVQPUvezVloKiNqRKu/DiXaCMENRjsxQNm7UJ Pq1im9avbuX2g1+/guKYssrcl8+5SKQl42Rqh3n9qlQGrHcYb4CrAJj53bBq 3voWJOC4RzpKuGLJvEyzbW/Xr1pk6DjAYgAGvXRwDGQdaTNn+ymD0wBgOhkA jCPWP3e5zCqfxl3edABWTcCyiER2u97LADBaYfTyi67HFWY8hQcDKgndWY9V BGBVD8RsYXgJqqPEEwCb2mFo0/rV6rbdwKcyL7H5zE5JoRm0KuEuc3uhrcwj W6GHTs37wHLyRxvmepptRmGdmMdWZrjCe+65p4gLzoJxriA1+8beljTeNbxt 5XpLFO9llgn1ehu40kMxS5Ut44AKwrl48SJzXTodBpY5LcybLzYKV8V48gUP PPCAHkaOHRezqK7nQdybTzx69ChOSyeeD5rLmrW+tVTI2pcb03wRVBCF8D/9 9FMGHKctwZEATvPpmX2dl3ysnGD4KihLJwT1zDPPvP322zph7a9du4aRO+64 w1QGfpqjUikmAezQFjQUgUZQPCBKOvyuk2rO5oP8W1zWkhrVEqV+mi+dU4Zm mX5WEKR6gqNYiQkKvfztb3/TQ9kJEF/2RUS22+t2845BkkdtGzR6QKcZJMBq DZjWFnZat0mjXJwyvP7668RIbagEQZkCkjS5HVFckse2IA2E5lOnToFPem49 KgnjEe8jHqwakSYuCZspBJRa0wNr+lBoHskBWhinBfFFUIUeVT6gzLEkSZcK 3vMZR4BZkeEuy+8bwlMTjxjXA8U279B4WSw242s2kYL+Uej6ctCVzdVEmNCW YT2M43wfdXJNQfWt9xrLw2XqMUtvCcotHp+i1n3z0ksv6Rl5nQjuoSiH/wik xEVgqVVuDBql9LlmmUrrhGBxp5N6aK2yJfTNckrMpqQwBQReLgu3Mz6Ehv7q li6j+ek8QwQYM0EmHf3t1y+pUaNN8GkV27R+dSu3H/b6VYH6eGSXOovE2W1W nGVBO5zrVyNEmZvjSUeJ51kAtmT8Rms7vBIXzOl7nd/x6tOerF+N2JhqKmFw SQC2YVf5uDI4LQBgy6xnJnM+t/UKZCDgpgCw2ZD+orZKdbIMANvsFSmZ06hv vvlGJwmKQ69w1YoCsEHb0CiPP02YANjUDkOb1q9Wt+0SPhUZzoB4YFkhdr4D 7Mvk5Lx+Y2WGl88KvfDCC6xZsT0skn+rFrGha+uEXZtfzGyrSV5UEdDOQjsC naxnHnXLAXk979x0TqccUKOCIZvJcvKnnQHXCSDh9kpvcLgMtc8uaGeNBzQ6 N1H8kh58iZHitc6ePcsUl0a+uodlVtcJWfEmPBpqOw/edkzLWRsKCtkGqvgj S8uV/PrXv666YoElOuR5uchqkcyemyMotrRQLqJmWrHAd3Bks7snrUOi7cyZ M+RZ8sxKrow4YWz6CdAiSWS4EuIqrKstuZI06sQ3JGCsjrYtmcDBtLoeBgMA zBqHUnJOhOHXQIV7jbPwbeuMs/DlHCuNv6k04xTJvUWLtYZZ9T3El0lsbHHm 91vFolF6KKSZx+cTCdztJXlAVZF4vjd9o5jO4jZAhSnAINHpCnyiFe1/jaiz Z599FuWVAuw4KvVwY7lAOzeKZf/qB4P6AUVQVQhiaKeTnqV7IihKTiXGmrAe ym1VZjkiNS4J66Rka8ncLOjkgw8+MO9zWSzm2niQy/bgrpJFMC9Hjx6l86Cg ng2B4LJGUA9XopDG9jR1/resET7ryiSadPIp9p5e6aRMDibRQ2ogsBML6UPh iM1pgsJsU1kG0cIjEzi4NVKPeqyyNqYep5uBHx0So2sMigXstA2NKcbHs4/s DpwumUvQ9fSQqaGZJZfo9PEO3mQn+LSKbVq/upXbD3j9Ci+cDoQDIdx3332F vi/OXr5ZO2zrV1jgID755BMool0eno4vaI1iPQBb3AKxHIpOquYGnu1gla+2 +/WrkpFyqWPLEmFLArD1raB6YqEGvK1+Kjy3HoAt01xf1R6aALHwuTRqweiz bW8BWL1BAu2KVtV6SQC24WlBP1WasFJ9QU3yr3zS6gKwaNPthx9+eOHChYq8 l59qAmBTu+ltWr9a3bZ7+MRGscyQCcvGArC9r776amE/S/bQWjr7VpLDe++9 l6n0l3ErofoIsl2wQfOPG1nly3nO5Ja46bbbbjty5AjrzW+yb1WgKO/T+qin /FGbRLjolFkl/EqBriuEvfTSSyX3duWIZqnDSglzdmw1Fw9xdXSdua6fkqJX L497YjYNBK50ojy7XQWQkVRqdlIMpHNcADnsP3r4X72V3orwtyzaEvQq/X55 6ct/nqCOHz+uHyRVlHaWjNFDGxYEBb/xRwx+JVGKXiuDOmYffPBB7qDkAKO3 zSoYDjzgGhe7hfdvBlGlQ/JH2EiFavQFkV2tWsyug1WRsGYuME6GWwarl46V B8dj21j6GbLixH1vouc0atAwskaUDKFiBGb58uXLsGUhf2WebwOLAE0HmAcS eB5pl+tn1/1mN/tMol91BQ/AS0j6p7VW9CDVNYRnYaRl2Kyh3NTQVTichAd3 9BO/iDEEFz+bfDWmZpNFVNuR30e2Ga/ogNubuLBT+d9KFucCD6NB6WFx2v/7 Riszg95MH2TVcVRgNU0o+0RnNJ5//nmKSv46MQVzUeJRGHmdP2UTKrKJKjz6 ixgPoGe58Lwtw8XrsCIUkJvJ8rwQeIXR2dIKNyzoYbal8x4iTyKO8MWwFHq6 ZA/rp3KCTyvXpvWrW7n9UNev2mH5y1/+0kmx+++/3wvmloWqN2uHZ/1q7HvO hShz5RDC66+/viASabPmeoKCSWgCWXHorPeOKdzZ+tUsAMPap59+WtAyAMzH FYyU714GgM12yF1evXoVsKfkoCYeCcpUlsR1mY2/WqgDTj579mxujqDyuZsR cAAAjIRhYHOHJPBGP6CvD7R9SQAWhZ31g51efPFFr1Fl1oVUzYIXjdmVzFUE YHqj4W+88QaE86tf/QoMK4MrBicANrWb26b1q9Vtu4dPxUGxvR58nqVYWbaR z9ruElYBQvxawKBSNYwnh8JOMoDff//9giWsMmsFLdh8wMm92Vt+ilNo7wD+ OXXqVKcL2zyaXXPI3nqjx9EozMfOF6qkN678V2utQA6+mB/EeOVZXVx1DLZL zzpBzFdffVU/ZMLZ8d2dl6xyrs8///nPC6MqaxbLzIuRITmwriNTYptHjO2Z M2e42mNrjbvkXErtiMdycS8ABiOFJufLQxEU8Zb8XCe65VyIvaIh/DJvWHjY XL2PPHhBWQkKqX/605/ItuVHMq9QYwmU/uVf/oUMUd4uCaYIp04IClR4//33 C5kmZzfiZWShRxX5wAMkSQJbrjvps1N+J06caJWPkuOrUGe9ETVXRdPMzgKN KvYMeXqDeUwKRgCeAsPMYOdM20LihSlA22T6LG1XEX10owQg0GBRav56TEyc 3n7zm9+4V/8ewLafSKadvrEgFhhrK7Mjt+TcoQyiRgl68Kg3QwQqOie7oFJA muCJczEoMqRNE5JVKzy0EQueTfRQm4qBmsFKcxIjvkoWUfo4zYMWUtUGGqRF GCxFBrFXJxSbMQWQUGyCwleC+sWNNkRNJUAgD5F+3KKTMkjMCkozZR1aMR0v v/xygtJPIZQFmZelv8Ro7asuLqlgCHRWNbtMd6bPxPm3JP+IWSYlxazwKRW1 xywVBXehcXzNlYTeVpvg0yq2af3qVm4/1PUrwKDSM6wuYw728LZLxsmsb4dh /SrMw9QDugV+VCINxmBvOWXghFPbFo+hC8IBIWA/PgUSGGVrdta2u34VDbMA rEMHHHoxNlw8EJjbLbpmSQAG0rh35ASDbLnjYvCKLtsyJ9hsg4sAFWiEDrTt uxnQDSwdDAAjar0VtUUr4BO/lhprSQBmOM6a9y/VFfhUqT4CB5JLJjaOUaw0 AEvmNIpZoPNDUJUgnADY1G5Km9avVrftFXzy5LIhDGDnmLzOc3kLCuBu1hDg qedEOqbd+kzVOhhD1qZAlwWxWC2FsYrlpva6WkJynVR2zTssW8qdsdXtPhTp umHdw3rjYtDz5JNPMk1IcjtLXp4r/7LtFdLlBBnYSgbrcy6Gqk0WrLHblU25 c61VD7ESaZX84xNRzpyePXuWccZ4NXk7sx/GyDfxKR344rDcixJ38dct8myZ e6pAcb6yJQtq0MLROLYPU/E1XvM/+eQTo1eibpQsXK8DrmG6dYUeXemBPhCR 3vSpZ362Qo38L4XhLFr+IhMSBnVaRRmZtareW1J6ZGy5YdfuWAntUcLBlZJC b3ro3F9rdOaLug6NWgClghlUxfyal9tvvx1f/FTbZK0imjKog3OH0tvPKm2X 6ev4IUmSySjm0vnK4FPpF9avzY4NxDbF9EkNgAfScwv9aXRzV/Y5VKENvIcN 6MmWHHX8lgXwqD7xxBP/vNaqF6NzH44ePVrmK3y1ckiwAGG5AiATdxFImetK pDBqI1awxu2uXF90Zmz1tsYLWXlwOovaQlMVQtGTLrUWagYJqhKl6+d9CCql CrWShqkhIr3pNvXWWl+Fhzu0u0CjItVlJJARKMddtcJ7Npcvbj7kr1vo1FOA X8xWgHVx5OSWbYJPq9im9atbuf3w1q8GAGPT+I6yJLF1XMDyixVz7aavX7WT VSXr4n84uIr/vvjii4yt1+cddBsA0yFfyedCuYDQtlZ11rdl1q9mo61gtg0B WPAPTgaTaOaSmdVnAdi5c+cAMO5bb9UobyNsu9M3ABgcAkACObgzESDW7Lgj HimwFAB77733Su69rwBsJJI14jJ6vh6AwTb3rDWEQVYmUT+B7UYfYGmlAZjO PS9IAv5pwvJpNyYANrV9atP61eq2vYJPDELndN566y0uL9/HdzC22+1zNtfT 5cuX33zzTTaQWei4dJtB0MJmQZ6+LxKM4WXiWlNiLflTto4vfuGFF7izDpj/ /Oc/53ra7zBWFdzm3CvWuPgqq/IIaOCkyKoTgpyjDvmRRx99tO0Pf6s2kknk iVyfjc1lGMK/AAPX43q23S3YxBe7p8Pnnnuu8LM2wqpg0o6Dv+2D+Pz444+7 pVSHiHHv888/38aZG3FkTg204SpWeaLcCzKRLQUgMcqArypxmEf84qJfq0BX IJxu8z4bngF89913MVXtRZ91SOydxyyqWYd6w0KlHiscWZIxTHFJ/DLCrl27 RgH8NS/GDbi+9tpriNxw9bLKzh9//DHUBF3oXM8mgkzMXSkI3OsDr1qUV+tp pskFC1KAUkUSNssk73ofWqWsN4LqvGr1U0bBmrKKlTKrqi7kibayDUAUpZ0E n1BY5Z25cUuLMQ5pdpq1iESaQyYE5SciMmXo8XnUE/RTy4MbckSf+W7KQ1DJ gcK0/qY3nhdtRA0qmJHi8UY9x3hMtmbTXQAJXPHOO++Y5eQAYMD25IzNqi+F xIpppA8eChQ+++yzlbNMydFQ8GFleggZMdjXfziNGCtcNctLqNXrkjnyEOm2 qjokr09CQyFKKi5QhjqzVkoHv45y0us1qmeENAi2lBRJIABWSgoqt+G9C1r2 pAV2feKLGqBtSTi3oE3waRXbtH51K7cf3vrVAGBAIGvJVPIm2zoptmGfN2v9 qhytXAbvxlCXb4fj4M15sYrBLbm2M9vmABhXBfjt4Ozh+rZ4/aqFuEpah9VH 1qYOc5W1aQCwsjZ1sGuxm9sMgLUCprfyGm2r7mRtPQADd4fMQ9T8qaENBGe2 cFp62FKT4WtfARis1f7yknq+HoDpvPU9w/3rWguJgVXQC4RZXNyqAzC3o6cw xS1jpSYANrUDaNP61eq2vYVPrA2r4i24A1aMKoO8OC50Qcve6rC38hLysLGM eZHMBU3N3uJffoQJ9erdQT+uquw9pZrnkXmNMjeisEIqLC3LXCSq24d1Hbmn 1sMnvVVt5IMPPuA42HyGugJn+hyn1yu5SwjI0G3Gts6zn9yHW3gi5o4J9bfa uzwIlvWgt9tvv73URh1mv/POOyuDyPtUird8427ku7n4joS7HYCBUlqRm1vF yjUbqD0RUuXWkydZlQeSNyTqzpiPBAgl9OYBSfJ//I//MRu8lBOZg0++xDhn TQil1tcJlkeepdodd9yhf/y60QRxc+4aJV1mNygvX748N+/t+5gjV/J0euA3 dWh+ibfYodamyv1FyCOFQgcAARuCGho1e560PFqEY37Jn2smFuzUFcnrrVSZ ujLvZesqwxIeDWGyyNBkFXrdScw2PSvfzB3P5ZBvG6vyMYZ2AUbK/08foJ3K c5txtI0DC7B0if3xTs7mCEdz+bjazKJ7wFtVO5GnTxRSVF92hneMS0oNWqa1 GtYwW5Ulikr/UWKuqWJJCXQL5iGgnAbFvVfGCCIih0COx5Nq0eeOr8ISHi5K 0j5mgnIvUERQJAxm+HKWl2bB5FZQoJoLbc9RGESSA76q2N4mXcJ3ZTnqDbp+ S7ENSgSPpHwdz8QXmimAW3aWXDeNwgWCDY1rKk39gLod9DbXJvi0im1av7qV 2w9s/WoBANtNtzdl/SpcUSRPNX04lIK3vZLDOYuDnBe3BQBsl2SvX78aaaCq 48P9AdU4Ct0hAGAoPJ4zQhJUyR8t7+CWB2DbbYsB2IgfK+uI78vFCiiGgYPK Y71xpNbfJwC2JS8LAFjzBcIVaNRxTtjj9OnTUBaoUNXvkcnKjWZ5FQHYMm0C YFM7sDatX61u21v49H/XyiUzzq+++urTTz/NM/72t7/1zs5s7iyz1sAP7DnP yLQyzqUd4EHOnTuH7KpyXF8LxuZ5X3/99VEJkUUt93sdtvfkDZ2x4nc63uV6 pLLMnfVmHvXMoPmekeRKGHC3rIdPIznSqGDYpgl/V92Was62udabMidYBVvX 6PCDDz44f/58MUgAjJ+ITleVz8sdkyEbe/vtt3MxbWmRQFtLP11r9WyU3601 1thfYnfliGhiTjnuAqRzOgSFvKK7MUsN+M2xBlUi+r/97W/lXSQHBI+Nrdli NDw771nnOZRZ+FT0XdtnZELO1UMpzTthFlrmy/w4lzdOw3Vs36S3W+R2rFVw mUdD1ZgCYMZlwABmdaXDdnzKRjWbkiLnhU4kYfny5ctjt8igBAWTuGuUvS5F GEYiCVNUrri7VkGLUAJswAB8hUUrHEOweOHLgC7fE3WVplHbQcs5+FTqfsMB NqNg8ah4iDbqAXj4tY2nsivM7j1VT/DixYsupiTuNe8jfai/ZpPWVVGIwJGa ThIs/XeX79vjKyHbiJEz1wXbhwkx5Xpkd1jSoIipomXB5yRDe6k3amk4hTdE 67HFByLPPOKRbMsb1t5cVTjbx4w1ujoHn1oR9cEsEAgIh4WC39pKM3oVkE1x ZYPwPgr3BLaruuhG01RaNoL6PzfKklYOdTYlBVKpCkzixg7S7sxUpnuleyVV BHcOYrvb9xu2CT6tYpvWr27l9kNav9oSgO24Hfz6lSFgTi6M5TdiZ+r5Gs6I q+J/qyW3g55zMYsB2G7aegDGYXFb7W92kq6sTT708l5K0pIRldFoycqJywCw CvfsbF1uMQCD1vL+lfzrgEaBTBy6mTJ9BUeBAQDqyIBByLDTngOwxW0xAHts rfnc8YTSWJVb1feAt7u4dUwhAO4yWfAS0BJfKwTAFrcJgE3tgNu0frW6bT/g E6vOcjKM3nZZ6XK57zgKq1Z+rcuXL2fQmIh7772XO6tSXvVSeQfmlNFrqaf0 gAsGdctXX33FFzCMHO6InvK38OCLFy+yQoUAsdXc5WL41CFKru3Pf/7ziKTl LI4dO4bgem6Pg0NxwUcffcTvl4X7nnvucTHviRed+MuAwwB+KtP7fffd5wJu CKnV+/N93ZZzu/TvPJ0L+Cb/lp7LoP7tKB+vxICbIKMPFwkJ5NY3ZKpz+gSF HnPKi/FT9UwaAFjx4UTNkxKpgXhJI2K/yimY4h9fe+21qDI77j1+/Hhhbzr0 PcpLdXXnnXdSSDTDDE8++SSZ6/D999/Xg+n2q+8hJU6ngjJ8GcIgED2kFThy F9yyeEulDdD33ntvlBDqOCGZ07Ryy5t3CoCMI0eOmMeQHkqoGTBTfYGSRURz Se958BJgzmYSM++mxu0I4/15TGiBUy58nep+v9bKrhBJVMK9ZZunjabMvVsW m3YNOkmDVugf1tJ5T4eH3QxipLQPekYkUjvXaSqxM6pkmiMXkDZ2SB5KQRIl dIG5Q6EJNbOhPn/17EoXYLmUC4A0fsP/LcOWmeHBBx8kKNKgIZ5oFJa9c0N4 A8yQtot1QramxsQZF9zyEBELmls8RHCFuTsuumGWhtFKMItrxODRvBOOLysA gSoIFpEEVaasjs0S/m6MWJGNWGBe9GxQwmG+9uqldYJPq9im9atbuf1g1q/a qNoWAFu+Hdj6VTtc3OUcAOO8wB4E8Ds7Xo2pFQTFMy4JwLbbWr8CWrYEYIWF v/POOxWt3tbqWTuhhaBvBsDow465mM1tvjwAc00Ld3AdEL4g19bs+tXeArD1 XBTNtRiA0epKgbfpCbyVuRdUCEvDVERhpiJgdqY6ublaAGx967RLp0EnADa1 A2vT+tXqtv2AT2UM+Pzzz0+fPs326pxd5U1mTzNtt7U0xCR2zppx03mpk0Zl Eza2AJgXXngBJGgTYXGfVcrgZxkxomDz9cz5IpiJ6/g8m9wWiQ9M9COPPLLA f9VnkTC8IWoL2Q0LcTEMfosbbYGxzAxp9p+hNkS7LeRWbJJ/eTf2vDh8dPLp nfAyX5031HPZpbjd1vd4Rq6K07/99ttbN3vggQf4TT4IayNqiOPmaHS4OAw+ pszpENSHH35YjBk644WLJyXc5YX9ixfEGLHoJn/9izWkjrqKX681fX711Vc4 wmMZz86cOcOzFEfUef9ybhMUXkiMmygngD4LrnaxuwpwKhhsMSAMV7Q4WfC5 DvFSGZfii/Q51g8Lhy6PZfFyeHeLOYUAzS8Nb6evxsWjBNbyK41CahpVdqzK negcjDGcieiAf1tvJGkq9dwK6hdffFF2hZGBcwFfbfORYfWMRvIHfRq9BSio zLid2SyRqWsAD2R4jtxbtRpvH7jACxEVA2nWuOZ2WsmcEMzvibVmgsbsJKVK apogrFEVaMe8E1TFa3TYDuZipsxR6VLL/GBGaG8RYoUd0ijzfunSpQSFVHM6 SvZs1grjp8k4JagC3XuCfKhEEbLNUZH51SjczWtLQXc0oSeCDni/o/y7D06Y HWKCTyvXpvWrW7n9MNavAB5Gkqdj+ZcHYMu3g1m/4hG4TvDypZdeqnLxAGBs Nd/XaawdR0m1mgFsjPNoSwKwbbULFy6gvMQ+iwFYJeSWz9o0ywU/C9uYkQUA bDfZhAra5y63C8CqkFj1wAVRZC6AOUljzwHYrJRKnQFbdnJhAQBrY5piNGjL SuU1xRFp6Kr9tVkCfPZNFcl1vkIAbL24YCE62aGGCYBN7cDatH61um3/4BNb wa0wm0w0u/f+++8bawfJG9c39rbkk6wc46//1tIZ1SNHjnB2sBPrsV1b5/rO cXPBTz/9tA6rDFggShbbv75nn0e9wi3BTIs/pME480rs56lTp3LKnWQ3SgFa I6ipeoJHjx7ljJ566qnXXnttw3z11Wpk2PXMK7GZcFdesgWxahoWTqO3xqqk CI50jpiSalYnsfNTW3LUqUnulZzLI1StOoOOVSbcIaPCKAjgqTugxzsjldfY MNVDMVGd6yyHhs4HL4Sjc73ddtttuDD71UnBFF5czI+XXnK7alw6TXLm1/Ci /7bGCApfKdiYozawctznzp0zp5U53hCC1jPCilge0tCz/o0ytuF8c/vtt1cf 0ySWNm3JxBEjwUU7fdwoOMFNk3mVHxGMfnwFmWjFqHdZlt2qP58/f54d8JwG 1OdG6VQpp9+udI8GjgJLnZSsvGMq1wZiKWFbE+s45zIPZtqIHRoOb7ecG3wt E4LhfEOkZs0DW/7V7UIms0PCcD4VKicqBS6DHC46L6Dn7SYI3XAsRuObb74p S14LmF9++eXy5Z6XaRN8WsU2rV/dyu0HsH7VCbXqlzE+fEGJhvaQl31dvxrF 4IxiCjBS3RxejK3mvPi7zkDtcqAcKDDASv/iF7+A8RjqbcXzLGBhVJQDjE1B e6/cGUdWSE9LTIsB2JajjKp54B+0wItBF4ZIUETHqe1gLW7EKcVCpy1Gzqv2 duNiuwBsbpQx0Pfff+/RO3nyJACmc72VDtco/t0NABsFEMdpRNgvTL4AgHnj gN4vXLiAHYNuWBZ5ZJ0aA4X/3djbCgAZvF8JADY4SqPKUgKMeS5SqgmATe0A 2rR+tbpt/+BT+dL//Oc/V0qPM33qqae++OKLjvnvpufWbdra8Db95JNPPvzw wwXHtk7S+tUOTiyWHInDYjMZNxiMWM6ePcu5EBFf0Bt6wTNwyNWrV0sXsLjb EWtdCJPOeWeS59o4LFZahyUXKvt3KYYMhBc+moOAFsrENed68sX1XBqiIr4K keIOYAyOmORHok6NmyvHuP5///vfv/jii9wE7IFfXOtn8SbjbOg4qq5duwbg cXClzWwdwyi4MCmVjCn1U8VHRtKA9f2PjKl8DV4KM6NCZrlUlhULrv+SaPnc MbdO52GhSsFLzjiRuhgXxEXOpGGiB7Qoc/6xY8fgQHP08ssvQ+YfffSRJ8Ut 5rGdvtLjb9g/dkYmMbS98847rV6GMeq/pbkC5KqWWDVnPW95VCH6dV6dR+qE C520VoYLkClxmYvScFFdLwJkRbfNnRtNeokUKjEwcvLPzngvKVg2j7xz0Vyx cOda88EoWCAoqBIX+NU5hWzvchl2hsQ82h5hwAm1hN/iIUvFkviGMmSvQnqt J+8AOyW3ousbgi5RAGDP6GRSEYFd2queUELzGJa7r8JVPWW76XmuTfBpFdu0 fnUrt1Vfv2LSuXXgoQK1J06c4IDY5J1lV96s7d/6VYV4eEA+i98skxKw1HYY VJAXaINvNwMx9ZBMpV54GTCg8I8dbLfNtpyLKdAbsw+pYmEc/4eXXnjhBe6s Qj+l2lgAwBY3ctAJrEg4Vc3TP9w10n3n4rfVZwCGkGEeWmQWnnvuudJuPPjg g63JtPXJ78MDnU1bHoDNts4MEgUWyH+c4wPGMKJ/o7fPrvNdArDZQttGqY4P jhpoPQArGooqIrJagbOzMwCYgXh2gtKDUcpGq+eOT2IEgm3ZZyUAWMnqiV2f RfEBlrSLGCcANrWDadP61eq2/YZP1ViBaozCdHio2eo9icK6voY9+KPMOJfd 1obX6vPnz/PmuelRQ3C7zV3sGy9WQUAiKjKHP+UmeI0zZ87wQcVgM1wjImvJ /scekM450zaYWlziUv3t3H3OrvB1PmhUwl0c/VWCyoEK9MCfggGtj+ncB7yY FLxAaydPnmxx5urVqzgq/KbiI3OhWdf/33I2fCL2Of0RX9SaSVUFK0PJa5T1 dAR1z/a2uLVWZgiOnlMYMUUtwcUFFgxNhi4wWTwv+s1IdR43pHxskyHJLKCN uyelsjxV8zEufAAMzL7+eVXzFcBYhvK5sCi6CnoZpQNrzUJzPYraYAExbTml USMibo74WfrLRQ+0vPjiiwDMiIhrriuGaGo8Gia6vKYedjcuQB1zW6JtJoKp oMsrr7xiIjwCxcI1SiuKv/nNb0BoxOh/+RybPaSNRbzlHjHQ448/XsL/KlT6 oHMiividHU9IkoE0qm5GjGJ+W/3uoWOmltymXKZ1QpnkO5BCRGWi23NjO8Gn VWzT+tWt3FZ6/SqYwdR4A8XFSFOzV5ZztP1Yv/rHjfJA/DLYw/UzniVb8JLO k5aMdE+QamFLWMg74wIvUM3OTPR6APbhhx+y+V78R5xPBeAMx42W4GjH56QS VAuVkN4zzzzzwAMPFNZlLl5//XWoYMlormUAGBdZvD0u2gOFl4D8VuG++uqr JU87rgdg8DAYWSEhszA2EOltm24ff/wxYW4ITXcDwDoEUczYc889tyQAmwVF FH4OgD322GMPPfTQAGBVezRum9Hk6eIlU6Zfv9kArMOJnbMw4ksvvfSXv/xl l6cFZ9sEwKa2oE3rV6vb9hs+MVNMB7/A07Ebv/71r5999tnOrO2yZ5ZWPxzB T9ZamcBZQn9/+tOfMlNPPfXU2bNnd5N3a0SVt0dw6tSpgljYcJ2XKEkrkpnz vXLlCnlWr3ZBt/9Yy+WF+LfeeqvKsPr0lz86c+YMFu5aa/fdd19VVwitk+w+ QFZu4b8ACZa/9AUbchfxhZORErGTCWeKbD6OuPR/7Ngx33Tur5jzcopWFYXB rw7je++9Zyw+JcRFnhDRhQsX/JqzdkukQlAdhfONf8vA4IIyRlKAQr45xwKz t9wEbLiKMusBwYgsj71vDDrCv8tdULGSytDAPG2ooRZGLdbauKXNL/mkrqKf hOOaKMZmGeJPnz6NbP9WsJjozDKStlx/K+zKM0V6L7/8cvtx1YscOakMin5/ S1xW7oXySlWbwFP52Wef6QeioDC0a3aPr5oyg/6SwRK+J7oY9cLSKnKEHSzo E/GdWdgsdjq1qcRw0VZtiept1Isss1axdihPkqU+A5s97G4hdgJfDHJGMg1s vvPOOwEzKhRHpW4gt2p0+sCSFB+44105akCYZGhSSCZtIUYEe7IIio6Vz2H3 NqoDI3CgzjvN0dvEDiLtl2kTfFrFNq1f3cptddevAmD8UdVp2Xxmc0+g3fq2 t+tXATBgJhhQAk+NC4N2WP7Cb9r/2hP6YSeWHwZAP89+8eJF3+zgBN84+zYH wGazNrU71o4e8FN6qx1HeQXAwAAurKp5pRyvwBwutgzXH5QX6ka2fFPFxDcE YDBSGRsgFnALpyNPV7Usl9z63BCAlSXAh8cff7xTFQAYyG0UbG6YsyvK//73 v+8AgMEt5AbYeA35/PPP8R4jI8p9Q0b85AKICNqhNjAbnZkDYD10ATBkeNMB a7/88sui6Ef1wC2ldEgAGO7Onz9PXTujsWRA3ZZtAmBT27JN61er2w4APrFy LP/p06cZRu+2HnDG3De72dhyL1+D+Lfeeqs1pWeffZY3Z+q9sPN9BmKsWMWW L4az2NmgAz4VisPIe9XtPHhH5js2Vfog5rd42g5hzYk0i8oFkLZOyIT8jx8/ zjvwFC0XYKdj7Fgoa2J1Q9pTq+oxGphiEvj66695kwLAGnEu8NiIkEbJyY3F j//xj3/kfPV87733tquiW3/HibwyZeGUdy40i2v78MMP0WziPvnkEz0gvoWR sjZ10N6VHcFrBQYwABKqwKi5rHVFKJG3jfJQ4lzerZHPoeFIXidJGxeG5q9B CLqE2nJJ8YOIIckRPM+/G6usldeuXdMVCGFccis3O8GSM5bjF3Ai2073k3y7 fvwsLgyBQRfjjsqNVco5ac9u9pliDF6+fJlMRv0XtOmfcIoWMwT4hItz584R VEn1ydOVeKFRZlkPnQn11NAuOob41gk79TZC0UaeLlJKAqitdoyLuW9DG3Su KM/6LdFOO5JYGIMMEVYMvJ51izbq2sacL6mWCTUpVAX9vglLe/SAosLV5nSy +W0PjpfviCgTgeBOVpYIjgQQ70sCBKUq37yD53c8emXoJUy2Qp/VG60w95tv vonf8np1CsaVuwRRJb4r3T3tMgWkuocbi3Ntgk+r2Kb1q1u5rej61RwA4xFY 73fffXevFnzm2h6uXy0AYMy+1/OvvvpqD08VrQdgBOUlnZPdbidFi5WkC6QB A8AzPn0kapgFYEFEjnU3ZC8AYFeuXOmAw4IeZnFFjr7dQ2CS/+XNq7u9AIBh c0EO9vVtZwCMJEcWsvVwiMxbxNsBAHMZtAl2LtCouSCxio8bDiIiqI4iVpt7 MwAGDMNay68bH0IA1qPnp72qwlmbANjUtmzT+tXqtoOBT4xS9UqKk+F2uYAd J0lgfPgmhsg7Mieiz2effbatDX12wh2UYvYNxGq1CcLv+KkywdsdcRY+6VM/ 33//PSPMsxuIsWJ+WeP2X6rm1gn30gK0+VVCsD/96U/elPkCrrPihmfOnPGN N/0SCAAMYyz+scTaOCq/AVd18uTJwnrbdCuYpzTv/B2S2kjKEehTb08//fRY zSC3FijKLfn666/7PufOEXPuenM9qggNXwVQFWHlb5VBHlhrxYxBgBDCpUuX Pv30U2IhZ9foUw/AQ6tGfNOFCxcivh2rsEo7RxcvXqzYcTnDKxQC8hFCYAyP 5rqMmq40yl//+ldzDd5UrIS3Nd2G8I1riKKKeyRckcf2rfytLItGc1DOZZf7 i7uhk7GJNe61fBFkVWEUcwTzIB6yasXD/Jo1pKaTnRDUjyuhDp0XnmdculF2 BXOhK7rqRuIiXvAJQgBoDWdqoB33pksoMS+5Xf/+7EYruswcIdgctUJYrQGu 0+eRksJApOqvQWlR+a/8Sp3+91pLJ41O2jS5mLQiwQw6O5ZpIl6ypWN0xk9G TIAoNxcmBfuldKsqZbyXFr5wu068pn7lNCs5XlODtsILX331Vf2QT+Up4XCX VTpzZ/DJ1MAS+jTX+mm3HYUtzLJ7xaSNbcGS67q4+uM7GLEI+XZ+KzNRSWu8 7xN2uj7Bp9Vs0/rVrdxWcf1qPQDjhkoNvbf5ZEbbk/WrZQAYa7+H9ciubwTA +OVl0hDVcpe8J/DM7RZNPeATh8j/8iyAVlmb4KIAGFy04/WrZQBYESybTXcr hIAfaYNJHSsYlCOPzugZO/yjC3S7IQBb/vzmbgBY61d0rHWqMtMCBhX1HgFI pcTHAhDoS2hwSwDmoViQn3+8FABgCKvcFZLMbNi+fUCzjJ0FAKwQ/W1p1CEE YL2z7JUBmQDY1JZs0/rV6rYDg096Zgm99XvAK/DBaO8M8DCGX3755R//+Ef9 PPjggzw1xzRMXx6BSWeHmcq77rrr+PHjfAoLDHG5sYCfbcVibQafynP44Ycf cmTe2Su9VxbojlBxNwatRq2//CbXw5NW/g/x3BMby1OMnGAY4SMuXbqkN1YX d7wJWAIzeJt2Mb6I0ZS1HdamD6fvS75sNvs38ozudjK/4447+NlqQA++fHYB dFHYM8qZeoIK0bXcVJk8P5UP6uiNVsxVZyfzQXzi//yf/9NnnZAGx2G4It4r KEMaeu6MXpuG999/P48ZLirdRNtPBAXVjGxXfGvLU3MpI2gv7rhd1yD1448/ 9g29IqXOSLZBdvT/beXG50D1WRh8KbN8RlgrXbgYMcxBI7y4ALXtrLmystrl GSh/vmtAAvPb3ly1/1yZihqo3PJBCDdWvlk/3Pe/rTW6REPwYkaMUla3scHn X9NNJlgzCnqwT86oqqYkvkAaIMF0jI1LxCOP+x7cke1/X2u+RwbCOso6YtIM 1K70mMrS75eFnnJSP2KkbwTVCt71NczQum6Z9kuGVkYsqotHw5VvxDQhHiqD 0Mo+qnUwgQDhnJG3ExdUEbNo2wF8apOxJCEdATYvlc5BIUYYivE4ExfCqOLg 7t13362w0bYGvb52bpqitvNL7J50M7thUaE9bBN8WsU2rV/dym0V168WA7D9 aLtcvwqELAnA9qoV4qLnIApXyPvMAbAFNxaNk7vkPZF94sQJoCVfCYRUtK7M otzuWE0yLy6DBnewfpWgqsuzDAAbBM+FWvF0Y3sUvOlU4DhVxwU/9NBD2CEZ 3rYc426hRW2Son8WgC0Q1MjaVEWn3QMw84UkZBeMhM5KnBf7lM48/PDDxAth 6nMZADaonROUyR0ATD/gkJ6ZApPbIYK2y40FvC0DwBa0ctg2O4cWgO1hmwDY 1JZs0/rV6raDhE/BHgO1/F6E87YSS1Z5EJ3uBQk6jMYAct+z2KMQ8V6lq2nC gXINbLW7oKCLFy+yz8tHVmwGn0qkgCRjccHQwpUrV1577TWujUfruH17fL9d axxE8WD8JofF23JJOp+lRJ8EokPvzvrJ18MSeuZzSxfJiXz++efFM7vs7Nmz 7TcZomP4Y5OoQ4LGJahS+sxKu8IcPIve/Kof3I1cST7M5sgqwxUYo6u2Y9rB dHG5s4qsQ4CL0fDII4+AAWZH/y3fRTxoBCR88MEHYZsRS1YmMd1Wt0UbVQX5 90Kh5lJGVFCGwMvcVfBbxa//ea113tM3hINsAinvJZqBDawRr4HIJ8rvvvvu BDV7iCDgAWxwfy1XmruyBGC8cDtDlICrbSx46cyZM5CAWaOE0NEolANsjCI4 rgSf0NBGdrejDSU+N4O6StoDkrWy5HMXmyNDmxd6zmnOpqQY9HtwUA7quKzZ QbP+2+mb2xKl5yg3QRRMb0UbRrZHTw9NLvzpAt/PzkgAqeg+QLpKhYXttYdb aaFUFEdFkbmSQFBoiLJDjLeJHcOn0FpFmcvNm1JROQpTqc0KCswaPf8iwIgu azrYClxvC2IZGgI0HXg0TeQJElfEav9e8a5P8Gk127R+dSu31Vq/WhKA7Xnb 8fpV5xz5KT6dJ+WAyrO0AIDtVSMTXgYwA8DQjPiieRefkCpyCUDid+Cxgn+K gCqZgy9BPj1zl9WVm+tzZ+tXHU7sQOipU6c6nkZQPO8CANbWHnc/zta98MIL 6cYvbjRkkzw5oAflHD3uqrgHFOXuuUsO111k5cpr164t1qggjdvpHt+qZy6+ PGa7BGDlfUraBVm99NJL/OmFCxcQibByUFSXvAj/LQFYLQCJ4CIAdVtZ8ABY wBUXxYmBakYxHdWG1uGSAGxDjSJhqn758mWDJqhDC8D2pE0AbGrLt2n9anXb QcInpoNBY96r+vqTn/yEKWaclz+4Xe5KpunBtcZ9cCJlp9zw+s4tsiQG4iB+ 9KMfdaabaytP4JIpNJeBT21VcPEdxWrtqMiZO9faj9ZaB/ALaCkd5YYjGsJ0 8GI//vGPH3jgAaiAA53bkCo3PkZ4H7aay+Zz7777bvbTuMVl+cxBM+M8Fw9l xJEpa9S2y9FwrJE9EmH5W1ejxh/4RAiAn+Han2p9rMQCLi7JlQ+uL9t5lVbQ 2dJZdVtayDIvr7/+OrKrz4tOg7b3VL08aolsmGEUhRlbk6UmaNONe0VGBV9K pOCvrorv6sScfjhNxI+Up0Wv1Vx/143m4tkKgENQEVApk84tQh3oLIEYmkd8 FE0raaeHqwTmo4QfYPD111/jqFhx4AfXBezFeEWK9YzsMo8l7YLuqo1YRJYP Y1zfI5uKlqx1bqevpArGpZn6LGavgshDLQkqMENK5a0q7ugfN+o26qpjhhCI 20svD49tqL3tipp3s+8hJWo0k1XjjqqRyA6No22zuO4dwKcmi7RJo7eAgu3x CLZRuXJ3bIZkSlcCtFNLBNMZwK9aA1sOff3G4jkkD7N50j0dnvSvvvpqb4+l bNgm+LSKbVq/upXbaq1fbReA7VXbwfpV7q99zEowZ431wIVxqXzTXqWJnms5 X0CXa0MtZw1/Gm4z5zUwWFmbvNcjGEZ99NFH+axSeibtItU54hK6btjbdtev Gr3qNjrnat0OEoCdDz/8MDxALVE1e/EcAKPDCIYH4AcOaNDcLhsko09ulwTI ZK58c8ViwKF2tbDpyg03tQeqKZ7fXeSgT2wCBlx8+LMUu3sIwFw5MtVX/ts7 xUcffdSxhRJ+EhRg6bIuuH59g6xWKNEJgukegYxM9UV2oR/ZxXTp/PPPP58t 37wDAJagDF02LVrhei8IBipDV8l1DyEA22WbANjUttWm9avVbQcMn1oBMBCz VmwSl8fUbAmBikri2U+dOsWts/zMaUVDFm/TjGJqXLl3Ura0bRqmWw+++fDD D7fcFFsSPg0k8N1333nn5fr5CP60KrrFTvcW/9BDDz355JN8ClfFcazf/jMR ZII77uyZZ55h+blI6MJ8zTq1PJQe+FYXkw+ZYMotZGvcMkkCEihvZ4pV53ld A4XqrfP+xRRhjcGvVG6LGz6gluvhudzy8ccf/3Wt8Yl4JFUjfvbZZ7jAb4V9 +eVQRIcLRjyY77H8wgsvsO0ggesNWoGV559/3gVjxaxx+0xQfuVhEVkBF7eU 5T6Xah6rGde6WStRGHnjjTdAhW75t3/7NyAWzR3DRDYkQFDcqM4RST6N6F5I Bs0FI6FWPyWXgD1KQ+EbjDRoJZ59cPuohEjHyNaVGITrMEhzUEuG7fH9txuN YN1FPn4l//b4PvnkEyI1lqlJ1GFajOu56ttz45b7FNxq286NqJ2NhDc0/XEN 7lqfpJCpBMjE1Xo6xpZo25Szz2Oq6NEzHcXIpQnrIXQ4XFcgIuxH2cjkzhtt LOEGwIiFKejQ6GaP/7bgUw8g3nFUwYhiqFI8sm042GmxoSuisro/BYvSJRhv SwikW5KhrgXLka0Z8aQsWQxol22CT6vYpvWrW7mtyvrVzgDYXrXtrl8hlbnm CstfVDzzE088USrREiv1Br0fxJe/AsoCEsAeGIPr2TDbee/afH3hQNwc2Zay idMh4bHyU9am8kLkTTajfPn1q9ylbsn2pZdeAg7L8kRQFY4ZKR0qVWPo2arE A4CVZRQ2KCiIkKFW1/BBA4Al8NnqgbFAheAEN0ImUCvUVMLY9dyFq0OMhiCo knIXAF9AXUdBcV020T0BYNC1oc1pK7eo1c+olmiWSc9wIydYIehzRxHdwjV3 RqCYtNJ9GNSrUIIa1QPbZZ6Nnyf5HQAwQ5fknz6kVFVdBKiIJcinE6wtSMh2 UwDYjtsEwKa23TatX61uO3j41Mo84zkyPDPLbFq5uzdrZbUCSPI+LWuw6suX q9B/tXQ7Y+4NuqqFYAP7VnLpzWz48utX+S8ury0e1Hbcm5BLuVDYTDnA29wB tDiRUYZvdgoYQGYQGEAtLPTaa68BFWGJzQjQCf8LfmDz6NGjHPSDDz7YuDpp a4nbrWYcL+ZirBHCyLlUPcHO8lea8L777uP12vjL083BpxEaDWjpp+UyPRi9 DaY241BS5nNgyYg6rJQhwfp+UFhrjcUtsE3bjrANBei8JIhFtq2NJF79j/xj aCA307pYH4i9TJhurIexDVdBH1SNkCTO1+i5cqobawVfFed22223oZbL5rz4 UKLoAB1R+3Iu4qutST1UfxAZ+gcnigaflW37aFAN4btsruSQv4MG+BNIIB9S JSuqPuLqu3LclZQQAHy+9dZb1HXxygyqgDEaVXoHyGS2fuhsXFxIj/ChslK2 0p/G9eH4WvMZtcRCmKNo9WYqvSR86sFpRxhqBVlRa1pNaDCGWMp+sHwYlUlv 155KnD9/3hOt8wW39zKCcertLmwig81ZbNn2sE3waRXbtH51K7dVWb/aPQDb TdsxACvOn59lkHuZ9dN+5NuZpYGT4v1BFADm0UcfNag3+jkiW2fwIs8rQQvg EKQRSOAusUnIo1LzKG2zTFtm/SoaSozJU4xtRx6zxJ7QSOtII6VVu8CjKvGT Tz45qA2AIbjs37xem2iLHVAS4GTLH25oZIBks5zOhjCBcx4Wfhx5nHsls40O 0FIJJNGQzz//PJEilVQDYBQVctgTAAYGgKOlb/35z39OyIb4/vvvDTqivCKA 5ANgsChqBwAzbqualZSC0ww6G2q14WQR5m4AWAdPjEvHkAQfks+ST+7NAmA7 aBMAm9oO2rR+tbrtpsAnozDy/AvDyBdUpIx9a3dm7uLWZJjNDqy1oME48Cbb SrlZRBMg5EbulRkvSKm0S7wMCXA6G54oXAY+5eM+++wzl506dapiJfwUl3Hp 0iUmnXg5Dv66+CjwhkvtSDgvgztGr8jhERxeugkOrlCWHJ9rfLPhTJX7HQEd B4MxjKVDvu/q1atwVP3M5q2qPJ/ry3n129/+tpRTgAohI/u1114jdrf4NbfL LxvFDM7KtqBuX0JieDdHox+i40f0Y5ZJo1igsQWGKT6uXFXlVK+wCzoRU+mQ UnJFrb8ogZ9NBGQCA7QO9ula47nakkMAaGHGNwuzNx3nzp0rH7tBXUxEvsRy sdZIxW8lC404ZAWK+IDUito8t9ZKSt9Ryi7orhKR6b8zjKaYe62aJLGABKP+ IOc+twtW6CCmArct9gLkBjWVhc9VbBGOKotCOf/H6CkhHkv2XpnFMjCYGhOB d3/1hoANpRQeQzMu2vuDuKCp2SB5+M3tYK0hcFrui3Sg3O/mqBAv0+Spr8oh RtDZju3ly5c3gxnLwCdklMMtuFjPZqdHr9OgQN22sFObnm6ktCXHwBcuNtz0 vL72CHRekogoQFksULsf6R02axN8WsU2rV/dyu3wr1/tFQDbTdsxAEOqz7wA 1MRPdUhq/4Q8m5CzLEzA0qhA1yIDD9JGKs/IcZc2syyd/u08VwfWSnzEjC/Y 31nftly/yl1yxNzESOMJJJhfThxtZrZUSwM0hoWQWiSbW3KvZ86cGQCM//36 66/1PFJyLabZTAE2NMqcwgDlJ58rOqNP806eocGqz5RmvJJ8WMCmD9w0HEXg GAnfunhESe0HAPMI4BcUL9yI2NGpq+rrlUcLqeiBSQDa9otD+FVLJ6hSiCwQ FAIQVhjVjgEY+lHI1BQIV5HNJZ/cmwXAttsmADa1nbVp/Wp1282CT5Vg9oxX MLej3xsee28NhyMu/tbFfFllRHY2dG+mnUNn5Dut7wPbW4ZzVmguDmoxfArg jfP4LuDmGLF77723KsbFd3WYkSXHS2E5P/nJT6oxd/fdd7fc1LpEEbYlX0IG y1/lvg6tl71zPTzobOa1a9d4jeq+BWDKPV4ITbtXJa26Y63dvtZ8qFALD+Uu uMXsdOBupERAJ7/gXg7RKBhprW8uFmsWPvFcrtFJoVncKIxUUZvyOM1VM+Sj wRJydiV9aI+vMnZVPyzHlItpAgxT9UAowoyUrsrouW/TCk5sGNfXahsYkKs9 duyYsSCKtt6q+6yfMEMH7ir/h4aymZUaotCsV199FacdNCt6bZQL1Hzje4wQ GuFDLwYaW4QEO8o3zypV246dYoPrIHAaRVxlCiWfDm6Y7qo6Am/0B1UjO1at nAPtTlbhEYAv8amH3ei01I1uR9hmQfsm2u0djcQm+otUHGkoTC5ACByOY5UF s/XuYFzUFlpfktgQqTkqtI8ykKEvA5BzVbkXwKcULzKw48EpD0YhXibXKHRj 8ebm4qbzVpvx5dk3g5hdn5q1Tf/KLXm/u+uuu9LMAwYwE3xaxTatX93K7fCv X+05ANtB2ysAtq9E/mOt/O6GAGy4S5LksvFSZbSA0Kg0Bwm0jMNN75iMzdav ZpNH0TS+zDVFyHDEXCfQxVO3YMLP8l88b1W8S5pUBgwue1QlNpZbiHdbK2yR wV3OAbDASYFMPkM4roENID3QlE8HqJDaeTTOvTQU/vo8jl5WTa8cBW5x8b/8 y7/AIZ3U23MAVqDX3KGA8m/gy/MCtxT45zJQaltO2ZQV9wVK7QkAW37o0W4u AFuyTQBsajtu0/rV6rabCJ+KSGdzPOysLsvP4SJjdgcEPZyj70GCznOVD3xb 7nJ9KwCst3gE8NEsf5s7LLB/q44xhtgMPnUWkiPzkst98MWQAwenk7fffptv 7WzgiOlqCavTdnwZv2Mg92KwnaPq6D388MN64HC5xXa1eLEWHDgR13DobaYM SEYg+uSd0eB2LpvHNDTpIY93O3/+fCcE77//fjNO4K40yvPPP69bb+gV5mtB z1+/4pT7O336NKCCd0T6F8pqewUNvtTznOufg0/DwZXkgZQIuXWewsmMUl4C EMXQ3PR9aw1QGZCpYO/2/kC+8Wt7TGfOnLl06RL5wDztG/qGFwMgqc2VK1fm YupQgkgUmk1ckD//PhJ7tl5BhoWEFe89tvOQ14lIrROLNSSVnZKKmseQDCfe iTk8cqw6H6ujpZJYD5/CbxAL1EQP9WZSjG52/AsjkSoZmrXS9RN1kWmGrtAP 4nVYyUI3dpYThaThLloR9OJbxw6sC8ysB8H3Q0rNGrUv+s5APpcwQTOVFy5c KKdEKQ5cg4DnnnuuSP6eICpadacepQCPb/SADCO2p98GNKH5l5TG9uL1zeHT P9ZKYOs/jE1/2sIuUZueK8m9eHNzy2YUnZgO8jEEHk0uymeTnbqmtGx+osPl l3PLjhHjjtsEn1axTetXt3I75OtX+wTAttv2CoDtXysdwRtvvLEhAGstiHer dBF/XQgTd1lcdMmIRnqr3VC72foVn8UrgZSmkqdu1xhgQNVrr73Ga4MNhTD5 FXKAKEoqzkHzvyW7gKYq2jhyQxXivq1MYtUm5ig3BGDV3Q5XREYirQUVCoMH zDhc1xTmhHJ0cr6wVgX1sD+iwvjxMs3O5vLaPQALDBM42lwDJ7iGGlBXTM2W by61+/JPDTVAZNnpy4e2ewC2g3ZzAdgybQJgU9tNm9avVrfdXPjEnvMsEFGe iC979dVXGdux2sP6cQEtrdx///1s47bqFS5uLa3wBSWE5BB70+cI+FD2J2eX oZ6DT2NZhm/i0Pl9iOWuu+6q+glrT5iz60ubEVA2exLgd3Cn/3ZPMOt1nmPl CHhhlpDjOHfuXFt7vFVHrkbMDEoI6uLFi+7l6VzmxpIGjIzuXMzx48dbFOKR +d+C1bEfuGoBrUTod955J1GM2CEeeYQkEVS5nojCjUUuDUC7fv2qRAft6BkF cCKl22+/3VguQ+qoxnLPPfdUjWU24zdqjRiMqWY3bUEk4NFPFacmkAEVzp8/ /7MbDYUjdWrE8HQcnFnWv9Fz9KYYtSABZEJQKNF/OdJLVub6Iu3JsMxXWkRW NpG4UEiH9YlNpJZDzMW449FKoJRDXw+fUnvz6CfgxNSUE6wDiZThjRttzNTs lmgzVeiaKSh1p3FRVUavjn+CXoAZzSETLh4cdZe5MK20q+oAaRRc+h//8R+t LRvLBRSpI3U0zSi9Lxi6BT3zgiqS96SUxWKxMcGv/mkmgRtCD2RLyKaggQJ7 /q6HT0OjTFmLjcnZ24HnCBd63qsXFjSYlE7RohO/GJwN4DdQWUHQj4skvFc2 alttgk+r2Kb1q1u5Heb1q/0GYMu39etXewLA9rDx4J9//jnXvC0Axucy1Ojc w1ONywMwLpuguLYSJqCw7JqlrEQnAFNVYg6F/D/77LNdZg8rrom7BMD0vCQA qxogdMEFFwTYsYUyWfncPubgRYdff/01nZmL5Z5tnfTcDwDG+2OEWu5yjbec 7aEOM7XnAGzJtuH61UECsMVtAmBT22Wb1q9Wt91c+FQ80r//+79funTp5Zdf 7ngaX8/PdpKLlebOyubEQHFMu9ycWk9AaRN4ivI9njx5svI6gRBunbMgmZI5 g09+ZRvRTGKvv/46e4VmvgN8AhjefffdL774AmipaMuWBJSVizN1C58LiVUR mIss605BRzAbQMLp8w5GrIQiUitJw/WbR/7FTyO0m9zcwu8XTDUSp7/99tsl eho7aP+51vSDABClJF38oOFIoAwJMAM3Wvlp0/HTtVaGJZSU6oovw8s777wz 4NOVK1d0iyPjFi0/bmwH7cdrTZ+linKNrhJjDSAhihMnThA7LkatwM7o8ZjR EyV58FOnTuHdiK3e+AufmFxqw+FCJoZwF6YIkF8u+q6tPbS1RlGyTaMcOXKk UUzE7BYb+eCUo0xQujJQ9ZQL3AKrZrNRoby8DXQGDmlbMPiEgKLEuTz6Fqez GbdGiJcvTYQ5bUuUQhIRjnRYzgoTqqvqUZI8WEgxUtRRhVCf+geH0IwRGoV9 ryeGcI3Z7ziAucOpsbBgXtI9+lByCU0nZpmoPZimmOYUKDiWnrbU/CIwDUch e/TGzm9YlJT8ivgBn4qyYy78CiyZMpQjMuDqUXX92BHeGxtx42AIBgmq4kE+ oK3Dlb43HdQPhVQrCg8GvM21CT6tYpvWr27ldpjXrw4AgC3ZZtev9haA7b4V J0M4/ELl7ZYHYLgoLGcP6Zldv9oSgA34VIXoHDpnWpIoRK6vSrwbFdXVAGAc 97YAWJtorizFBNxVicPCnGAPnQdoy++0WEtn16/2FoARFATSxvfOpIRyPdDt TkwYFBLbcwC2ZJtbv7opAGyzNoDZBMCmtuM2rV+tbjsM8ImvYfx5TE6Ks2Cl 33rrLT7oq6++Yrd51ZIDeGHkHfbJKo5M6VwSp3Ds2DEOtCRLMAAPhTweBCXo 4SgRzNOdOHGiM2IEyHi6EpE7lmHbnaACZ8RlVJu4knz33HMPh3X69OnCtMAk RhLk+Oijj1hvftP0wSE8rJ/QDz65fWyl+YnXczsuzPICyFT4NKfAI7u4QjMd 6i8UimR0W+Vi0vANgEFQHMf7779vguCKINzZs2cBDORxbdUiARXaTcv5Vg4G IzoZVRHd/qc//emLtcYbkjyEUMZvOsCVV06lE3yEP2oOgjr697m6MDwIwObi onp0gutScfK2rkQACpFNnq4ZOaNGpUikFodWEcYSoaPw3LlzJaU019WFwThl ADYIf6RfcAtBtb5XSoSqBFazBoIlGUyhRJ9mkL+DbMup3qaeWetEJ0SHr/Kk FR/ouUDG5cuX0bBh1rihUR0AuXbtWiUmEYOMY2utJdAApClDCQZNOpsA3XkG w06mqaxofnULGqKNnN1ovkwT1LRZnvxl1P4///M/SYCyedyKwcN4jx4Gybai 7Sa0/BX0gVp2ZXXGXen7xVUUd9PAIRbJW0BKqNFMj8mf//xn6kpQpIpImJMo 9omGLdsEn1axTetXt3I7DABsfWtfz+vwgQGwxW12/Wr/ANgOWsfhOS/gEBkd DNwNANt9a/2q6PqdAbCcaRnD9mq6A5a0fQ8BGLFXvHgH9MyuX+0fANtZw5QH zRBYrnTR888/v38AbHEb61e/W2s3C4Ctb0Vekf8EwKa2mzatX61uOwzwiSHy 4MMkXpm5+CxzYSFFZLWmsaCW3J60yr9yHF9++SXH1ApS22cFXTDUTBNPytFX DKXc12ym11jWm73aZWnmbDLf2hs948wvwDxGOXnyZPVfZrNCje0PtPU9Ckuw WSQS6PL+++9fu3YNU9999135ChaEVV+fSdJVniJkwIqffvopSvh3wKwaK4VP 8+kAQPV5q7fi1X7kjCqWBj107I477ujEX1gO8nzxxRdhABDr448/pnvVZOF9 Ou6nle8UzcigHi4gELil5UQ+okRPzYV5OXr0aGf6Os3nLziENqO7oLowETNi t4rnd8ttt91GdKVbn81jACldvXrVY9Ie32wihXZ4h6C+/fZbQsYIQb377rvQ NR1uEVK3R44cMSh6jAiQVHQySImkUWkROCmhvStbjSwrFGZHhcSxJUpQbYku BpmV0SnIEJqFQJAHb1de2ayNUtStuZU/xMT5Mk1rcgmQtE3ZhQsXSoLx/fff 69ZM7XK7rXB0MiRh6krsCKgwU5kftFHP0ff+kqrZZxloprvSmf17t2olsAQd 5hQxjzzySMcZ0GP6EEwm+wcjl2kTfFrFNq1f3crtMACw9Y1H45d5QM76wADY Ynr4BfSU6nz/ANi22oi8KvvBXWtt9wBslw09pVHdJQDbqwiWFsFKvfXUU08h xiTuCQArymhnkjRxIwB+vwHY8m1EXlVtvFQeWtlf9w+ALWh6rvziyPR1UwDY XGvTswX2CYBNbTdtWr9a3XZ44BO7zRKyhx5/JDGPvLB3Z6ay8+wHZhvLhY6Y kirwrUgqu3iH7tvyKzlAReX2MF53tnXAirt8++23z5w5U7LuMi+FlFquGesM WiskJRODnYoS2X2pIL4JZvjss88qHlcYc7X5Oso3WuIKZ/oAHhw5cmRUORwr IaMmy7fffrstsBRoJBMOgh8vq7xp0i1iksAoFziqFjZrrbZVRrALxrJS+b7g PbQ9//zzIER1WwAVrnlJ3QuqlSj1gw8+6OBbETujMqDhjNsiW5Cy0KwgXLIq 1K1FEtjpzTffNI+ldNjlU1AFwM8//xwsLJoojWpEH4KRRdY1ZZEKyz355JOU kNghyX16TcDdN998Q+zlS09E1KzE/lGFPDSzDx49iOUgy2DRPc9j1YiKTPuv aw12IkwP2oFRsmGb4NMqtmn96lZuhweAjcYXV3OWwWeBWdoDBmBzjUy8ll68 eBFm6I2+nAMjVen+AbD1rY2MNlzaZOQTQYuBwUYp570FYFvSU9k+LgBayDcF wDj04E1t9wBsW/S0CwlykAMkA1/NAsKgVxiD6HYPwBa3WcIqil0AfAAs6Bgx vYMQ1J4DsPUt0BhhVTw8ceJE+9SJJYU/GAA211pPO3fu3FNPPYUkSKz0vBTs pgCwWjrm0bt27dpDDz3k6fOsTQBsajtr0/rV6rbDA58qJYOGRx991OthUcSl 5t7zAN3FreB5dpvrZJ/5r8cee6yo3VYeKpzXthpkta9migfBu1G+++47buvq 1auAxxNPPFGKgLFEU9w1wk6ePMnbth1ZdoU9KRWUhy3WqOKJiCEcLuyFF17g 5R988EGubawXjXUhrcTmCCsLaKmfSts1MuQvjwqiZDY0CwDDLJZhjBLRUxsq ffz48dapZtes/mmtjVUahJWg/o9//CMY/MUXX7THVxqBsce3LVk1ZUBXs4Y2 ekJbOlPws5/9rJWiFtAiZqyw0Xy++JVXXrl8+TLxJqiK1+wJyMz1l6RUz2j7 4IMPzAsh/PSnPzWDA14G5HzpeXz11VcJp1pO7XHv37sMNqmZOWUK4Njf/va3 I+OZv6T3m9/8phT0xHsApZlnW4cgqpf985//PGxJiy5dugRNHUy6lQVtgk+r 2Kb1q1u5HR4AVmM0uNHXX399lIRm3PI+N4UeNg1g4D0ffvhhGINv4g68TZ89 e5Yd/uyzz/YbgM02s8MXQz7oIRlkjGLK999/fxV+K75WMqI9BGAbElPuVvQY cZTg+clPfjJ2ytrK9CXQ6AK+HhQpkGnHAGyzFiqDEEr78M4775gjIjJ6hW9+ /OMft56W0/zlL39ZiaKCrAAMz8KeALD1LTwG7RgFYaO+IZIAHrKqfAxU/+yz z1J+9BBUk7i3AGx9K4umEckB9nvooYfMGmhB24mo0K8i0Mqvtd8ArJZ2sUiA KFDR1iFLFVpuDa2k9wdvslrlI5lSAYOI9I0MJwA2tR20af1qdduhgk+d0Dl5 8iSswqewmadOneJ2D6AG61zrNd+4rCJ3hgxGCR6oJApxlcC5IiP7miowStoz qlR0QAVeav2qBRm0+RdhMANjXg7JgNNeJTGY21lDDPwzEit5ned2W08ba0Qd 6IOgHnnkEYRxhfAJwghWD4luBzO7fo+voCzTwWucP3/efP3qV7+69957y/85 6JmlrSKPCANmLl68CKu4nQOao20HAerryYNjMX7hwoVnnnmmDaPWQmcJC9Rx fzDwe++950EAD+aI2cNkFJGn/07LGrQZHPQUSgfjPf3003/84x8Jp6yke1gp aTPaevRo12uvveZBG+u0SILoYLxPP/004Rx8lk60BYOJCz1mjZr59+CLNa9v E3xaxTatX93K7bABMC/IHLq3eK+rTFx5oQ+eqgHA/uM//uPKlSsMWsGu3NM9 99zzhz/8wft+ixsHQ8nAPNwlel555ZXf/e53DzzwQNFNHROAOt5+++02VvYw c9QcJbMADOaBK+C9gXnGOcHcZatqEA5EUVU4TnzPKQmA/f3vf4dzSnlR2v9q VSNpbLOiLY/5s5/9jB8f1frA7A7i7Z62DcmDFiDnFq+ee+65X//61wgrEr70 F5WtLPUuTNtiyP6p/SwA++6770iAKKBoCLB9OiSZzccee6zwQhN9MBnRr88A MBMKGZ47d2720MfLL7/MUpHnzbJUhMBGUTPa1VFimv/NN9/cdPR1fQJgq9mm 9avVbYcHPnGszDj3UeG2aspUP873vM+Bha/zXFU9fumll5566ikopYp7/pZM qYotpZi+cOEC/7KHzne28VntZxmFSXz88ceZx+qzHD9+vJjnDuuV/MpftFVs ly+GImAbtp0n2iUlwNL3339PW3g03UJElYwpu0IZ1O+8887wCcLuvvvukd29 VF1VbzGtjzzyyDPPPPPqq69evHixgwmdnd+ShmBJmy/819WrV/XAvVa/hp7A kDR5lp5xlhAlZZscq2q1wvw4aIxANWh7+umnz549++abb77//vtVdob6lgQz Ad2B4jr1SYVMh/6N0onL1j3KCNqKKHG1QBqdBPWLX/wCRyANwAB0YRZSrSrx bh6EFvq8kowdZBrV2diOG5AVCktSQXoI8xO0WfrZtgLdvk/7blirCie5FRLW 4UH03HOj+Yxmjx7o60r6cGCWwWMOuaGNntA0hB1fa1UYv1nJjWfJm+DTyrVp /epWbocNgPE1/BTvw6vyX8zdnqx4bKvNArATJ07wPuTDgfJT/CNPBNLwYvua Raq2HoBVuabifcEMfgooahEmce1HBp7NAFg1+3LfXHZ5AGiU7zl3s1mJ6tkN zR3TsAwAa5oqLR097SEij1IBM1iAH4h0r6r1zbbNABjaCIQioS2cg7YBwKoU 4+mjdcUU7RU969t6AFYmt5QKYSQGHptiGkVEf/vb3wh8D1NsLW5zAKw8CQ+u NTNL3/yE/ptS3e/6DQAGn//ud7/zKuFh/OCDDxB8s6JD59oEwFaxTetXq9sO CXwqMPvSpUvMZse9c4iMOdvO1KNzv2143pnnYp+JAjElb4cH0MNUMuA+e6Ov yG/HwAEJXpLH5PX+z1rbPSVjawYlHPGHH37YSlpnBtugqYpcWSBa0So1RLuB pa3wWv3mm29eu3atFAezYTyLJTmb5KEMBpWpNRGvvfbaqVOnWOk2sIpuKlaH cECCH6+1VquCCq1rFb0W5RV2NNfvvvtuGRjWxxrVEsXI7UDIUKLZqURj4MTU EILRx1m8cXRxtOCTy0YiiGLsa3HRBhOyIQqi+8Mf/gC1ckNffPHFKMochbNE NulR2PnK4vnPnz//3HPP6SpVmT2XpxXY33k9f6urOJsga3z/6KOPloEBXPzy yy///ve/h/eWD8qa3YvECFAEA+MOGgdLSlja6cVWRCua4O9ITBEqhl48lbQd d+X0iIzdv0ekkAkQQHrvvfeSG5lUW6pKiFWybhULYu/UZ6XS9zsqLCKT3mOP PdYqMcVDJJI8DiRjdvYP9y7TJvi0im1av7qV2+EEYBBOrvmAa/ktBmC8ACfF Ie538p8FAAwxuUuzVhLUV155Bfzb21W+3QMw4Jl//PTTT3dPw54AMO77d7/7 nbv2VkpzoVY7A2AwPOXfP6VaBoD1ZtHfN954w8Qd5LrxlgDsl7/8JZUzg19/ /fWBUbWeyFkARut8ALoOLDJtyzYBsFVs0/rV6rbDAJ846C+++AJcgU/YyZMn T7766qsd0jlx4gS44nte0jVwwj5ZKv6F4zPouXPnnn/+eS+nBu2F/eLFixcu XPA9b8ie//rXv7527drVq1fPnj379NNPV+flkUceefbZZ69cuQJj7HidrQxg f/3rXytg98wzz/z+97+PEra66sYlTq82nO+9v/PUBAVL9Co90h0UWlNIkg9P PPEEPFPiIF5gM2fNF6CBj+DFzEjS4LYatAkiAR3yvCUsBQn81FpZGzcwZ5WO QQiIxTeAKLhV+RsfqudYin6f8Yg7c02kb7/9tls+XWt//vOfiYLwYbbTp0+b C/10Vw1rkPY///M/lzcVPoEnXYBZgBPNhiMW0uNQSlsBNvgXynI7mkmJkH1u nQ2QAGnc0ppJ+2LJnxBcj5eRrgGFhMlPeXZw2i7k2BL1t+J0PqBwxDWhTQ+U Cobhf8nEXTqpoqJviltDSbV49NA81mc6ibvSWSCAwmwGdQLAAKc3EYItk1tF /VLaJIAk/JpHkzKqQhOUae2YKrHjzmVuSSyu1BXo/s4777j4u+++2/FRDg+L 57pHz0NUkU36gE1TT71T5t+uNVIi+WBzuo0MT4pXDPZh/7bhei5MUDlPSAYl FMAUeFTRTFBlxIWE94mGLdsEn1axTetXt3I7nACMAWlL7sBoWBKAcY4c8T4t NSwDwMpQBCHw0XwBX4nmPcxwtVcAjLPmnSGoHdPQOhXHumMAhsgBwDBCUKZ4 T6TUeksxaXjcJQDbv/Wr5QFYycHgH7SZ/YPMK7UkAEMt4tF2MFTNtfUAzLuP ty3Q95AsXl2fANhqtmn9anXbTYdPjM/f//53DzvjiRJOma/kWVh+RtXnEydO +J7X7gjVnkdhjWzt3377LTIMxwR1Ir6FoC+//LIMot7og08MaZuVvuGdCwgZ ZabL9FhwzjIEjI02NHDHZuH111/ni6vnW2BMzo40zp49y4nw0ffeey/y0FYe zmvXrrnePPp76tSpt956C9YCNsiz42ner1H+8ssvk+3XX39d1d0RTTQSBZRF Cgri33FHGsBSp97gk2PHjvlslBJykgYp6VzP1bDWCeBnsngZ8MmUGYjoQEFf oqTSPNjRQyCwVSOdIDWIWKni9957j4R1W776lghCa6XcbLcRPR2+03yjZ26X lN555x1Q008RA5FW2+6DDz7Qm9sRQBqu7DgYeVbwJR5jzXBjS9FYWCbScaYP kZ4asASnBIXCYpZGxBcnazoQ4PvimnwmVYIqfr7yzcEnuv3RRx9BfWBhRx2r XV7cO2JGXcIixIiUoFprKnPX3GyOHGWGMymlfdBnm4+kTZEQg32wykC+xxqg BW5RSKy1dIwREoONjeiWdlr14MPjjz8OXdP5FkVHjN+Sz12PHsrjnQyR0cHY KvqREiNQGUdD+xKzVJRymrgSztMiRCLPw0uMu6mpvaD1XkM+rbgyEVgeIN+/ JqjCSab1JkbXT/Bp5dq0fnUrt8MMwA6MgO0CsL1datgWAONGh1tEz3fffbcb M7s+U9PeArAt168WpBLtrCLHyhG3LpQ0dgPAdiyoQeqg02QRvskqJu3pp59u Ta8jeJW63hYA2yulmkuYtiQAg6ZM8RtvvHFgFf12BsBu1vrVhgCMDhyGtFf/ P3t3+yVnceR5/19fkGSG4cEGYXz8sLYxYDhHgAQsoAVpLI2QRhIIoZUQGlng HYSFb/B6vMzs7M7srrk/29+jPDXV3aXqR3WpK1/0qa66rsyIyMiMX2ZGRoyy BGCLWJb7V4tb7i986myCaTOx5+JiMvdNxzQFH7Cg9n2ZJsyiQM42HuKYcBhB UIGtZ6DN3gwcfFIKkkIHFH+bfCbhE4vPMPn1888/b3/G3N72Uclx7hmzK9sB OuqCTz75xCsF9+5wAbPqYdfQdu3aNfiNfSnGY8HBLly4cOPGjWIaRIwJk/kD J7zrMRYT8Lt16xbKO1hBeec+JSn2Tf4zF1bKyZMnYQC8k4AH8IIY//qSIUOe h1955RXmox0MRGIZckAGM1doUEwheMAnwulcVZ+ixE8W+IhnxE3ywQk9Cxn6 sihMKvf3mZXyw7tFo4Ny3xf//OGHHwbeIgOFGCEi1p8anzp1CvFFY8iXL52h bwRO2sh4/vnntYU73Vd6F2jt7NmzxPLCCy+079c1Q63kjqXCQij4dZAXzXjx oRABan7xxRc9CQlQGD8RIwHSKILKXYq4puBT2XZo4xdffPHRRx/hSCVtjY64 7kYHdjoJRWQy0ahfoQsP6GiA88MPP8RI2jKOI33oVDTHLa18+umnlKSzP22h kAqhofAdSKJ1tMK7DYqmCEImQJ1btiPKFg0IIMzc58p5PXsyYdwNK6RqgsQS nQpRrkcAqjblSgg+4JOuHEMPJZ4cik3g+MWFriTk7QpCUlolfMGfXWP0mZrF YBgVyyRPSrSCZAhq6xHnNlGW8GkRy3L/aj+XvQzAdrpsBYBtS+ubAGAMZdEt fM+al9Bkc617kZHtogE7qPJtB2Dr7V/VesCM1uVvpioQ4siRI6DFoEGP+JfM 877WR/DAnAAsSap2i55XGVl4CYa8fPky68YW55PWbgZq0Xb06FH9pWnE6Eo9 uCEAtnWlakeoDJXAZLtqcwKwbp384Q9/2DXPq80BsN3fv5oNwHaZmNllCcAW sSz3rxa33Ef4lAMwAs6fP88gWvqxPubzyYWnZ1jY3G5LEVKczO3K9suK4Zr9 YrWL6MgUsoO//e1v2dzRCuO7JnwKgH399dfgFovZcVg3+JiAzz//3K/dzh5V jdOZzrm860mvs7/MXGdY8AkyIEmSuXPnTotxT5b2F5EMzZQcyqQMyXhXPWpj rNuu8brviz56+PDh0sGgky1o0+OtlYJmapCfDzL0SJsSx48fZ7/0C+SQo06H WeCBnyCfqejrH3300RR8imXkAYHI6DwRbEAqu6kJAvdhKl5WkaBKxeJ5QCXP +Y4diwyvBqJABtpUDpa0s6Fpr6Af2iSHKfd+phnWwq8mgIciQNKHzj3PnTuH HsLRbltYxYOqILKorQdWyginUBx4FGKHLuXkXDQD1PqXWLTbZmNH3mvCp2Tl Sc+DNJrGoxpUe+bMGYYbp3ARZWuTDTG17hnkwdV+UqF3RyptLCDYr35SQ9s7 yICydGtyQDMNnwQnHtCt4E07VAqLXFZr1BYclUZ1YNfhJv3JfS7lLEDW6shU HXZbtiAAnDOgcnUzuht6k7cRSWw9+IQSTxILlKh1PHZKbkABhyS8LdGoyl6E ryL3agKF1GwsW0ZGQr2vI7CggxC2+15YS/i0iGW5f7Wfyx4HYDtXtgWAbaJs EYCxhvqLoIqKudF2JyM1mZ8Bj5LisYOsG8SCweJwbjsAm3SyqnUcjct3BYTP QUjTwQYFqinag1/1EXoiwDN+pTMjv+FqAFYOmtUAbEOCKhMxe3rr1q02W4q4 RVA6Yrj06z5QVj/qTb2GAN/kJQ4mQT7YJO2hUesBsI2WEYMrjSIBevvpp58y wSRGAoXTJ67O3/X1mgAMwYDo7hx7bRcA251yTwC2d8oSgC1iWe5fLW65X/DJ 5AO6MEkMXNf5TVDda5tacppI2VlmusvsLClzs5XgDPlvqNC03ImSKZEVBld8 Y+Xb9fNJSDADPhX2sNOccZ6CSDMtQ18MKD95gGy9iEeY4dy5c5qL97y2cl+x GGejP//8c2CAjSuquXeZNlMitACZFKFoKnl0h4ndamSD0MnWaFr/5sbmw6VL l1gr5pJBbwuIVX3ybsnQdzENAPjss8/0DkuBVIR1FVHraEY5kIALQMUUPXVd axI+wWaTMAm6UBWxwDYj4ijUigzfhIW0jsjTp0+zmEj98Y9/XFz6Aon/1d3S 3Tq14fHGjRua6PyRZEZOvRMnTujN1cmsS2ICcRXPQROdfJUNpyip+hpaK2Bs Uc0nC1n5yQMn7hZV4boNpcdXSgeUhVwgNFCTVJHadT8Ea3E9+OQzFS0RjLfG bhU9GXCRfVRVCY71eJlivn+3RLa/rQjIpFshXoGR4BMfkD163E94n9T5hEYm tFGLnsQv/AyelVsZzRQbjP/kk09yGDt27Bh9LkxHveNLNeMCOwWApQ+ep1TG 0X9cKV3khOvMP8Qylf1nBnwK4jb01GngdM2hiGpo1gpqN31O15giauR1l4QC G+OrsVNLIePLk7kIapry72YAme+W8Gkxy3L/aj+XvQ/AdqLpbQRg85dOG7cI wBjBcsP5fv4VdDt1WGOnGHRNd2ZXWMu80Fk3hhiPHmA00bC9AEzTjKwvdTSM WlDN4b8EBmDtjTfeYDc9A1b5275iwAaWKLxqgIGg9AgKR35DDc0AYPNIqW2r YilASrB0PmltLRZ4k7bUU6AFyaCBiL788kvP6zj4JHVqRwun7ReRQOdZo63t 2r/qXhuEg1r1UM7OWOEfPVUaID3iAUSODJUlcJwEYAVG2Ok9GUq7jQBsp8v8 AGyPlCUAW8Sy3L9a3HK/4JOJ9M6dOyb2tk0sPBnE9fwlUAV1ZBY9zzAxcJsA Wp2SdOxVdpIf/OAHTz75ZJGXzJPdd1td7TzwKcvLXh8/ftwz7Q6xYsUUQjC7 EGLUEJyAizYZHnvsMQR067AnJ4+r8gvKycSTzI2JkejW7CkCxBozWgJlGAnu Qu0//uM/eoXZMv+jZwQEqOTv1JV8Fgpu0UQxEIATaAo7LLLHigsKJ6wO+zBO zfyqaY9hB5hhskm1SKTQxQh/WpiC4WqFtcOHDxcQgLkspgFSO2grMlXFi7hD CapgmEuXLsEGSMUj+WMQdOm6H4uc/9uaygDR4ZRGjbhP7XUQFCyEhgKIjbhY o3U9S1Be7LS6MoIbjIR9Q6roUY++6DxXL3xxt+hT8Ek9awLRPKBIj+heeuml AKSmYQndik7c+eszORRPbPiM1TrKdTRNK0smNEs9NNSmkz46cOAA+ATMgE9r SolGed4DFKCLiuQWqaPf20EibTCjMGXFlfU8xvWjYa5dvePJPKbof3dICwVP PUhjzWE1D3xCAwHCh4aegZw/mGq1UjivrtludLporsAUTevSqDXXejcTCcRc Sm/pHr7SjXatN9ToVsoSPi1iWe5f7eeyEABsu8qOArA1y2Qwom6NbRGAvfnm m4hkBe7Z6PAgmgJgbBlm9XgHc0xVydS6QcYUUoOubhVVe4cAGBtd68m/UPlw Agphqk8//ZSUsrwD2OTppGZVsS+EMHlAk6OU14MKQGYAbL2thjWzBw4AhmvW nNHHsk7XegkfY41RHsCvczfWH8G5rKdU0ALcQphren9tZf9q0KxdQoCdEINa kKNoYKRKCDBb8WZLTBwNcf31118HwEBfekjOO532JYKHy/r2ArAdKhsCYHuh LAHYIpbl/tXilt2HT5oo5sC4aA85mDxZohnYKbiFyI4MvPub3/xmymNkRslV 3tLbRM1Mhx/YcdOy9TU7CLQUgGjN1+eET7nEt5TWkEnMbDay7Pk8Luxb17MF 7733XkigYFlmZjPeZOXoQduFCxfyvQFIgJMZXBfTAKmwFu6Ye811yPjTu0XT pAc6YqdoRUUgL1ZS0RU65yqCQW8VEx6uYL/ypQmWFEihtLa4UJumhzdXcauA mRFIYZzxEbsFflmYS1UDHBIvq+pF1t/fAmSxm22/AAlXrlwp0UyJKTs8DSz5 jLViensFdMn3bD2VBg4RrOn2u1SCwsnkNeOkjwGtdYLVp+jHSPZ0UAu3AIQ+ QCOex5elAYkVropsVYXHQkUVVlRz+AWfkI0pOKdR0GbLd3ehjn+RWhZv3YTr /Kg7M/UXJf7FQkkwAchaJ9hSZHYmWHh/jY5jX//qIGqg8vXgUwiTcqqznN0q RAYUNAQbkKMD6DSOiF3rOCKoTm+1gmbyLMoK9n3T0LNeMK4p7XqhOOeETy0T UNWiTDd1aFv2Rr0A3Bpf6JxzgUarIUzDUyW4JiW6Z/kzYx8MC8TYzmfxuOiM SXWns72PsoRPi1iW+1f7uSwKANti2R0ANlXyMfvqq6+CDdrNo3srAOzmzZvD QK/ZIhPDDE0mxSNnzQ0A1l28t99+mxkFgZh1DzNbCPjHu+XWrVvAAwOqd5js BLWNAAzsYSO0zsYxweUZLApZ/k7Z6FJXo5agCMEzhT1HIUENcBUNxIgpTTCX MA8WZmtUwCaftEAgPZzKHojafNJCqiPuekmLvvnmG/+CxISDZm9FbX06GRNs deub278qEldbkW1p0uQB53JQhzyvX7+O1Fz6u905eW9RRxM+8cYguKL1nXMo 6rZge5glRygBIpSyjQBsJ8jeKAC772UJwBaxLPevFrfsPnzq5rWpqfDXrfdN nrMn8K7pMUmt3zsGmieWe2clbSuVMVCjxRSCDVrszzgkqmwIPqnKxIVUr+RH /dBKydGIuTS/YYGdZbmYjylX+UnKTdfdzPKKPuqAcrKPxvHiuH3vFdYT0NLE ww8/rFHv+nvgwIF8nACMscuRd/ft27e9gjuyfeaZZ5BaZCd/1dDpJBiWK0uJ 7SYDKbDILHXHmsyNDn30bunen78qmUreBwtpnSYAhMAJ49jDo90ohwpQC2h5 OE91eBJiAajKbV0cBjR7Md8n7ZItY4ev4OigeZDtS9BCHwFs4Ae8p8URz0o9 BPXLX/4SPmHii6cUGilPk6bzMtKbXYT0Ihr0EYQABxIUQB6IhWfyw8fL8ObK jyt9QMDoEWNwYJ7JBJEdnGFcPWlU6Qi1rgZ1+lLTkIZu0qeoNbQpPAbBwmJW DL+sJIyYkdh6Mh9lEasmQULJoYqf5hWdRXRr7gWlkCrUTaVP+sEPfjCimSXh onUVTxWCnZ0xcKPwqehhusn8YLHwxBNPEA59oHVt081z76MZw+oD/SUYpdtd nLzni50d6w79SwJXr16FtHfHrWIJnxaxLPev9nNZFAC26XJfAFhwiBk1aUMm bcto10yex1FQZBMAbMoAtXUzcIUp1ytZ3lOnTpUUj9FkhkYa6EkAxoyO6FVV pQYAoLwk5INaBui5556DQ4qbtDqS1VYAWBtBJMBoooegGNkf//jHvVLI2bff fhvwCCuuKXDEqEGdYAYww/bhbnU4gkm3pQHASn2STxoKh1fYagCWM3NQoT0Z KEv3gYJtcXiRoHQWE0yMk7cFV5f5968mXfiCYepHM1iFWRpFwgWUQG2RN8rF vLqeAjJ4wJPeIlt9AdLPoHMrJXGVLZ0aa0vPpoQo10cFT5u9HXRf9q82DcDu b1kCsEUsy/2rxS27CZ/Mk7lRdcTDWpkSzd6z8/RNvs5YI/LSpUsdorHFJuEZ 0EtzRchkgq1nC3SQmzTzjXfL5ybw2QTMA5+KrqBCExcK33jjDRY5c/zXd0tx v1WCclMcxlmWNRfvqrLcNml3gsZ8nDt3jgUfB5RF01JD8ZpKz9FOBTYPrxQ2 XaMPP/wwUeOabVUnK8Duqwrj4caibhLIiRMnAIARaSqHbSwcOnSoeAJvvvkm OMF2E7uHUcWiDd+eoh4xN17UemRDPqCyXm4DStMsuE7xb/n+iKJs1CSjrQJz 9Xkk+kE2UvVmibaLEcqSwj9qhlgw6GF0ss6lAhzRoshEd+s7ZCPeh8hmaBjx MK13obU6qxMxOkNRUatD9X6bYEUSo4Hq6Spru5EjlKi/qgJQS8dTiCqkqqdQ UbiGuApahQwEqAEBv/jFL0r3UzAKmoPxd955R0P0xMMqpFEeG/G46h2NqgG/ eoeGdK92bGYSFIIpST75lOSll14q7FiB5YtoMbzyjh8/rhLkGRolfmpkUTaf UV44d80ZUCnkmuOFuAhNPfREL0TnwYMHi7fWBloZeYB80kYeOa95qL0J+FSU LTqmp7wL39bX/hq8nSPPCKGAWZx+8MEHcUp5dCXUOqfvlqYLYFtINKUEAbsA vZbwaRHLcv9qP5cFAmCbKPcLgAFOWmwG1qK/DBwDCrSohynM36mtmzVbXw+A jb0mPYWqyaR4GSkQaGQPJOSSGLLgI2TlFABrk4Ex8j3aGHqk5rRPYuxIF9A0 xAax7/rLN/nMbxGAMUyYIqjh1B3NBKXyTku1qF2CWi8UA1F89tlnBOtFSAZE vH79OmrHPVBvebdrjIXKJ5ABwIZPWoKCBxCsg/JJmwRg2euOaCnPyZMnS/kd zfVsuGV1KIbVZZ79q/qlJOOo0jV6pMyM5QaCXQkquEhQhbeiUWvegY1yIDxf dCzrRJXvRJDMtvjAYypX9qscwo0IGI8+mGoaeveccHZ//2qLAOw+liUAW8Sy 3L9a3LKb8KkIUaZ683+2kulkm2YflEyWThOYthEmmqk1DzMNk4+FLlTL7EIX Zr+sc5GLmg+zyHO2OwM+ZeNGdAVWjIViTIvi2D7D4bulxMc+BIfMycVsnFpK l58ODGDNo/ns2bP+bQOnc7eiaTGsnZqN9BwjQpRWmFotHjx4EKrBNaMwZdYT VCeV5lv1MMpFhHjiiSegkaeffjovI3WqKh9pRhDYYH/JP++aHHuefPJJBORi RKNUhUcgJO/oSsSz2hRPLyDbhN/2kSbKM6jRcZdQ/5bMt72UKajZdpZ6cAq2 5WCWP1W5C4cnFYSDbPYXVEO2BzQ34qt7t5j2/qWToN3q+3SdwdXFukOFzz77 rPq91YZhXBT6gJLAV3QGhMgHbGQezG3eKCAcSJVsO4nOoat9nuJW6VDSBplQ TtoQiCdz5GujLEF5HgGeJIcuF6x5a5LMKYCVSxtuiMf1SCSkuxGgLwic8SVz HYRZxA+/LDWQDJo1rXM9CQZPuXOHxjuJBjIhNFCNSjyxUrprqUU0qESLR48e NVjy9u/ENkFNdvHm4FMQDrQGL9Gsrfbr9K8Ki1O6WlAd/JXpoKC1WDAxTs0w s4tKPv74YwqgBh1EW8xCBLLT6GsJnxaxLPev9nNZLAA2Z9lTAGy4Hp0+ffry 5ctFTLon/bMBGAKYjzavSorXXllOzgEwfQpslLfFbFwI8TUFtRqAZRwH+Cn8 e811O17HMYhTgUwJZ4sADPpiZHGN2Yz7bIWslxk7cIhSaRoLKsfOQKo+F6uB 8EecikngAYcgtVRH2kXwmt5T6wEw7UIRxPKnP/1pfjM9IwH0cLjKt82oVLln yqGTHx36dW6bZgQ1T95ACo9y+kDUhsObb75548aNbYcE6wEw0Lq9oA8++KCz wjmb3uX9q20BYPerLAHYIpbl/tXilt2BTyWSMPUVQDKvdQPcfL7mUcWMUvQG K3GTKvtlUc8imJPN2BGfjbt9+zZbCQDkqdWhEjsLvOUy3SWpORtdEz7l7G1F zwyx+OwvS5H7E1wBz/jG3AuBECx6PBlEYblKkZY7t5oZ9xHQoHyLn3zyyc9X yiuvvPLee++podfVpgZixJGp8scrpbSMjDhUw7J7Bs7Rs8QLM2R5Ue6tydCj xP7HP/6RJE22bFN4DwuMlKpGkpoRFfzgwYNm5sfvFhUy4igs5Pu1a9c0+uWX X5ISCZCSdhE8XNxJAKcEpXL0ILugVVopQzRri7xcqogU2qQq6NHLwRvdNwUC yQS/iEdPcU0ZPmADrkAVCQ9gVjI+n33ItZ4Sap1WnDlzxitIRQbKfYl4bY0w FwU9QFWZgHKRInMkFb8CzTqoCAOE6fvJNDR6iuQxWJ5EutRekA7S++ATVaGW OiKo72HQqD2xkc0wn6s8lzBV1M1jd4vnc/XpuqVKSGZ4T4UhCYe+FdNMfxVA vsAX2C+462uvvUbU6XCHucSInXa0aGlH0u+//76fkEfO/h3LgTSqoYeq6iEK QvNYcTb8StRXr17NEQ4XHVijqmAX6ERSsWGDkVuBTznP66DuM+bD1soChciY hEaa8++I9KVRYtHieqEh1ivkQCbffPNN8eSLwVXA2x0NPbqET4tYlvtX+7ks HAC7Z9lrAMyLuR5Ztne4M/uu4mwAVpqSAh9lSnDnQ2bOkwOAFXOAQcwnZwC8 UWYAMLZJT6GfQWSUVV5avRHJSusI86sWvZ7RDD5tBYDRxhHeqmD+s3VDnUAR gvULOqGvjqL0iM/aQmF57qJcc1neAsWnD/mkJajOy6aOWWcDMDUMWDW/Rq3e vypSlnoQMyLth0/0C8qxqVN8n6OXx0aoh9kaFQAjEL2mH8shrvV5dr02VGYA MKIGgfIQ21AIqd3cv9ouAHa/yhKALWJZ7l8tbtkd+KRaY9n6N59h5oANnUpf Mn8pLKf5oarUyQDlY9xN/O7ydz6V4xAwU9Bpj20CrU3CJ/a3VLw4UqGGzp8/ X1aXNkY8g7C3336bpSsjbS1mHLvvDzzkseNJhDEuw0da5VnqXIPakrpz546q YCeQII/xJ554oj2Nxx9/XCXsbNjJu6gaYImR/eijj5iwAwcOEMjly5dLrxOq UScAQDjjAh2TB3K8++67YQ/Mqry2xr3CkduOYJ9//nkoorgQhRVVYSFhiQuK 6xA2QKhOgtJlJZTJxSgQQg9HIAivdPoJVnmgYAgIgxuZ4LirqLPrh54pMVAp p6HWQol6kXxyxBppAbs95y2SVCfhe5Fl118FNkfY6dOnR5BSxIDcJS362c9+ pgb4H+P6EVTD3Xd3T4FVUoAsIlW/hvJk01/Ioyf1DjZLSOTLkX+wCwUY7Nb/ CK3wyL8v5P/MM88M2KmnTpw4oWefffZZDellFOL96tWr3TYtdhbCcErsJQr3 6wglEZAr5aIXLTRGaqS813RBybhJG3ngIh2mwIjRm1pX+TiVplEeK4ZGm286 iDAh+fajvluBjsa+AUXghYrNb6384MYyCQD/Had6y78EtRX41FmqFlGS950u xqZ/fakvOlj3N+CKEn3nV5B+0yBEhXoc5fqL5Ivhtl442W0pS/i0iGW5f7Wf yyICsPVKu2R7FoDNycLYKlkTgJVdmnULLzFbxT4qKR6sNQXA1mwit6U5AVjR FQqbGXTJt8qvnrl06RKLXOrk4BMatgjA5ikdLQFpawIwFY4IpeRTLFMs6JfS 8ZBk8VRnG6lgVVfyZwOwjZbJ/asAWA6BviTPMBgxduKp+IxTWlFc+nveT5ws 6F8TgG0XEmibsXua8wCwDZVd279qUGwjANv9sgRgi1iW+1eLW3YBPqmTQS+E JotpgBenekPHJVMlT3hznYnC8rDUMKY79oXdhAE6oDHfXrhwgbEo+PYMUDG7 DPjEIpAVu4kdMAYvhXQAMwAG8EBzjPjt27exXPaTSUiQ85j1clYGFgJdCrCQ owvsYfZ+7rnn2r4oiMHIXFOOvI7n8IsMkmRtGfEOg2qx45WMY4nhrly54pXS 88EAxT9XDxEVTLWcvwcOHDD34vHZZ58tqgDyWK52xtDWDkYxyQubkAd7UQj0 r+4A5LxeOmaQptDlgwuvIKB8cKU+KbpCJyxjoy+8DSeA3IhHwDjuVJuOUIO2 dEF7EfjKX6jQE5rLjQrX3kWtSlAObhXjNF+jDtdG1KkkTJ2m0jR7WEcXospb 1IzYO8+dCjwbqPZ9ccmYMPpQYp2RiXKE5NLRviz/IAopgx4p0HonyDgtjpYu vnHjBuG3geldovZwWY0m9+jqPjVoqE2zET1ej6iW3lpcTEK+7hTkp6QviowB pJWzifTKYIj9EhIVToTACYrmaB3lntQpGiojZNmdLCvQrMI2iCYzkrdPW5x5 g8UIpZN1Sg11NEz3CjqBAAoAQW0OPjUcjGJjxPBEVX4I5UZUP2UmW32N8iLh Y7zgupueo7RYQkYKjxfyp/xanxF6a4tlCZ8WsSz3r/ZzWVAAtroJFoTVY+/2 MgCbXQoJbsncLt+aAKyccQxTUdAhipInlr4wJDMJwFaXLvJjYRMArMR2DBN5 jthQxZ8f8MljaC7L4eYA2DyFhP/u7/5OzesBsNI3F3MMCNF9uYeVPTBXq1K3 rFl/+2ME+/HHH2NhTgA2f5ncvwqATcKk0jQHAwJgQJEuHmeCc7bSnhLeZwCw LZaAjU4Et+YEYBsqu7N/hQsdSiHp53YBsN0vSwC2iGW5f7W4ZafhUyFDTURm PwYIHjAHzpM38J6lNDqgUYto8x4bYd4ranqZU8y6mGLot3LSwT6a/6ECM2ph go4dO4YL8Abq0BbzXexr6IKZuydI66wE2mFfLJw7evv+97/POmtC5eosgFVx mTBYWr1CLeGRncr1ughF67UyQoPeunUL3GLaHn74YcASqUUxKgZUwYjysMrh pzhURFp669L0sFxF62pbCVCEjtTZ7TaVQHevvPJKvawV6EsrrLZ/i92NHWqQ J8+a4SBWF2hBL3/wwQcEzq6hViUI86+2oCbfoLawpeNIVI+MI0jEoxbNeSJp t20lQuia3vA16iAVtSP6FhbeffddLJS0SF/oL/iWRt2T8jaF6EM7YMlhBHwo CVHx28s/SFYjO6GGCgKvl42d/P+7B1dMA5iwVMhkMugvbka+TKX8UzN1KgVS CJPKdZdhBlhtR4vYS/WoLepd5poRN75NV+1qpUCmhIlHv/om7oidft4TXiJD 72AtDFlQtWLpF8oe/Wr2VxN5D27l7LLAcQ09HaGhrnn61+jTBSSJDGqzdZBZ koVuwhZtzDRLA7c94ExlCZ8WsSz3r/ZzWVwANupvkV5qYNN++e/2MgCbJH4k EOykT3cwDXndrAnASopngu1a4owoXpMArEhQ5d2b9LnaBABb0/aN7H4aunjx IhvKYuoIhjjrDw8EPDYKwKbqH47WanjjjTcAjBkADI8A2OXLlwcAm0e9R8JK zN64cQNwBY0IRy8YJnrZl8DJnJSvVz8MYFBgYQCw4a6/JgDb3MX/YrJ1yF6n QIMzdu02WupxikFvsVAoM12g6996661CmG7xsGwX9q8ag4WSA8YMB5KfDEyx KGUJwBaxLPevFrfsNHwyf+arUxhtExRDPE8K+3uWXHTQ/Morr7SLohRjnIGG EMqcUua4TdSPwi5elUZNE4zagQMHiCvfHotc0xTIZOLtUAkemDPMYMQXMYkN BVRAI6bHX1yABGxoZ3AMx4kTJ7AzAhAhaWQ9Xh37Pcq7oUYCpK0JwI9F0wUP PfRQYImNO3jwoA/5oqOhM76y5HTGV7aaokZMBnDoNtZ//+//3a/o0YQXX3vt NdaZmc4XqJBTQQ6GT+VqLiJEQQPmBJndJUQPCbDIOregjo+tlI4si+aBBWCp xNDlzVl9JJq5J3nEk2HOP1D3pK+RJgqZVZj0gn63QaojKMOcgQuCfCUBpId3 7twh2LgomC3Up5Vi44dX2T7wLLBK+EWaLUVRHV1YTsTji/A9Q6Tov3btmpoB 3Tz61EMmQe4QfpkZcVEaSlwQUXh4vUTP7b+NDIblsilzYjtaUCVBFQM/T3Xq VFQ3zxAprud0WQ8SmyuwrEfQhh2DK9evsh60qVj0dfqA+M2N64SZs1lhOvBS OC/16wii0wX4nT+88IzS7pxu6kppx+J43Jwz/+yyhE+LWJb7V/u5LC4Ay0aA B90Ez8HY/A9O7H0A1rZeGf3QDzaUr7ADrPaUsmVTAGx1UrzVXbYagDGI6s8J RxMDJnUiuVEAtpoXDyCs2BSw0Aj4qS2NEhQd8OsmAFi8kG2BiYbj/WRYiRJZ EtQUANMp2iIonVI0rXt2ShGcbt68CcwURDQAiQWWGv1+XTO6+zyljVYSYPfH TQEArPM4csMCTaNvMNUUANvoBuy4X6lnaZEef/fdd3UrgWzLtcGAJdUiKJTT VaiyJAWhF4IauHcrDe30/tUASB2SauXNN9+kbDvnpr5zZQnAFrEs968Wt+wc fOpgjrEwgTPQLLW1m4XbtkzgWVVTHJpNd+btQyvF1MFkb4WXccaUPzxDzCJg Id+YdhtYCrav2N1tj2yo5mJPWUEjkoE7efLksWPHyMci/aGVAj498sgjbKvJ 0K8XLlxo22q1GZ08QKyMLQ74gZku3GjJcazN1fzQRAHVdAoBgj0FeCwv2z3h wWi0e2fdhcynC+WxoJAYzIk7ZGA2KDs7cOvkSeLghXXTp+VohgQevlug2YMH D2qizSUsaGXNPZn1SjCDbEfsdAgzR6NkpcdxoS9ee+014I1IWXAkjax8sbOe So9cNnk0UadTp04VFZ+6akL9pUrEhWGiIwr9NAQ1e7CM3Zi6m06mTrrbuFat +usO2kUBilfGvMLJwX4LHP0yDzvFraKxWjl69KiqCArxoy80B3BSNhrrMVLt eHrUfM94sKNHADAgFqRXm6FxcKWAytS1rEx0tV5I7TeBMD2PdzD19ddfB//I BwvaAqIom++7KbAtk5WqwFfEP7tSzBvlANpizVNlCZ8WsSz3r/ZzWXQAxu4U Epwt6HKfFajJX0MLAcDUoP63336bxWSCs2Xl8mPgWB8W+dKlS20IrAnAhmPS qHwGAAuxMJSdLv3oRz/SNcWp3hwAixe9bMJ///33dQT7NcwxuwbdMce/+93v WJx57O8MAMas6Iif//znbVtpIke78uD49erVq/p9QwBsstERQpPQSIy4nlop mNIF5Lk5Kz8AWLEdyv2kTrpKaQfAq6+JcXVC802Umjt//nzXTpVc37du9Osa RIIof/M3f5OgclZv6Pl+o/KfUXZ0/yo1pjNWEARFrygSNQbMtrGVXStLALaI Zbl/tbhl5+ATK2aNfOLEiYJvnz171hy1aUfcUQohZW3bnbLi1bAOL730klba ByjX24Z27zuPY6mL71SIoc4TC/hTWB722oeytHSFPxfx2QR7DGhUMwkASyp/ +eWXUZ5dUz8YU5RICJC97pvcyIvCdOfOneDHJMGsmHke4GGR1Xzu3Lnf/OY3 7HKiKEyrQj5tK+XB8vTTTxORD6bZY8eOlS9mzUAK6xWd6HnA8sMPP7QqZ9fK 9KdaHzpbLIiTpv17eKXgt2R/SMXOapBWLHHsENRI5FeMCNWqjfw79QM50J8v ln9zmzly5Ih+1zUbOlwO0RWSgq4Wfmo4kCMe2ixSvW8iYHjR4F2LxRJRw5Rb eCCwWGcGQkjDuwkfO8899xyy8aJyn/UUGZJYNx8tCuihrqFm9+yR4ZoFzOdX Rs659NNY9aNco8MLCw0ISPcYXNoI5VKzXLPIXw2laprkqDQ68VJHJPmqasdP i2Ua0iIC8KXTy9qsZhA0Qd0ThUICBUCjKu+88w6QTz4Fky84SdcriLTYGggu S/ic/V7XG5Wwk94speOIgVb8tCKcfPPNN1tMfxOERuGtW7cMlnFajTUi3cZw 7kv4tIhluX+1n8uiA7Ay2Zktjx8/buH5u9/9zvLZzLNR59X7CMCKgsgEZNHY LAQUGwoayf2p850pgmGYnKxIg6hJGCS4JwDTdOGncqcfTlabA2AFMh3JjusO zXV9zwOlnJun3+8JwIpw1TGof6GUEBcJaKUENOv5pM0uKCSKCxcujKgIGKEA xUsnorpgQ3WuBmAqL4hW8Vq7kgCAwdjWC7DWCG+1dXNPfyiDJqAvYmSO1zyA 3lDp4oOqdGsJC0o5Xcob+gBjFIR/G0HFzu1fIVK/t8FeCnX10z29sKNpmneu LAHYIpbl/tXilh2CT+Vig52YBlbbHGs+N1NtZV7tGMUEXkAkkzbr0xkN82qR qy3wxtqcqbpy5co818zHSZOZx1xatj4IJ2zTGZkKi3sJJOCFSZqNzSbP49pV KKftpUuXALzChk8mlMGCf/P3ZriL5d40CId4+O233y6NTn4yRVEgB4AHcihI kZpZAS+CMWDY8B06ePBgEQkOHTqkodBauyWaY2F1yjzXzMfluxzj2XqC0gWk 8fTTT+ewpE7/Fr+o/ZlQXHn0/PXTsWPHkAot5Jv9r3dLHNEZVljNerBQ3sWM Ku9eW2T5/Ki/i3Ia7WS2awuQZCBqHqgwmCIEyJacR23d3KzHg7VdkSNM9ETJ cF27evVqoTDCnxXs6HqC8it2WH80DxCrfmhKc/rI98TiG00UQaJIYvTw+vXr 7fXlaDRPrvOc5MkwHS6ol35pR3RkkMxvqvukxV6Dq4drVhcYB6Ku5H9V6DDs 01jq6kWSNzoQn44VQKy24gUlXlGzfs8zvLC693TGix3YrGTQkHmBsFoOdNev aLTtuQXe7pkh/buJPSWckjyCEQkT6lOdXjQSAqESnVxvOk7sKK3OKEzLJeyg uYByW6l2sizh0yKW5f7Vfi4PBgAzLfuG8d1obJ89BcC6R/b666+vzsw45WQV XPFYESlzsvKimu8JwJjCQmjO4w01yvwADID0TR7pTMw965zkaH4AxtDoJtq7 IS6mWp8BwDCi/lu3bm3I8lZnitoR2DwAjBjp3raEC/ju3wOwrqNiJC/3TddZ +Ii2YacAmKFt6Ono7c0lOsoO7V9NATBKVboi64Jtqf++lCUAW8Sy3L9a3LIT 8MmwZaCt68sMyHZ8/PHHvtyKU6uZIS9Ty1XrylKEWNWCDSZAlIMo/pZkp2tx HanMyG/CXhegwGzT9beR+8xcWu42ZoiIcjlmXjs+W9MS/WUieTRrMgL4lOaM IatmVgAwMFF3Ud0zZXMDOV588UUIEBdmdX/BLQtzyLDXfcAvqt555x1/2XdT 5XBOJo2imvtJi35SIZjBWBfdFHK7du0a2siEJAnKWwChty5fvjzDZT3IeufO nTLbkr+mc0NCmBZLMO1XK32U46ssyYVy0mIHRnqq6Pr5txA1SZIPghlHMLsM gwUGH7++9957Gj2zUjwAfviVHAjn66+/VjlBlboOI8SII02Tqi5jCmeAqyLD U/sCuUMCvUsy169fL+qsSlSOR+2WqK6MhN040I9IKoNhHjteRzMR4ejVV18d vxav46233vKisUBQgBmxqBB8wo5v4EZqQ2IlzibkwmQlilJOl7t5PaASsygn K5TQVTVjsL0d9QNyFI9ugIjaHZm4EexJBNfoZA7HUgihHL+dWafJ2CQKLHSC rCNQSIxqht6J1HhBBl6IothotCVdNTZVixLdVEZLqKyT9DXVzzBMIEhtQ8/Q Lisl2rAzzug1RBPos3HU7cL1ej/nLkwhDJtIQslXKwUvICLufIlZv2qXhpOe Ojd9o3Dci0S8Cs2NqMUCNd6uLawlfFrEsty/2s/lwQBgX3zxxYxUMmuWvQnA OrgZHjjqLJC4OotWSqosXUiGafjJ3cKOIKyUf7MBGLuzoQhOmwNgoAUk45s1 6yysaDE5IRxPdsl0fgDGyG7xnl37iusBMMqmfr25IWub/xim0pbwzD0B2Hbt X60GYAQINtCoeU4e1yzBhvKVlxwhzK+/AmC56O9EOM3KDu1frQZgberuUGqb 3SlLALaIZbl/tbhl2+GTGqxGAQ8zXt4XxY3cXM3jXrwpmlFjSVUIlXVSFjAb ey+Ml7aYP+2yhufPn7dUnDwQnLwL7y10eoD5LphPSWfUDMywO+rJI9cr3377 bembJ+HTZOypDrDKY2viYtwZZTAJkYcPHx6nV6hiMU3aly5dYo86G2Jhuw8F FbSQV5uqdI3HugVWTAaE5bnh78g6V1qcYpirHDBj2uAldXog95sTJ0609fHd XYNIbtBIvlL4mtoVmYyBkNfNiK7glXK0dZiIZn0N9Q0AABO2kzDgE+nhFC8Y 7wi1u2b5m+WXDuMVMt33dS5bCTtBoSrXTXoWdx5g6cADuvrdxPX5ETCh42Ct e5HWFXxpyhcrjcIyCcMzIzMgsilt2xTxUnyPd955h7TzyKIteNF3ECM4RIYk j6rcmVIexCclbB48eND3Regi59IZk7+aR/rm4FP93kEk0XVcSCakhClcI0Pr RccazkuTfNVfBFW0DfoG4YAEmdEuHdBMXYY8/HYdz8PkjJdGgUY1h3IP4KsI /0Fl3HUg7kvqTcKF5+3A8cKFCyrp9ihV+fuVApOcOnWqfi9VomrjiOi06EmK UbrzVlidcQ+/rNXw6S8rCaN1X3t9Iwlje4yFjEMbcN6h52ovr8mbg8ZOO3hq 7hl/aRc1CFQ/sVLKOaXOnAY3veNklKGcJlOJsitqa4v3E0dZwqdFLMv9q/1c HiQANmdtCwHA8nQqWS3TwJDhtDtoWcnhjI3sfMvZC2TfE4DNL6jtAmCTQbri izxHImYEM6yqjSOyUr+a5wFgmyvxNQ8Am6cM1vCV/9hIjT3l0J4HOMbXBGBb 37+aDcA2WroYm997eXN+9rOflU4IeCi+RMEltkj27LIT+1ezAdjiliUAW8Sy 3L9a3LK98MlCjMW8ePFiLt8gBFPS7ftN1DZCXhdVoIOhzpvYiNw2JhMKF5kB JtEua4UpD6OHwepsLn9veMn0Uuys1ubmT88DJybqvLkYnXzIO7LUyhR8qrZy vsBCDD2Lae7qAIu5LKyQaR/xhABQYQSMNEtjqlv57KZ2oUGm9vz584AHTHXl yhVABeAp4XW7Og899FDRMiv5P3tX0whQM7tcjpviLagTmirrHzPR9tQAmUWu RoxWihIP7TCR4+6V58kNPR5gi7ton+eMjiAHMtTiiHtQJsSUZxI+6SMV4teH oqOroXRyudb/1d3Sbgmyy8GXuEaKQFXlgYYpgGrq4G9cPQARz549G8Ikh0Io lAgJXPlu5Q6XruzAF19tBna7DdmIRCoVGvtCxZUiChARIiX2YkqgDdd6v9yL 3W4rayFGyv7jQ73mQ6rYtVZiRwPlmYJPOCoGCAIwWE5ksJlOps+FodALv/71 r9tO6Tx3BAX1GbOeBJyik+jG/m0Lh4Arsj2pdWSXCbFAVek/y6tzQ8seKM1Q eqjLDqyUtl/ofMqAL/oD4WuXMpcKJ4+m0mWSFWLymwJZDQpPGiBdpiuwBgnj S7tJuIuTOnc1fAqv4qUopkSqWiPorbfeUm23Vou+Ne4teqYzyrSFYmvdM0gl Rmo26ZygLzqeRkZhQDrG7Y4h5aQqwMkmTlSDuHrWfIK8p59+uuC92xJUeQmf FrEs96/2c3mQANjsslgATFvMOiLZCBSykvmrq5MoIBkPeAVVZlqdWCY4tc0D wGaUduS2F4BhFg3Xr18nBxXmNx6W6MQnv30/4Yg84aiyB84JwOYvAbCbN29S gHYXZwOweWrL+qOwm6F52SGV3PRvp9j6GjvI7oysSF/bvn81G4BtqKCQ5Amk FDbFH4M9EEzlymO4vcEH1ivbvn81DwBb0LIEYItYlvtXi1u2ET5517zEjpty DWF2vLObjdY5jvzMz0hSIQNX6M7Dhw+zs2fPnu2YbM3XrWq1C2B43ltlPS5a FGvy6aefwjMsnZUjK1P0g1//+teqLalZ0GiqTu/69eTJk+ATC45NraiNIc6f HB4o41u7MeVkMduDNIgpelUAo20EeMOXBSt46qmnTHpXVwpr6xWr+OI8I3LE LGpXpCM/ZHi34DxmfqJQf/fiNV0MojfeeIPoALw14SVicIq2FvsQF0YAIXyx 7CBB/agVPyHg8ccfR6d/Sa9q2xFa3XfAJKa6agc5EN2kw1KBmNA/9nlwNHax OiaDHiGo4n3BM1SU9SzrkG6lUWv2e272foXuwBg1I5hikwkyso86jv3FdZtX hw4dwh0zSnS+x/iaNVM20Jc80UwVISXkjdpKjNjVuUJqxNfwjiM6AoE9Os2k M1j76quviEWd4BMK1wSEwbPOEz2ZqnTwingssJL0mYhyCsIm+VA8jdIN+onC 9cYIrEXOBGWlo06V0zoYUm34RY8uQ3BJBmOnFAO5ZuVD1aUJ+k9pDZDh4o5O Xay29qCCo/QwvylAGrSjDGRSBH6Vq5bSdqCswu4AephO0ky0kbBBsWYIF98Q KT0hEAQUrJ4QUGtklQ8RbV3cwJpKVJX7mQ+IXPNWYLuXpGpUFre2kUWxE9Tm YtX+ZSXdOc1JZ6IQDfOEtZ9dlvBpEcty/2o/lwcPgE2V4Ruz6ACMnepiI3vd dXJ9xwq0PdghlxbnBGCrZZ6gsL9rAIy4RiSoU6dOqRZHDO6AImpjLucEYLPL OD7bBACbqmc4klVb4a1IPiyRY7neB4HADL285m3N7d2/iiraOCcAW6/kUoha XWAsQEFlEUqp1EnHvvnmm+2N0D67bO/+1QA/8wCwhStLALaIZbl/tbhlu+CT GdWKmAXsqpFFqCHcBfYN1cMKMLhW93kTFdkGfAJ42GXgBw4pP8V6Rw++ZzWs +llwk3+eGD6zkj9dKWUeUflvfvMbZr38JoxC0aqnIiYRiC9HptqgkdeLjROz nfXki8K+mLiYHnNyJz5Fq1bUUybia9euFc9nuKAXibGzyCIP5OV+4sQJtgPj MAM6MYXg4B+j5pXSsRWYqJAUI12y1mcnMi7vHlILK5EDfKyNqE2Mi1/1BfWg Kpgacden5K+2rrQDQgwTjJRTDeGUTpGgSkitQs/AMEVsQAC7f/HixXxyCpRU +Pfiaf/HlaISZMxwxg51dyuBJhOy7iCZokhVWwGpqhC81Fk0qtR1M47JRtoX eFINgIQ6xyFm4Qh8pvOBQKCRrDCFBni4ME0kiYaOhsv6R9qlv8G1ynNrX53B sEzfGC+S2EcffURvtTUiiXUNrWirHU8bei0EZowRPWgcUWxEer6LCaV5Kj6V mnHqb/mSCEqdBdHq4l58abe4Je2CGh35oSmY9RMhX79+nTaO23ygI44gf18a DgWMJSgwkvZqrgyD6TMlLHSJf5FXeuvVC5wC5vuJoOBGdRo4BaEtZETEFEQL g8+uFCMLAXm+rTdM6BvKaZQOxbIK009YDqgjFv2y0aPDdo/xQu2LkVt8s43C 3amyhE+LWJb7V/u5PHgArDJsscmfrS8qVOZyQQEYmzK8ktTDipVS9v333x+p ZBiX+QHYKO2lgA2If+utt3YNgBXnszSL+fNMnsjkJ4OX51YKYs6ePbuJ23CB DWLXYp7MYbDSC84DwEY9nikqKareeeedUEpKlVN6AKzckQCGrl8vlc/27l+p ilKBQwAYYFA2wI3mfGm3kKrTqOHyTaNy5QoiGnpbP+faUNnG/SuUU55c73L/ A/zuCcAWqCwB2CKW5f7V4pZtgU+547K/hq3afvjDH54+fdo6d6OzdxsFXcwv ke4jjzwCroAT7777LpNkep890Y2gAWxi4YlKnmIlrp6CFbRRw3SiMNfr1TWM 5DVhMMYILCl8pVI2Oh9iljk2vRfdCG4ZSYQ7JBqBFHJ37yZ7UYYOrZQckPLV eeqpp4oT3imn6b2E1N3s64D18uXL1tEl0GmjoMAR6IHBdAFqi/kzQ9SxiVSA 5+TJk0/cLYULyNOGidELyIAHgIEpGDAZoqHkNYXQP378ONpyVoEPh28VQYFP BDUCmUIX47Dp9u3b+AIkPJNHUIeD/hIRPAZBgZFrhjNaXTymZyEZMJ6oQZR6 DWsPP/wwTWA6QdNSy83WyRGvg529ePEiHVAbqsqyd/DgQUJDMMsOOwGEk/Gm 0GB8lbuZyaYnJbbDnbfU0D1K0As+8WQAXlsjANSUtge6qJAuo8NPP/10NJCz 2tT80EMPgRlXrlzpSGvNSqZKAUOgwVILqS3J6zK1ERRV1CLa0sDuNuK02w3d 2vBYKZC+t1K8WyQNVXkgh7qCR43Mg5E3Riv2iZcMuzZbhsGcvsaJtq4kzC+/ /DJcOhUpa5QAswqpGayuwq6s/vXdQmjFQsF4Yd7vOVOVxyqWjVDMFhv5/Pnz VbI6zNrs4sl8C0unWK5Dgt3KjYAlfFrEsty/2s/lQQVg6gEtcGTyBwlMlSbM /IgWF4BNCqoNQ6bWk1hTG5LYGvx6ZU4AFm0wA/tIIK+//nq52EJ0uwPA1uvQ Nuj0u0q6kMjobyjbYOQVQ0yLMEYBMZhyXVZ4rvkBmHoQ7JX2wX75y1+qR/et B8Bml+3av4pCVBl6L7zwAi3SiXqTss1fQxqFgNIZFLchrIgvXQDz7M5twdVl G/evCv7ZTWSTgOllG4N/7oWyBGCLWJb7V4tbtg6fOhNhODpYAXXee++9TtPm rKGwP2a2S5cuMWrPPffcz1eKqpjajz/+GH7w6ySEWF2Dn5gk7MAw1pgmye5z HThwwJK8vR0Ih+27ffs2u8ykdtF+GIVMbbn/4Baw4Y033ijukPk2fKIq1TLE bKWltLa++OKLzno6lVMt+FdW5alQPwVSQEZO6WrLexZVuaN7RdPeVYMKVdWJ WN7CwYY81XFRzPPChvsbqCM6OKEY1+vBjKJ3tq9SlC0kJaXAADBmKV3QgE58 pkI35H6vFWa6bNfgBAYhigTVDUdK1R03z9y8eVOL+jG+SH5sGgQACr5Khp4x 82OQrWwPBGFtiRAXwhjTcR9wtn/RZ599hrZARduYqgJXQjuFH5+dy0+HFuJV PUeOHCma95CVLoBVLly4UFQNSAxfU/GmurgKzJQtUd/poIsXLxoj3Z5DDy0t XMkI2hC2J7GpuPqTAV0JygMF5iLqEZyq3blCoxuVHgumrjf6AnKkanUAED70 0EONl7aPLDQoZ4C8SgKTTLNqkUeG3YUk7RyKrA6KQ2WkqIEmUNSOthn3X//6 137FvicNkA61ia6x0+U+31CtzqNPnjxp4Kiqq6a6oHgpuPYTHo0acugSx0ga pRTDCnnlaDa/aZ1yjkBeVMI3BKUHdYopy/PrTYDj/iPJo03vQHR03hKjwL+G EjL8ut4ctbqEyTVNYhYIhG/2Q+f8M+dUWcKnRSzL/av9XB5UAPbmm2+OgI2A QZlhC9e50ACsWA2a7i5eWfMYo6Iserf7d3MCME1j0KSdd3GRSEecAaztAgBb XarfW6w/ZdCDpO3dDWmUylGYMsBgdSLc5V/9wsSzlQQ1Q6OmABisNXzJfChX EdxSeoLVAGx22a79K1oHXcCl3e7EHcs7QpLOU7RLH8rQ/fzzz5fLMtyVoDSx UVeubSzbsn+Vw3kpfhqDPkOMMxDXIpYlAFvEsty/WtyyRfjExJirVXL+/Pl8 EkALxnf+2XtcZdI6ewQJWM6DFmVnLp3xekhgnM4UXYGtL/ZmYQ0KqtNJFoPO erLI1suTi9zJnCxmmxanRTQq4lC+QBa8+ZYcOHCgzIZMahYzTxKvm7uKroMR c5eldyElrbWJBZxoe+Ghhx56+OGHVcXetVmhqs5rZsCDyZAUv/vd7+DAErqN fQZsakVbZbgbyekSXZzGJkvtATwylzCPVXNBA6oHneABGQ6DPo75hqDK8oNm Nrf4ojorf7BkhdOunulQoiCoNYNarFfIsPPWfGZyCRthDVhS8td6CbuT/zia HIKilp45c+YMJPDY3dLeBYSgXxBPYYZb0VCJoVHAFd2DaQv2zuwWPD9XN12J GDJkhed0vKlyZAPwetBIUWdx7Ec/6lMYhu4VUQowC72MhIOT2QrQCTPDTqBv lQwXI9+0hQW4kn9eT3krDVKrIX0AJ4r4kaqrwXCmGBALaRvgQc0Z80MjqI0s r5AM7p588slySj60UspdSFsoKvYNEDymIcH+kX9QVQjTgx4wnA3DRnGKqhJr kBJ8G2jFA+myzKhhUiXiMUes1CmBt7EGU+liw8qir5XCbAe/vBaLylXezO4v GFN6Ngg95yG1VuiYPip4C7W0UjC4NgdWl/BpEcty/2o/lwcbgOXRZHJ77733 UGiWHokqFhSAsREM3N/8zd+QT/tgYMDx48evX7/eptB6r98TgCGS6S/2Izsy 9nZ2H4DlJ0N6pTgkQBZ2tvVfLSiVEBQriUcMljUPqMP4mkEsv5sPgIVMwoEX L16k+ff0CVyzbH3/KlIBHsMNyCRzYwd38+/wTAIwPUVQuZ13rJbT3ZyZPXeu bMv+VWrfpchutuLu22+/vV+bcjtUlgBsEcty/2pxy1bgk8kH7GFGGTgGxcg1 QXWyNo9B8boZjN03PQI8zz33nGmNBS8NSkcqIwHNVFG/uUJbhc0sMBEaMMJq m2ktLZlOloUN/eijj/wERBXUkb0DAEZaZL8CEoUdqIZOQJAEQTFtyAPJCn3D dPoJbcwWxr2uibNnz6oW2EO8Gn6yUmrOhM8SgTfqRxWYwcxBLKdPn9Yu+otm YKK7Z0QvLRaSQuUJSs0q0XGffPIJU168L+iOMdWi74sAQIA4NZ2SqteLA1bc gBL0FF0BRFGbSqAgNfiSbtTFv//972/cuIFTsK1I2iN5DWngi6DYXx2hHhWi oQAXCGuLaU6387C0SsiNpQuN5zxz4cKFUrEUSzbxkkB7IHpEt5Y6/PPPP9cj GEEbW4kSJHnd9570Exa8XuqTUmmDi8FCr1MYnOr3DjTV0H6Lt7Dv4bIclmiv aN76cZ7Qo5UcxUF99dAo8ElDhI9CWoHfQjahsMhaxUgvCIbHkNe4oDnIRmd+ d3ikqDdv3kS8EUQbC6MxYn/5EvFlTuyGhfFepun/eLfox3Iqkbne9C/G1UDm HtObWlwP5o10z4UZqSOuXbum9VywwFcyLFGUjlNzZ7uNWaTqdGQTsuGmp3Ks MsCpVr2p+Am81+nlZCy2VYhIbxYFpXsfuNMvZTCE8wlKQ1D0qVOnVJ62q1mL zTxFCykBuhUNORs7a+5EtZRoAiETlRe3rVhkuc8B3nOeLaZ4uqYdPzrvdf24 0cA13y3h02KW5f7Vfi4PNgArkyyrwRx0U7tdlChfRABWYusgAZJKD1RmkPUS 5cwJwIoDFj1FbWrbavcBGKtN1NRJPZSBtNOoeWro8DEAVkywPK7bawJ7Ot9Z 893ueM4GYKUp1BfffPNNTtfz78tNlq3vX3mFYpMPYEN1VUVuM7ibLIHAKQDW eShBdfGho7RNbM1tb9ni/tVfViKEwGwD3uQz/+BtXn23BGCLWZb7V4tbtgKf WLo7d+4wKJ2tWMQxxPNMuSMFjMmwrCjFp1IJY2dpyQKuTqE7eRe+4EIsvomR 7S4UZ746KmF8mRVrcGYuIlv/aoUd9KV2fakVFr/kcUxkwTxzLPevL025pm6t dKMNQrASh5F+v1JAF9aH7cuxGZwo/U1OXyZq0zVKcGfGhpfYKXXmG6PCEtPM FlRu+TnVAzDtmWgCI3liB8CIAsZACRMz/JSs03VuaBYL+e625eKZ4v8Qiy5D TBfcwCdIjxAKvEAlmGN0loLNTxDFcFD3WBl5VDL2RtSjQ0f6ZtBufnX67q5r PXGN6KPEzijgEYPAMEhWvKCcqfJI11DxHAjKY1Qix5iUygN6mShI8o9//CPw o1sxmOeMjtOtMDw0QgJ5RuEUICl8hN5ETMpw9erV8sSphHg941dtdVK2IaQx CZ8wBYwZfeRM3/zbxYQRsQ0lbda1gab1QotAp0eOHHn44YdRgkfSLoQCzWxB 0eF1sbZK552rkubgTP1LJXRWOvOLX/wCU6rtqojezHuNEMhKd3Qhcf5YTwVU R/aPVooa9BR4Rlu0axzphZIP5lhVhKvWPsaLJ+ktpgZoVxCAfgwSVF8WiKwa CK1dOE20IsALGkrkZEh6mKr8caWQvyZ0K9n6qbFvFCNsrG4KtLWeR9ZfVjIg mE8oDGkXkkVfIIBKEHLoep75kH5iCvukkQeFYbtRZ4wlfFrEsty/2s/lwQZg jFF+O5NRrYoVAGwsIgDTUznzgz0ACTuyepodHlNdtJ8TgBW7ftLVqivwewGA 3XOP6C93Y4KxhgEwbE4CMLq9pgfX4Ne7iJ8TgG390tlW9q/aygMXATDCP3jw IP3XQXi/57sjJtgAYISMR92KZUMPgt1KJPltL1vcv8qXvjFO4U0OINyGwiws UFkCsEUsy/2rxS2bg08eK9bByBDHXHbXaTZ2Gu7usEd+Jpq2cDY9AhgW5mBJ UaynXuwcCubJhQaQMB8WBhCeMa9aooYEmOzSqZg9AABTioWkhSojVUzLYjv8 9G6BKNigDnfUzCQVeaCMYyZb9tR01OHdoUOHiKsXC+mjdZSAcFadWDDVqwEZ asBLWyKdPhQaFJ1gzz0T05BwATnhh5bY5YYruclUSIpul7eUhmYJM9cdzeXZ MvKvkRsCQKlPP/20fTxmtMPH0tp2j8zzoBHYUPa3kteYmbVelp8c+Im6Skjb 6936x90m4FMO9uT29ttv06inn36aPOlD1w0CvZogEFItY536yyKUlxFO84Fp E+PFF1/0gAo9P0JS5NaeoLrB1+WCYnR4q4x+KqTYfi13Hi7AV5R0QaDYSuRG qbROLJ73ZPEc5uR3Ej5Ba6VJCusWsqnk4Gk73N5eTSlpSrqEVFwHCZA6MhgW catIYgR47do1cHQEaypZTzyqwbuq7c7CCEnRCigey3hOsXV9sZ6OHj2KHl1w T+f2FjvGEVUhJX2qLV0GzCMVENUL6sEmnTFwgJzczwrhng4rbUUaPnpK01TU ZKVfvJuDFjb1pi4gyXzGOoVH7S9Winfps+/92qZWdxXJHI8YV8+IuKLride7 KqGKWhzx5ynh1PxW7FyiaCFpoujuTIlBC9JCqvdMeF2WSUIAaJMzjohuQ1B2 CZ8WsSz3r/ZzeeABWNPsuNNXGsE8iBYRgBW+4MaNG8x3DrqTbiQjRIDWgbQy BXu3KX02AOvyoxfRr4mRjnlkD7yPAGy2UhXpPeQ5HLYRwG6qcDIm2ORbgTF9 VARICjBSKt8TgM3J1GyaN71/5V16RSc77UIq3u9ZQztXGNFxdMCLYTDKr0PL mViCmz0VEmor+1f4pcw6t/NibJo9yl563/3KdqIsAdgiluX+1eKWzcEnpodl NzX9cKVAQUy8ifeeKVfaZjGJsa1ehCi+//3vs7alRFkdpXAczTAZsJNVpBdZ Ou9a0hYJhyEAEopRacYYicm6QR/ogiUCMJg9sFIeeughWMg3LIjXoQU1ABW5 M3WVzAIZj2hj4BBp+fzwSikIz2OPPeZ1AMOsrnLQqzV1cZmiPzDTTSXTWgnU Rgyu9Ur7UcCM3mHEX3755bKwgU94ByqmAgiMg1Ht6kTEaOjxxx8v1INScKRc U2Anz3Tbbpz0aQ6/3eX3DJkkpYLDe531AcMIqrhMo4bVZgizm4BPnXJevHgx m/7ss88S+5ogM05JQNfgdEQdj9Ro9rpOAeSKiTTCRnUQnG7kX1R0d5qQSoxo 6qAj7ETfuoC5Js30rRv9ekc3wckwzJz8zgOf2morzAXAgyNwHXmo9VYsGz6+ pF2IIZOWDCO7X8d85aUCP4pePt7tDJd2AcM6dOS4nOzTtg1LOF7CxHC4AeUV HXRPX6zioiA+Fy9k5BXZK+meEVpUN53eDlL5oYp7VpaohsDRo0d1uieL81bI VgXsBxh0KFFQCXhpjPSKekrlQ1Bd7huRsipt9zW90J+SQPnbzQUtejEcPhVy bagl/OkZcLTtX6/TItSmGPfMcujXztm1jnFgD9Al/PnR7BI+LWJZ7l/t57If AFjhnsxsjIj5f7xbdKzFAmDwj25S55Rsh6DGngxBIYy5LDyp103Ob7/9Nms+ wh2sBmClRxwbQeVGxCze9wgAm+Ia8ez7b3/72xMnTky6tUMszFm3FydfGS58 RfEqvBVqNZpfmT4tMw6F6Rb/7J3GzZXN7V+1ldclR0ClWFV0j0bNfquhp1GC 0oMGbH715AxDdvFwb27pbHr/qkFkVMJsJeg09ErluaME38eyBGCLWJb7V4tb NgqfinvAQLMvHd6Z08zeTNiM6deUBcOAN8x3AQRYuoJnFuWmdehkvPHwg0nA pOctDzPKRZgpwHVXhJhIy8aytHje7IqLHMvZBWaxwy9IoLgK7EVe7qxky3AP f/DBB6Zo5p7NVbO3SsHz45XiQ8GIvMu8qophLRIRyrXe+VdhhcKQgUAsdPaX HxQTcO7cudUJoydLkRaYM08WkgLNmUgcqa1LSZMmgH0pFBIuSh5UouquA2Cz HMoHDx7EyIiLlRdT0VYhNJ2olYIRJat2wBh3xIOd1u/JuZ6a4VK+CfikuwlT h2KZmdMcrdDWemG0PQ/VoIeUXnrpJfyiE9KDnfBboiJdnI6dOnUq3S5OJq4B SGhWW54vSXFAovRGuD58+DA18BZru/oaxSgdwzHQxR554YUX5jnYHe/eEz6N Wx6YpWDlG6J4R44coYfo7NqIDqrXGhrQQsFyIX/62ZVJneJ7bHbVrssLVi5F 7FcJcELH6CpBaWtwHcZGLcXTKdRAPV39U4rKVWai2bkgwTbC6faEtxoI390F hK10CIF2aQinkDxKtIikLk3Us3R4KGoHtX7Vy3Q4BwAD+eOVYvjTalKy3snF Lil1wkttvEVFCcok061JJPmrT8mtCGBkkodDx/0++Fcr5FnE+MlR0OKl/FnI QE+kEhfxWhqYrGaHfQB4CpzbdKeYCkq1eU+N+m4JnxazLPev9nPZDwDMK/lK vfjii6hVm3nVFOqx69evm+IWCIDhcQr8qI2RJVsPwFe5SSPS30J44atNM73m yUKBBUVWA7CimzI0GTXohd0kZFBq7wCwZEuSui8TmRdfCfhoxeqYYAEJXw4A Rg9HkE+KcezYMQS3C0pQhbfq8HEnNnY2t3/Vfiz1wyzd0E06fSpJ9GSZ3O8C G/TyCMpqFFD7ERNsT/lcTZbN7V9hB4LCYJCPhOEfyr9d7nN7sywB2CKW5f7V 4paNwqfcyI3NFnRFwF4vaOHIRWLVZultBsuyP/roo5bMzWlFmOn5ETQgM8GK sRQMYlEcO3HzoYs5LB0b57F/XSksiM+FjGYHCzWAwpHZzZo9KPX9lYIAs3H3 j2AhDUELbGgUduZVi1BTF5H89Q001bbAjICf8U4s7DtS4xciggrWe6XNisId sHRgz5MrpdNVvaPRyd4Z533/9E//VCgk2AmQGLGhrNmLyKqGJ5544tChQyjH I2bbCYGp0Fb86qefftpKv1hJXgT2ip508OBBv5JnR5bzmJ6Nwqc0BFXF4i7C NpC23l28kbRl7F/pL5RjAYPdSlDJ2KUhExVevXr1q6++YlKRR1AQeLGhRnAn 36jHv/AbCYBwJS/OS2e9dDmIoW8hMc3RVWhkHrO1OfhUyFAjrs26NnBoJoSQ kjeW9Z2aAWNagQWj9dKlS3qkeBQNosLgl7SoEBMvv/wyLErTvvzyy3GMPtCj v7kyrk6U09XFGbGegnAoh/e02D0+AH62G7mf6LYxXtAz/TVYVsnIIYgpPyGJ Juvo9qPARRCCEEo7pX/VUBaqwm31lhmg2P7Uo0hxbQ6Xyoq2FN0Fwb2Ia6LG cmmyCartysmpoFPp9gwJuTAgXcdAW3FZZ+Ry8m7OosjTNWfPnjVUETPP0FvC p0Usy/2r/Vz2DwAzhXqROT59+rSFLWbVzwh21W6xANhwPcL4f/tv/81PJInl NYNzssLqLJZFnmzgDWg0A4AVDzyPMlKa8mKaXe4XAPvJT37CMnZfvv2cdm96 vlheOh15FA+DkwBMJW18bSgVzlbK5gCYwaI7AEhdTCGhxBs3bswAh3nWFdSL cDpgNRAgRqiD5s+Z1fE+lk3sX+V5hWVYUS/r4i59zB9eY0HLEoAtYlnuXy1u mR8+FYbI9AVmACGZ9Zs3b7Jiax6RlKLL1F2cos6VOpBiK73I7ueNEGoy2M3n zAHrCQl0AIewAlS2zERkqWcUlAMGxUFCUpGIOrBjWdhHwAC0M5NYyEMCPjCR 6gkgBY08nGNGZ15WwZOBFIAHa1gs+FuYU+BEQ8U9mCFVvxIUIFesrW64Q1xr mmaiw7sFL9NfDsFwqebQDBCScAeL362ABw+rSuUEglo2Jdl2/sW2ohxO8yIp keeHH36Ia2LENesJDRYOqyNCyIGgWPDCapl18atFQMK/vrd4twAnk6KD3lOj NgqfdKimccGso0SjI+jrao2Ciwr6+vrrr7eVASQTFDxcRNYiu9KurgwUGJau 1sX++kxcvgQn9LLmKBJBdYWtq3bUW7Ul8VG5fpmxK5UnGKUdwRCuXLlyz7Q4 G4VP4V6QoOBOeNeinkK51lFeJiYy76y87anUW0eDxN00mfSzGpu3WMbmZLof jOtxikQgg/f2S4us5S29VkwSpU1O5BWTbc3ZgEwMKH3n+cOHDxuMyJiRsvy7 FSDkLeTprBdWSjtyyG4D1hCYippVYqZSDRqtpT6kXRgkMQuTcyvlvffeIyiv DN+qEgAV79diSuVGfbrRHuDVq1eNYl1WYsfCiOluk5shRv7h9vS2JIzeLdm6 cVrsMgpJdVW1XmyrsWtXTPjiaBXg954hT5fwaRHLcv9qP5f9BsBMa75XIQtV cKcRyWohAFgh2bFPFASVYRoJkb0+ABgjxVAivluQbEQdgcf8jmplTQAGfaG2 i5Ds6ZxHh5X7BcD0Eaiga8bl+qL0dwN02M2iqmpOL/u+jUGPjfBWu5OWbisA jN6SMCyhhvV224KFOn0kcTY6dApRF+qq5Il73xlpo/tXut48ZugV8rT0mt7a a3G9dqIsAdgiluX+1eKW+eFTOycMYn7OJmS2e81jmpG0pTgA5j2zvUFtRdmd HaaZaRvBiIIE7clYOcJmzPqjK4VJ7f5awUULoexhhhK1Fy9eBCFMqgziE088 gZGDBw9q6Nlnnw1+aAtfqgUnfDCrtCHD8n7ve987cOBAsYAeeeSR8tqEc1jn rhQVEvO7lRg+XveTx8CP2ZZuwCH1PP744+UgBo1W2+VwY/GfyyL9zDPPjHtS oaAM+jgYzVkdbe056Ih8bzREyERNgOyjfikKa7cGyAHXud8T0YgTxb6UdDuv FavmySkXqECVOr3iLxZK2TPb5s4Pn+p98IYieeWxxx579913Ebx6T6PjTlAB 5IavQIJ6kLj0GhQET8IDIwxFcVDJny7Rn0mulWJGgdZXrlzxYo5GI0ABzAyC 0ihSzadLv+tNwilY6xT7cUE44Bmqys094/iyslH4pF8QQEMQX4Lyq1evjr4I XBU9Q+vk6RnD59ChQyMsmAFimBMdqImdYozQEH2qC4yjAjd5S0d4mASMO4pR qPZJLyON6qZbt26Bu+TTbU2i1n3FBlkThXorxz9jk7qWNQlTs/HqmvCpsdBd Ffqmo+mwuaVrhvqaVIvbZig1U6EN43rcmLJGK5JwW175kpEVPSGiAHnn5hS4 RUQJOrVV8ArjLr81q0iq2+qDyo0A+I3Zdu1MHd4iWw3RKE0YvHqKkqy3NvFi 4WLyGUO5yu8Jd5fwaRHLcv9qP5f9BsDMw+h57rnnyqlhyvV8TjgLAcCQxGbl 1kVQegFJw/07izkFwLpI9d577+GOXVY5W1NQLP1IGmRCknDaFADbXNkLAKyA AIHPIiBV/1+vFJ8hh+KLqnBDu3PbVbYXgFXyNxsxweiwmktAYygRlNd9mQP2 zrO4PWVD+1e5nLVjCRrh2rxhEO0TKLIEYItYlvtXi1vmhE++Z+IL3dmmB0yy OiWuzyZnU/eFCxfY6wIXlDVvXNKxxMtBnTlQg+9VC5Oos7AML730EuPbVSDm z2KTxfzwww99o06msExqXZzXhMo9DzmYY99++21/gTST7Qi88KOV4mGUhJE8 oJ4XX3zRVANEtYcz0sGwZd2fCuB9txH41OYJsjWEAE0cP34cp2qYstElv8Yd govzGerAY+AtCFSgy88//7ysJZDDSJ4IBHZyB7Egj5nQKYCQydOTfurAiwQm 8wSh3wK8vSy8szJE16HbVDKdMgdZ8quq63Uk4DFfztCo+eFT+2OezKFad9PG qa2PIBZB5QxWsj+86D5s4lclf/7zn9t68kGLeV4VGwRkaseArKDNIlrgve+J RetnzpxRTydiHUPDVOrBODHqwWTodezrrymoE2jRtC7zPHmSlWHCiM9I9Dw/ fCr3ov4losYIwU71wkh7RIdHTLCiu3cVFJvF5SgEXAPHGEFAkW+NL0IAy2lR yY+8krQL2F48VUqYMv/LSun2R3dMPJNOEnJH56tjPelc4i16MMIQqRM9OcPQ rwmfRtSsUrGPpI2XLl0yS+jZAwcOjCgiVCXfgPyymhBKDu57dWLNEDADmD2M r47Uu66ieLFoFXhECUHp2TEp5frVwx3BJ9KCj+nlMnuOWMT5TpTfUIf6t2yM U+4EuCsxInrUSaq6Tw2zw2ct4dMiluX+1X4u+xCA+bWA7So0Y2PczMaOMOV7 H4CZvUfmxPIelmRWd6iKaVsTgJUcWSvFDWOj4/r27dtYhjeK9TQFwDZX9gIA wyNBYRnvnh9hOdum09cFTLuPOznbC8AqpWxm5XFK/8cFB3ig0To8tPe+29Uo 8+9fYaoYHVgGMumVMQ6R7uXoXttblgBsEcty/2pxyzzwqaWiiSh3hcOHD5vH GKCpY5pOstjizz77zHRn0ffoo4+WdcW/KociRmIak7+mQYLCLBdZumg28Inn c3E34UNcRfj0JAtS3OmDBw8eOnTIK4jxPOMI25w+ffrMmTOMSwgKZvjhD3+o zrLVPP744zlX5KBuhoG1CnZUmIJiRa4ppfnhk58ICiorwI62TP6TNnoyJAUG YYYW8kWWKLnzADNlf2MyACTcddepaBJPPfWUF1leWIhsxw04/+Kr3NbqLEwQ xAvSkAlc4ZUCCukaRBIgSKz+ssLlXzRpXlVenj4CJ2cgTT9OPTNZ5oFPnVKx dMW4xo4uK4rs1DMgAeFrFAZAA13FuH4ESnMYG2EowCEoC4DEI/bx1fGuXugu gxqKsAFQlV2uo0BMleiZphX+SH91h4vWUbk8uIoQpb/IWQdNLRzytLEWCNx6 RUesF5bku43Ap0KR6AKMtEFEu6Z2xgKZpKcT8TKCSxTdV3dTHm0Ri29wXQyo oRUkQ+tu3brl9ULR5vxDUJ7R7ySJ/XKsFys4/WwsG+9Fz9BoSmWonj17FoOr Yz35DLTDeJYzlFMrnjTM10M4c8KntKVcjd0ObhqhYPShxRTC8D6iZhl0Vjcl NG/qMCfQkA6X6Q8pHThwAJHYb/9zRG5BRhdDirJVJLG/WilESs6YMmMUmjVB lR4dL7pD06ptWxXB6lmdyTGHQC1q1/M6ogpnXJxZwqdFLMv9q/1c9icAK9vg 9evXuzg2I054Ze8AMPxiPIcxEvY8QXXHHC+ffvrpDAAGFGEB44U336HYR3sH gBXRKwCmxxPUPUMr7E7ZHQBGtaBrjLfFt1vMbWfZHACjUY3x2YfdD1hZArBF LMv9q8Ut88Anc29GmUkCP0xQDN84pjFxmcxN3WowdXdAk9cHAwr2FKWTvWbd SqNsSjcZFpCBOQBIWENWz6JPDQylfwtoo7kOsMoCo04Pv7RSCg6jhqdXSkGS xwGiRR+4ZQpVYZm/TK2WnGU2sUhHcMdP2g1dFO1nzS2seeCTFz1GnshGajhN cxoaa1K2AFrzDB4JobQ4qC1aODSFMOQxggV8gC6YQlX1ZL4ihfHBlw86xTdI 6vJaR6JFScU+gvFeLCyUQLxa1xFa+eKLL0ywKsnNppQojBSBTDmQl2hGP6LT k1rR+gxH93ngUx745JM8MYLfSWBDYm2kAI3FZdKzPuRshvjOkXVod8HanCkk 0RApOuFzYF5Ha9Hz6iQEb2Vn9cI4JB1hGUKwBXNQqC5BXbhwochpnWijvMPK QXOoWP1e1CmBW9RiYc3gJPPnH0xQzz//PIxNn4kUI6PCrn7gkbbgJbciY6o8 3YjRStpOAoSGdw9TtrQr5SmgVqFOdDEFVpUxqAfL6GeMFGY28WrCi8g2KEiy 9E/V3EZZsZ5IgBpoi6AsBwbNLQrK1lR+AZ+p6Jo3LueBT2qm1bSi66KeJDGE leyp1I1FCSuBo0HRJGPZ1eWXkrN3c0QpHguZY4F8VJjXoseatXww//iempUl ylACWkiMoHxPSijJG01bWCiOSpMPpaXGZKiSxmyxVrqFMRkKPr4sGz2TOxk5 jyhbU2UJnxaxLPev9nPZnwCs0FuFJZ9nN2PvALCOPgNgJFaORTWUK2ceAIbx ketkJ/xw9iYA0+N6J5+6veB9tDsAjM7n+7f347SvVzYHwPR+9393J5rZHilL ALaIZbl/tbhlNnwqkAJTNZJwseZs1pjAzU6d5bHORbN8/PHHn3jiiSYxxkv9 BnU+7Yx4y9sCiR86dEjTxRQ13jP0UIR6gAQmIL+Xovf0ZBEymWbYQ1uspG/K oVYghZEvxjPaUmeoqYzDU7yjHNmXLl1SA9tUhFKPrZ5y54FPDBlTheViZWMf I9nrIUkTGmJUhXJWPjeqAom3CVASn9LkkTlG8pbpSdaBWbQA18pIUd2RKPT4 yCOPPProox4zhXq3GDuaK0XaamoRA3uMA7iSu/lX65OxnupoYiwqlCfZJkY8 WLgaisyGT+3z0DTmnkYdPHjQkly/a3H8msN24TX82pYCvtRZIKBujSEAqWzr 8BQq5yDt8hbsxIKsGR4kpQ3MlzM6j51xqBpIA2KLUATIYURv0kmd60kM0hZd oMuIcexQtTPT3f+WBjDMmgeO88Cn+shyI1Ss+Dxc7hNUfmJ0OOxEApSBYIkX g6sDWfQWjgzGIn4UlAPj6U9JDVrRkG0RZXUEIj2mFzpP72IdvVVJDmlIbZhQ PPKpy8iBGhAUeDOJXUd65XFUd/nyZZWsGTN2Nnzq4C+fLhIwXii/LjCOVve7 2oirMYgp80yhOUr2l2sWIeCOqhtupGqq8bfNW0z5tdhxpgKiJr2QakOSNAxJ pBYUJd+tYib7Fcqto0nDzNMlC3WSgJWgqoxHk21dPCZY/xI+/PzsSjFYqOWa i74lfFrEsty/2s9lPwOw+cveAWDlMlYDQREa7KFHtJizNznMCcB2rux9ALYX yv0CYAtXthGAPfBlCcAWsSz3rxa3zIZPpl/rSuusw4cPs1DmZHNU63FPMgHs l7WbaY0RN73DPJa9LBfQ4id1snesA7TTmaAC8BSUxixnzs93qNA0DF/nXAVt 6BXfs57gQUtL8My/OYoUdQF5pgtW+9q1a+x1Z0PrBVKYLCZeP7E+5W7GI/rx C+atlsMM+ORhMvnyyy9b5+IFxiPY7tkVdhVJ586dIwqUJwTPEE5e+vADqABE +ZLtG+4f5aZpBd3t+w5DK57xcBfETp48yYaW1yw3D+Yj0Limc/4grOzPYycn zxl1qhBVKgl+FDceC8/dLRcuXFjT6M+GTzqFeCkJtIN4Aim6Y7ei4HBiJIoi bkECRV7SblIq66KOKOhQ6uEZ/6I/BK6Goiuodr1jr25kkE/3I9BJsak3SeK9 9ED6sZ2HcoVfv379s88+K91S/jBF83j33XdJ76uvvvo/d0sXLrzYXQyvq39q o282fKp3jKACeZEVwnLXHxF3sYkYetsReSEpDD2MFJJivSDqIXndWlwmsrL2 wRoGc0xCFaBeMqlGZXFO8ILxFNIQJqi8jMqRp0/L56ieGzdu+IuvbpGopH0e nUg4gXkaRRmMXBUm8IKwTUH92fAJL0TX+q7u8FjR6dc71u+yA/l7kfITqUZL cYVmbeGCVEsUpUJLoRZobe6Zf6xWCuvxi7slQVVGkL1jK6VVXtG3POZz2L6U hRo14WgUa14sgpYuMDMYbvqowVu0fPQYEUUtM8ZX53lcwqdFLMv9q/1c9jMA m7/sHQDmXVIaU307dewFmZOAeuYEYDtX9g4AWx09YO+U+wXAFq5sOwB7gMsS gC1iWe5fLW6ZAZ/MvaZrVonhtnAzRZvPzVS5MQAMVliwkxkeuAJpQAvPFJoy Z/VLly6BBMU8PHjw4Pe+973S2ZjrmAPm1btlUgaKjPqyhnkmtMYu5Afudd94 /dChQ0AUKMVYFNXBWg/AYzSzmJtwTsZIGYHxaK3KNF+5cmXqevts+FSiamv2 ECDCGMThG5+g4ElcA4r5+cCZgJ95vmzU5rrusjGUfi3qDgmAT0R0YqVgtiPR vEQKRKDCcvKCYQRO7JtwSkfhn/70J7aY/R1HSKw2magzS5RgLfA9U/yo9VLV rAefggQq1F9sHB799bBvRtIWFWoCJAgzj6Qtvszt/+rVq6WN8wAh6LISR3qF st2+fZsx3RDvFR3qReqNZb0AAGudHEg4WFKArBHXXd9RlWKGdE3Dr2V/HruC 6AznM/qUZ0qjZsOnDkAJp6DoxTkPEBaSAs6k8/nO5XSH5nznNL05CaiTNuZA BbIWMKqD9Q5VDVWMayVdxddwxzJ4W85YESAMfjMey3iO+JGRquP4sp+HA0uY SJj0WbsGEclMKvAM+JRGleCpnIYeIMzNSYBs9YL+LYaJ7jty5Aj5G5LDNcvM U9y5pqbGYzFeyvNoliuqFYIbtujpbm+OW2oo4hydKZ+XFilPw6ocXmYY+lw0 kpY/9AptBebSoie9a+hNweMlfFrEsty/2s9lCcDmKXsHgCUHsqrXSu1R6Ing 0A7dCpy/LCIA2/2y1wDYni27BsAegLIEYItYlvtXi1vWg0+Gnvn5zJkzpm6L Jjbr448/9qV5O09R1j8rxsqz+51AFbrTWoyhL3NZ0QDKBVOcTAjBcjiMZOnX wZ/535eAAZTFsAa0/OpdcKVACoAZtGZagDoKVw6WbDSQwupSyh6T8KlTp6wr UYXaTpGGNNaDT4We8G5BUAvxDSmhjY0r13AAsrMq8AkYICVV4ejVV18taEAR dbo4r5S1R1tDUL4phwuwRM4XL15kMoCTNgoY1nzUZ8R2ni2BHKFVdfPmzSK1 juPXTt8+//xzUoJhtIgjvaZDEc/Ej8O7ynrwqTsOGCelAqpfuXLFlxbpBAUX HT9+vMQlxaGiUcXiKMXbSOXj1wLM5u2jCQCMJqhKv2xaB7qoRZI55ICjlA21 ZaPDb0fSmgaDARW9UPiLrh+iynBgtgqcNY7nvF5ynwsXLvhmtDgbPhUzv2to 5E84eFSnVxBGUUsmSBpl34ZIjUpC2HRCn5FJvHsBhKBFZBgILYK6FkcU3T15 cqUYoV21azPHh4azf1GIO7QVLaQkgHm1mTroBoJJu7YKsZ5HE6jc9dUImwGf 2ic0Imoabd2D2NwoyHfOsC2P0oiaRQiU3/AvuWEZuBBT5C7Y1ZquG4WBeQo8 /NbMdd9fKdYCOeyNAd6oLxO3saDapoUirrRF1jUEvd8tDBLDYNdVypNF0ya3 sJbwaRHLcv9qP5clAJun7B0AZtYFS1SrlTzNCCE3sz3ib7OIAGz3y14DYHu2 7BoAewDKEoAtYlnuXy1uWRM++Wv5VlRGCzQIp/SvualYYLJrvmTW/QoRjatD 58+fLx74I488YlkHBT3xxBO5lU5exnlopXjA6ypRA6NZvhLEFDemlaA5k/kw G3zwwQcFQC5D3LbLgTnDI4uvaUR2oDZ8ZtaDT/nwg3N4JBAA6Y033iA6zxMU i+YVGAlTOXeZ4TECBBJgAYVwbeVLFFa4eT2NyKiwU+nSiKVwQ17x4s2bN9W8 Q1AhB3K2u84qhQqmGOgicBK+z8WdRjMGWe3JiXpN+ARnsmtE6nvMkoMm/ErC qqVa9IcRBK2BZ8yO7MweA9QZR4rUHS7tvvLKK/nqMKY75KBODuWMzoEcvNcL +qI+KrUKbUe2f/1U3r0ckBop1IAkYT+dqDcNpbIDdDi7HnxqLxHj1IYYVXvi xAlPdpGTxLprVnPEVe5IpG77wXfnyHrcWsCQ1xFIohUdy7LRI05UEc6L5et7 QxvLHtNNZgaCAv+AHM/oRL3fpg0ewWl4D1+E5knzAEkWGH/Ee18PPvm1+LFI Ig0aRRR+3XZlaGcvfwbLxkuXLun97uNYCKQVpQQtp6EvKQmSqKuhrX/95DHP GzW0peRE5fDyQOsFywSMk8OFCxeM9Jy+vNXsp2kKSQe+/vrrI0eOeL0A8jSN 9AbLS/i0iGW5f7WfyxKAzVP2DgAbQbkLELrLcpinPBgAbKfL3gdge6TcdwC2 QGUJwBaxLPevFreshk/N0hcvXiwGNWjE1nuMtbJwY9dYt4IDlIzGxOWbVm2F ylFbZ3wlWc7X+qmVwgiygP9hpYANbQgwEwyi2c8ijrkshA5iLP/Nh8xugRQC Tjvknh0Q0iJ+EdNBHsBmwdgqfjV8yr0fzWQCAZKD5S0p+WsRarYnopHgBi7q SPRHK2VEyyHkLg2RFWkEGjWRf4VpULvq7LgT0iCKAkrskBxywik5I0YQgDss FOOIodc7n376KQ1BHlK7NXb16tVx+LgaPhU7ixaphxBI4+zZswyft9QPJ5BS kX8K56VOCuZL9RcF3TOaJlWioIrF1ujMd+fkgCP93sZFIcIoZ/swSP3R3aJP f71SEF//ZsotJbDvLcuQI0eO6Nljx475NwCwHnz6/e9/D2DTQMOHDC9dugRK Acwk1sKETNRfAkS4vTsLO7egSA6oJXBjgSgMips3b+qLQgGjSt9N7eQUuZ16 j5mhvDy6nvQaCOUxpEUYwSBZYarrKiTpG4ixmzKr4VPrOO+qsHNzukRjd04O M6JmGQ6kQTdMHfq9c+rCxVgk6t/um1CAnByK9NsVFZ+Hv9ZkBC1yaCVVgBF1 UglLpwKLXblypbUGsWh3BHNewqdFLMv9q/1clgBsnrJ3ANi3334Lz+wc8Nh6 eTAA2E6XRQFg973sEQC2EGUJwBaxLPevFrdMwSeTlXkptxPWPE8kyyWL1suX L5u3PWnlVdgfs1nRiRl9oKgTPRgpv/Qf/OAHXaspqMLjK6VDHMu31rn+zZHg nXfeKU+ZpXEg4b7YRO0yXohBG07zjgaTfD8Fn8KZX3zxBVyEBfDJ9M72kWHR 4D0Zy0TB+jOIpAcM5J/foV6Ys+AVCcRbZW+BWplOaKHDxN0XhQLzmIQZd0gG HqYnBUOAD/0E6AZyEAwaDQAwBZ/0I6hDzVhADLJ0PniGHQwnkADNIQE/FZs6 B5WuM5QxvBRvKpm6q7g7pWv+WC6ndht6HekqPmNWlx05coQoinHkAx3QfX+/ UryCF9Ir6lGZvqfgU6EV6L8Xf/GLX5Tn0bjzOmzgycYUtaGBhO+n+3Jhgfy7 y4awQp23m9e49lfH6dNxJ4VA9K/PviElTBXMHIMdBBeHBPgpH5+h5OHicmBw NXzSHcajD+RALcnKTx4eOft2sxCFrjQP0A3w/tSpU5TBhEZvqQEec2agLV1F 6YZO+1pFQi5bVm4PrTcNn8K8K7lYeNcH8jEnGFMGArl5t5VFVxvKRrGETwtX lvtX+7ksAdg8ZfX+1X4AYJsrDx4A24myev9qUQDYLpc9DsD2VFkCsEUsy/2r xS2T8MkSzLIUdiowi7UkI5iLhW+K6JJdy8/cX2CJxW+ZxhCwg4/cLX0DPpVV ueszhcQxH8IGmmMCQI4//OEPeRbBCeWMu19GQbvFcC6RDa7xnqMFaDfgE+CE 2vfffx877LvFOJPHxAMDRUMq072f/nqltHQdy/mHV0qe2KVaZkAvXryoCU1r 6+uvvy6eQCb1fkmj9MEogRL1FyILuFHanddff92CmpSYsOeff/6tt94yYyMV zhnwCfwjT+v6LsjTASIqZmzXzdIoNbTYBw/yNlF5FyIoyYiu0FWp+2UiSaMA WXqHQOAcXYaXTuXawOnOF3aGd1ZHdW+uFELAIEF5Vw1T8Mk3ZS0cAVG9Qg8L OVVICjLpLJj6jRt2u1w6b2Wp6UahM3QQpUVYsd/JBKnjpmELqEZBPgA+F/K0 dRbuKEwxz60aOvz1VyVmJwI/e/bsgE94165WiN0cZeiB4kRHN+6LNMqrlTR0 IoHoVhMakkxuOhR5p06dSueLLtLNwRaVPnQTs10+hbiMBaLLU+vRlUJQTSlk ZehZXr29UtpYbvZoA3kJnxauLPev9nNZArB5ytT+1T4BYJsrk/tXDxIA296y ev9qUQDYLpep/au9BsD2VFkCsEUsy/2rxS0DPrHd11eKKTpD9p9WCkvH9jH3 llrMXGdVgFNLLU8WaSGngjzSJxEUkJBNLHmZpajprsDjOVfcbwGsUdgmQjh6 9CjJQAJ50TDlwSeLUGiKTS/FcLERPG/pXXAJXCfSMq/5twALwc6O+YKUBAuj evfDDz+EmkrOu9esQN5HUC7ESyZtQehT5uzMmTMloSuKLATIpgNLAz7BA7Ax ePDYY4+Fi3xPUB7o/hRxkZKfSIOIgCgyJBNL/k8++YRA9qYJKKuRLrt8+TLu qAEudHSbV+3JYBk0Ihy9TFVIAwQyRiwiCIpGGQJGhGcgpT/96U/UjNVj/miL Csm2JIBqpjPEC5+TyZ51Vu+uJS7oSUuJwkA1BEhj+CO1fGiXb2Rm9xPdIEzy oR5tbaU/xDLgk/lKK54sxTlxTcWA2iOlUdN2lknVpHfixInhe9Y8cGiltH9V LKy2sMpT345WU2hhtXqMRplwCMq8QcK+NCMRuIYMliV8Wriy3L/az2UJwOYp k/tX+w2AbbRM7l89qABs62Vy/+rBAGA7VKb2rxYFgN2Xsty/WsSy3L9a3BJ8 YrMAm5MnT5Z45emV8oO7hdFvyfm9lZID9gihMH5qH8MUB1dAGioEOT799NMC KXSkVaY8w/x+BVKYp5ixmbPf/va3L7/8cgF52qvJKd2kjbV333235GLkkyeJ v/AAOYCRYYMC4BAXAXoA7jp69ChgcP78+ZL4wGDEYpH77bffbjp74O4U5CES 6L1169a5c+dY9jZtDq+UdiFefPHFkn0XT4OIcmP72c9+VnbvzkC7ihggJ6Jy uMBUrCR4UE6fEV1hbx59tjtRtj7WPLEUIKs9CmyWjhyDHf4W7fOZu8VjFiwU zDPY9+4bb7wxXJJSp5/85CcAJzF+/PHHn3/+ueEDdI0I8HutFKuqDFlIBQ7z Qbp8+fKFCxf0L2WgGCUFaGeGcA5OlBy0JlUF+6+++urrK6WwuuT21ltvGYPP Pvusag0iMtmz2IlAcmIkE3pCLFTFZIhsM8DFixfff/99qNi0UDIj42JkMxzu WM2uk9G0WpO2mG2i9sFq5YMPPiCWJXxarLLcv9rPZQnA5imT+1f7E4DNX9q/ erAB2NbL5P7VgwHAdqiM/StgdeEA2C6X5f7VIpbl/tXiluCT6Rp2MoF3XcUE bvY+sFLgAUuqh/59yfYpk4EUSk/8n//zfzZ4r169akW/cKEDRildyzvvvMO0 WVYz8SdOnCCfYvv44PuOOANRYciOQX0owEIHoKFT61MLVdgpPPkv//Ivi3iI U3R3PVvevcceewwu6tCTDnS56bnnnivBHH5N5gNhjvjeo3QD4pe//CXjaC1P FaGmBQLVk0WH0nY268yZM8P1qK0GUqIYYJIx8uMf/9hfI+j5558HwsnH9/6+ //77ffZWiR29QuuOHDnywQcfLOgxaJn7LBAoPC7ojLWV7s4pq9HRNNKOTRNO 3kcjvw/tKpou7NQlCEOJppHSb3/7W+uORQTYxr4ObSOL5pPMq6++avKcjJpV mdzLasXarcO+KU5+DhiF2CWxJXxarLLcv9rPZQnA5imT+1f7GYDNU9q/Ymr3 FQDbaJncv3pQAdi2lLF/9dJLLz0wAGyHynL/ahHLcv9qcYvO6qAKgpqCAS2X xtKp6KBMnrnLDDaCBnz88celD5vMkbdnAynMX/7t3/7N0vv69evPP/88ZIhr IgouFho0GDlk1UkfYbJ65nngAdwqlc8XX3xx586dJMNudlV8QUECsmE/jNy+ fRsUBAghSZCyI+ACQMHS8NIzzzxTqrWBKokIPPjJT34CFYDrly5dsn6nNuBB 6Y0W5TrD6lLa5UJCwUWYKi3d8ePHX375ZfIZa42OzluzlAqqz/6ST3tcx44d 64xYVSRDFRf3hMsyQc/COUnGQMj76Ny5cwYITQhjD/+ipqB2aZJYflnDJ4Fe nTlzxrBaXOxU1KwkU9QsU80f/vAHY8qE/Mknn1y+fPn06dPvvfce5THZ4roT 88kJWSnee98Yd8ToX0JbwqcFKsv9q/1clgBsntL+FcyA/aKs71sAds/y0Ucf lbV2XwGwjZbJ/asHG4BtsbR/9eabb/70pz99YADYDpXl/tUiluX+1aKWv/zf //q7z4ux3GHff5goYxWZm3HZ3s3zzz///GuvvWb5aZFVGOoH1c2YzbKoPHr0 tV/96rnHH3/s0Uf/Kl+RySVkUmqhXZbqF154gfn74IMPytO311L5bFfpRuHV q1ffeuutn//85/nShCfzrsktZGxH+OybfLROnDhRvt0HVXOUP//5z0bHhx9+ CEiD0/B2txuG88xQoT7/Py06dOjwSj6XU6dO/cM//IMa7jcTO1LyPrLgMkBy a4QYS6E1PLKG01Grto7Rn/h/WzRP/vCHz1z/5JM//elPDyTYNlcAzKX+tPiy RD169CjoaOJ97LHHRtSsyTtEk6WLmX97+tTNG5/+y//4Z/Xdb4aW5R5luX+1 f8sSgM1V/vLP//Ttf/kvV06ePPHXKzEAlwBsnfIXGnXpww9/+tOfPPXUD/Y5 AJtd/vmf/un/+8NX/+k/vbkPAdhGyl/+9//+t7/7u3NvvPH6s88+ux8A2ObL X/7v//qf//L8r35ljjpz5vTNv18CsMUoy/2rRSt/+e7//K/v/vXb7/58+79e u/joo49YLE6tgKyVxu3vv/3bv7127Zr1lJ6Fl1i9r7/+ekEDKWy0/Ov//B93 fnfj4wvv/+iZp558/NGHJw5DSenJJ5/81a9+9eqrrwKTHWaxesnHIvSf//mf H5gD0NVFj2MNjyDiZ5/duvzR/9/eecBXWV5//E9bq1br3tatuIpaheKsddW6 wWqrttZa7VRrh7UOlJsFCQkJCRsS9gqbMMIMe0PYGwLZIQnZe9zzP7/7XkKI Ad+E5N77vvf3/bwfPpfckbznnuc8v/ec5z3P5oSZ0wfGREX0CWtIPuhFdq9e IaqmZs6cYZRsDh8+nJOTo3OfvtHenqNfvY4OoxlUZmam0Qlq6dKlkyZNjImO CgvtDS9yHUGBgSHBQaNHxSUtmr996xZ9sdrHxsVQY/WRnqD6gJ6pYR8dOMa6 o/nz5+uAUplt7BXVIC9DgoP7hIUunjt927qlRcdyamuq7Oo5DZ3ENMCqffLy 8tQ+Gng1vOzfv1+vy1atWqUXsBqcY2JiNFAbGfWGAB4SErxmScLhHasrCzKk qkic9ow/toH5K/+DAsw09bVSWVSen75307LlidN7hQRTgDWDfvv1dVJVLKVZ m1YtHDJAZVionwuwU1JXLdVlFfnpBUe2L0iYHBQUGBwU5FcCzBzqFfVSUVhb lLVp5fwFMyZA24eHLfEDAdZi1KPqKqU0pzr34Jyp40YO7Z+8Yl7qjrWVBZkU YL4P81dWQCe4Ggy0mjKpLpKyHClKlezktM2JUaGO3oFfBjtw9HZ8Fhb4RWRo wMCoPhPGjkyYOU2vKPWiScWS3yyGdMVtlZe1FaowneX5Nbn703asHD2wz6CI wJCAHi4rfR4W8EVU755Do/vMmjZx6aLElIP7c7Iz9bLTn5YZOxGZ1VA15eXH snNT961enDB6UPiA8MBejs9DHF/o0SeoR1Tvr+PHDF25ZO6B3VtLCnKrKoyS hH9NfPW1NbVV5TkZh3dv3bBw9pS4geHRfWCl3jjUSl/2Dekxc2Lczg1JR9P2 OatLpaYCQ9WfJj6sOyoqyjuak7xxzbIliWNiBw3uFxYR3EMjkhGaNC71691z y9Lp2btXVBemS2UhQpkOUr3AcfrFoKupqVFRbdyAuSxp8ZyE6XFDBwyICgsP +So0AI7Uy/GFGqpPcM8NC8ambZ5dlbVDio5IRYGoR6mhdKj62bizBMxf+QcU YCY5SYBJxTEpTKnI2HpozYw1s0eGBX9NAXacEwIMmSsY6rAc3Za8ZEpcdHB0 WE8KsBPU12HowaNKpTxXitMqM7YW7U1Kmh4bHvRFRDAFmAvVUc7jQ089qrJA Cg7WZW3btiQ+aeqQYVFB/fsEbF02w28F2EnoWWs8ry2XGvWoPCnNlNztNanr kqYPnzai766l8Rmb51KAWQLmr3wep1QVStoa2TtLln4lc/8uY34hI38usV1L BnZNDnhwteNnsx0vL3E8fcDRMSX4x0dinkqP+03OgrDcdeNK8tIrSgrq/KWM VQfJVHBIto+Xtf1k1nsy5dfOuEcrB3VOD7l9b9A9SxzPLHc8scNxz96Au1ND 7syIfChv4vuF876u3L+4OnMbYpr/BCiVAbm7ZNc0WR5cP/P9mpHPFMV0yQq5 7WDQXTsc925ydF3l+NkOx31pATfnRHYpHP7Lilkf1a2NqU9Jwht14vMTIJxq nLm7nHtm1CwPLZ/yx2NDns4M6Xgg8O7tjns3O7qsczyyI+C+QwG35fTpVDag S82EVyXxY1kfIwfmSf4+vN1PilxOZ13p0dr8gyVrhhbO65E97IX0qAcPB92p EWmp46lExws66A4H3FoSfW/14M7Osc9K/Kuy+HNZ3VfS1kpBitRWevsE2h2j Vl5drZcgJQV7l+dvmpw15V/pI948Et45Jej2ZEeXtY5HFjieTwx4KSuqa1X/ +5zDHpSRP5Opb8rsv8iG/rIzHpfMerHjh2rTh2H+yg+gADNHcwJMRjxWP6xr Rcw92ZEPJAVQgB2nkQBDhB//vIz6uQzvWtLvnszgW3cF/oQCDLgEGAy1d6as jZS5H0r86/Co4V1rB99/NPzePY679wfcQQGGgVORD4WgQ29NhCS8B4k18nHn 8K6l/ToVRNyZFnzbkcDbSqLv81sBdhynOGslZ7voUFrdRxb9VyZ1l1FPStxD zmFdCiLuyA29taxfp6oB9zmHU4BZAOavfBcdKfU1kASFh2XnZATwiS/rHCch 50nQ98Xxf9WOszMd1x10dFzreFhVgf63LuQCGXS3jH5Clnwum4ZAS5Qflfpq lzawMU7XuuJSKUqT9LWyrKckvC+D7pGoGyTgO05Hh1rH94odF253dNrruL3Q cUlZwAUS8gPpe62M+yVeuWOCHF4iVQVIs9u+vKUepUFYJ/eUxbIqDApz2P0S eqEEn6MeVek4V+2T5bj2oOO2Y45LxdFBel8EM459RuZ9IFtHS95uKctGMcLe HoViVh2EoqoC9Y21fWXWHyX2QYm8Tm1S4Ti3wHFJjuOaVMdNaqUax1n1Ad+T gO/CUCMelbl/k40DJXU5SmCY7+ps7lFYllAux/ZL1gZZHiAz35XB90jkNXVB P6h2nLPbcfcmR5dCx8U6BnUkwqPU2SKulHHPwaS74jFgVZPr0LO3MDDKo2qo ymNycL5sGSkz3pExT0vEVc6gH6gvHXbctNnRZb3jwWLHBToScQR8R2JulqH3 45J5ZYhkbpCSTHyCDmF7e5R1YP7KzlCAmeV0AsyIZmWO8ynAwGkFmOvoUOC4 lALsNALMMFSV4+wSxwUVjh9QgDUrwIwYZRx1ju9SgJ1KgEnQ2Q1OddJBAebz MH/lo2hQKslCbFnwiUz7rQy+V2Jukz6XSe8LxfEdY7g5Hd+pcpyjV9NFjgtL HefjhxrGe50vYRcjkg/8sUx4AcFcR2v2FjvPdxXH5NAi2RwrE15Bzjz6Zul7 DexwXBIYKSw1kU52dY7vaTBHaNKo1edSvFINpe9K/EjW9ZNi19paW0YnQ2Ee WijLAmTqWwjLMbdK+BXS+4cS+D1DZ9Y7vqOGqnZ8X51KVQGsp0+pGcMukcgf QZGO/Lks+DeqPDnb7Sw1S7Mle6usH4j6y8jHMfT6Xi29L3DNdCespKOv1nGW s2HuU0OFXgRD9b9dRj8pM34vm4dL3l4pz/f2+bQP9XVYq79vjmwYKPG/ktiH JPoW99CDoRCmyhznFTsuOK6d/s/tUYHflzDX0FOhFfewLPxU1sVIwUHUAe0p opxYpq4Xs6v6yOy/StyjMuAuibgaw0pN4fiO4UuqxjWSIzo1KKiQc+F1fX8E w455Uia/JtvGSupKV4wi3of5K9tCAWaebxNgxhU0Bdi3CjDj0OmSAuw0Asw4 VIapI+m/FGCnEGAncjJqIgqwUwiws5pPXlGAWQHmr3wRrMQulqM7ZNcUxOGB d6G4cKohdqpDB2b4ZRL3oGwcjPW0NaWYQG2GBlsNuYWHJTkWa0EjrkLoPk04 OtWhs96IRyXhT5KdjGWi9TW2u33eicYdJemY0ONflYF3tthEbil1rox5RpY5 IMNqymzoUUazhbw9sn+uzP1A+t0E32iFrfRKZ8AdsvATObwUwgAeZTthoGq8 LAeLqxM/lH43fosSaP7oACk17pcy56+4Q6csF2sVbKbJNZLot3/sIK6FVVEP 7wJF1GKP6iBhF0m/6yXpK1y8FKexIYMvwPyVPaEAMwkFmFkowMxBAWYeCjAz UIDZF+avfA6dlXJ3yp4ZMsG1WD3sEul1XmsCOELTWRiqGsbHPovV7AcS7TXf OaUkQzYNk0WfuRerB31fAluuMw1hoLNk1I0y9jncQX90uxRn2Ge+U48qOypb x6COHPug9Lkcd0C0zqMCvgOHjL5Z4rtDRKUssVsHyGMHMEwW/leGPSBR16Gi p77RClsFnYVbJFR9qcGTesiRZegGaRs0jFQWyoYBkvgPGdYZhgo6pzWXLW5h cKlEXY9GBKpXMzdKRaF91iro0Cg4JJnroQ9HPYHSsAbkVlwLGzEq6Gx41JCf yPJAiKjSbG+fnr/D/JUNoQAzCwWYOSjAzEMBZgYKMJNQgNka5q98Cx1uFfmY ldZGSeS1WFrcyqB08qHRacG/oKCqitER0Q7UY9tT1ZkL/iOTujW+17v1h35I xJUy+ik5sADLs512aShanoe2CUu/xrLqPpe1gaH0UE2uCmrbaNSpUYmwPuh2 UisZ62TjIJnwUpMl/a08jMX/U16XbWNwC4lNWjE4pbIIVfJ5H2Ftf+vKo00O tBq4FUvZ9yVI4RGbNBTFfkDVkrEB18IjHkH9rnVSvGmYOlumvYWLl/x9rsth G3iUVWH+ym5QgJmFAsw0FGBmoAAzCwWYOSjA7A7zVz4EWi5kYmHw+OeR49Vh 0rpi1jcPjeGqoMY8LbsmY5moDbLrZUdl11RZESL9b5Pwy9tGZOqHqIIKvwIK av4/MUHoNGHp6KRfdE0F9s6Y/jsZ+gAKUkFntY1H6Uf1uRwlmzV94VE2oDhD MtZjW5YBd+LU2sSjAlwF04irZch9sjwIO+mU53r7PM8QJ1p9bhkhiz9D35LQ NpIEDlergbBL4FGJ/5SjO60vDJz4rlXhzPkbbo1Bp4XvtZFTfQf36aiX6iX2 gURcbhMvwfyVraAAMw8FmBkowMxDAWYKCjCTUIDZH+avfAhj4fq2MVgO2laF v8ZHzK2yOlR2T7V8mz794wtTsHvLrD827hHaNocqqL5XyfjnUK+x+iL2mnLE 8DXhMqiThF/ZxobSQ0X+rPdQ3bDBIva83bJnukx9w1X1a9OhpxdB6lQz35VD i7FbsaWpr8Fu8klfysQX26yU3HAEfBetZvQqL3WFVBVZuwWKsXA9fQ02I2ur wl/jI/hcdDzeNBitGKw+9CwL81e2ggLMJBRgJqEAMw8FmBkowExCAeYHMH/l Mzglf4/M+UDGvYDaSlsV/hof+rED78TSx/x9WNJs0ey6aj+dg3ZPx5ZAkde2 wRrjpoerCNjvJuTt1/e39kpajd5r+8qEFxu2/G7jo9cPsYlJ4seSugrVUoti bAy0cYiMeAxV8ja3krFtccxt2LN462j8OosKAyzvXy+7p8FQqp0C26iUfJJH nQedv/gzXEVauL2AU2rLZWWYTPsdYlRbFf4aHxr3om+S2K5oDV2SZa++OpaB +SsbQQFmDgow81CAmYECzCQUYGahAPMLmL/yDZzIq2towmY3d7ayv5yZMK5z aNzDuMe88LBVk8bVJegxuHGQqxFfm2un40foxWj3t+A/aC9gycnOie93b4LM /jM6YbaTlYxj8muycxJ2a7IoxjYuS7/CGuPgc9vNoy6ENlsVauGGFTpH750l ayMhBdvPndBe4E10Cyk4aNVLPNWZlQUy5y8YenqJ0V6GOgudV1WQ5+1BoZ94 HOav7AIFmGkowExBAWYaCjCTUICZhALMP2D+yieoKsI97BsGoZjSys1uTB4d sE3Mwk9kc6xVC1vH9suCT3CbdnvUsxqOwO9JmCqoJ1HYOnbAelKzqgRr75cF YMrWibsdPer/pH9HVBi3j4OVnBac77I2yeahMvHltmkZehqPUmEw9Q3ZFY+7 VCxHXY1UHMNOzSMelt7t6VH6FajTDu2M1f4VBdbrFaNXWznb0RhhxCOt3+zG 1NEBuw5NfxudPQotfluENWH+yiZQgJmHAswMFGDmoQAzAwWYSSjA/Abmr3yC 8lzc+r2sJ1aYt+tMp0efyyW+G36XJTPGTsnZKhO7YdPYNr+j+ZvH4HvQWyBj o/ViuHpU7i5sUNKuksDtUZdhN5x10ShhW7FUemihLPpU4h5pd3fSY8zTqACm rfL2Obecmgo0N47vLpHXtO+Vi8PVWyDiaqyT10sAy5VKdRQcTpKtI9FKor3d KfAsGf1zmfkOBjvxOMxf2QQKMLNQgJmDAsw8FGBmoAAzCQWY38D8lU9wbB/2 W5n4crvHJYfRHvNqbHmswVBDooXQuFSeL/vnSr8bkFdv8zuav3m4t70ear1S aeYG2ThQJrzgvve/vWO4yv75HyOGl1lrexcnDv1+h3aWiGva3Z30iL5ZRj8t W8d4+8RbSENJa/hPUcRsb49Swd/rfFnaE11WLLdhUG0F9oSa9S7CbLt7VAf3 xUvKEtyFYcWLFyvD/JVNoAAzAwWYeSjATEEBZg4KMPNQgPkNzF/5BEe3y+TX JfZBT5S0dMTpfDfyZ1J4CMvmLXSDc20VelTunCS9LvDETKdH+BUy5deyPNh6 pVINp0u+kJGPechQesx8V46skEJLBRCst6+VNX1ct420223yjY8+l2Pj3fX9 XTdEWGfo6ZVL6nLZPlYG3u0hdwo+R+Z9IBsGSnG6t0++hdSUyaL/ydhfoJuH BwyFG20ulX0J2MdZvybiQZi/sgkUYGagADMPBZgZKMBMQgFmHgowv4H5K2/j RFO+I8ul/x0SdpknSlo4Osjg+zDisjZbaWF2eR76YS4P9MQif3cMP1eib8Ue FtXFVprslG1jZMwzEn2Lhwylx9hnZU0kuohYCJ1xju1H5T3o++15m3yjI8hV Kl38mZRmoV5jFWorZfMwWfjv9tkeqLlDL/HGPC0J72NbbQtRV4MwNelVLL/3 wFoOHB0k5Hz0dNVJpLLQ2+fvXzB/ZX0owExDAWYeCjAzUICZhALMJBRg/gTz V17GWY/ljimLJPwq17pQj4QmPQbeJVtGSkqSlTLGOuNsGSEL/t2Oe5Q0jeGu JqITXpIqjUuW6iC6abAMuVcirvKcR414VBZ/jpaPFqIkE71DZ//Zc1YyjsR/ YG+XygJvn79pNEat7iMzfg9V4LGhN6wzGq3kbPP2ybeEuiopy8YdChqj2rvt ScOhgtyiS/0tDvNXlocCzDwUYOahADMDBZhJKMBMQgHmTzB/5WXqqhHDd0+R 0Is8N9z0iLlFVvWWXVOs1J2v8LCs7IVeeZ7TmR0Qxkc9LiXpFlvBviZcIq7E Pewe8yhVa/rV7Jnh7TNvCXl7UVCe+qbnrGQcs/4oqStxL4ZVqCmThf+TMZ5a ku1wdWDod5MMfxCNRCyDEwImbzda0XoymGOp/4fY0b7Eakv9LQ7zV5aHAsw8 FGDmoQAzAwWYSSjATEEB5l8wf+Vlaiux8+aO8R66+7vh0NCU1AOLnC0knwoO ypIvZcpvPFon1UODoSq3KkutYF/ZGx1Wgz1bUJ78muyM9/aZt4SjO2THJIl/ zaPupMeMt+XQAvizVagulcSPUeFt142bGx8qPyKukkGdJGOdt0/eNM56KcmQ nC0yvKtH3UnlU8J7sjrcSoLcFjB/ZXkowMxDAWYeCjAzUICZhALMDBRgfgbz V16mukSykrEBhycrNXpEXieJHyFjbKGNXfL3ytwPZfwLnrqv+fihwVBDogZG pxVWsOsfWVcjSV+jn4AnaxDRN+N++a2jvX3+LSFzo2wcDI/ypDvpAZ05SXJ3 evv8TaNhatZ7EDMh53nKSh0QEqOut9JW1846FJQPL5Wh93vUnfRycupvcDlc cMjbJvAvmL+yPBRg5qEAMwMFmHkowExCAWYGCjA/g/krL1NVLOnrscuDh+VT 32tlzt+wDUetdXZwztsjc/4q455Fa0FP2mp4F8nagNS6NeRTHe6JSPrKoybS o98NWOe/ZaS3z78lZKyT9TEy7jlP2yr+Vdk21kqNBVQ+TX8HLY491rnX4Wre G3GNpK7w9smbRofe0Z1ycKEM+YlH3UkvJ1WQL/pMjh3wtgn8C+avLA8FmHko wMxAAWYeCjCTUICZgQLMz2D+ysvUlKEKsGWEp5evR10vCz+RzbFWKv8d2y8L /iOTunm6/Bf7oOTukrJc6yxfd2KToKCzPbIb+PEj5jbU0baP9/a5t4TsZAy9 ia941J30mPoG+lTo5YBVqC7FlcuwBzwYpjqg1YM6Vdpqb5+8afTyquCQpK9G 41NPupOO9Olvy/Ig3AlFPAjzV5aHAsw8FGBmoQAzBwWYSSjAzEAB5mcwf+Vl aiugCraN8Xj7hRuxWcmWURZrv7D4MyS6Pdx+IfZhyd8vFcesJJ9WBGOlsSd1 5oA7ZFJ3dDOwEDnbUIbTP9uT7qTHtLdk3xw4lVVQ+TTvQ4l7ED09PGSlDhJ+ OXp6pK/19smbRuVTUapkbZThP/WoOwWfIzPflZWhUsTp26Mwf2V5KMDMQwFm Fgowc1CAmYQCzAwUYH4G81depr4W8/K+WR6MS64j5jZZ21f2zsRSZ6ugoWl1 hCS8j3DhsRgeeJaMflLKj1rJUMraKFR4Qz3V7FGPoQ9gH+R9Cd4+85aggvzQ QpROPDn09FBD6SRbmunt8zdNTbks+Qr7mPe5zEMmCviu9L9dRj6O/bUtg1Oq ilCDi3vEo+4UfK4s+Jckx2InNeJBmL+yPBRg5qEAMw8FmBkowExCAWYKCjD/ gvkrL+OsQweG/fOkz+US5MGuAgPukE1D5OACqa/xtglMU5KBZo/zPkK48FgM 7/VDGfdLqciX+jpvn39L2DAQpZPwyz3nUbEPyYJ/y8H53j7zlqCCPG0VGmOq TvaQR3VAT9e5H+CmlfJcb5+/aVQ+reiFfa7Dr/Tc0BtyH26IyN7i7ZNvCdVl Upwmo59wVd495VQhP5AlX8iOiVKW4+3z9y+Yv7I8FGDmoQAzDwWYGSjATEIB ZhIKMH+C+SsfwCmZGyTuUYm51XP7lQzrIpnrsdzREi0xDdBqdZ1sGOC5Vqu9 L5QRj0jiP7B811rsmY6q1uB7PWQoPeJ/JTsmyNHt3j7zllBbKZWF2Dck5FwP tarQmS7sUtxcUFuB0r9VqKuWXVNkVW/EKM+4U9DZMuV1WfK5lTa5Bk6pLvb0 VkF6ibdtNHYHU5VLPAjzV7aAAswcFGDmoQAzAwWYSSjAzEIB5kcwf+UTZG9F onvIfRLggRjeAanpkY+hAFF21DotBYxWq7vQMiL0Ygn8rifiUp/LZMKLsqSH 9eLSgfmS+E80PvVMAFenmv47OZhosQ04VMCogloRgi/aMxu7hF4kUTfKmkjU /S009OprsKvLpqFYNuAZjwo+RxLek9XhKNFai+oySfxYRjwqvS/yhKGCzoL3 qrgtTrdSLx1bwPyVTaAAMwMFmHkowMxAAWYSCjDzUID5Dcxf+QRFR2RVqMx6 1xONBULOk0E/xu+qLLDYkmydbjQ+pK6QQfdIxNWeKJXG3CqrQ2X3VIv1XlDy 96Knx7Q3PRLAz0bzkKQeUp5nNZ3pxLFjgkx8GTf7e8BWQ+6VGW+jOGstnPVS miVHt0EVeGZhdq/zUehPX4eiv7XQGLV1FAqX/W5sdytpDIy5Bd0e1FCoJltH kNsC5q9sAgWYKSjATEMBZgoKMHNQgJmHAsxvYP7KJ9DQtHWkLPyPhJyP+47b dcSFXiyjnsA+yDVl3j7tVuBEO8FRT8qAO9vZUB2wnnlwJ2zvm7LESk0qDIrT 0Fhg9p8hyNt7YXbvC9GqdHUYtFOd1Qyl7Jstc/6CvYnbd7LrgEPlx+L/Scoi b59zS3FinX/BIRn/nPS5FD1129VW6rThV2JTsLw9VhPkrqX++xJkfb/2j1H/ h6E9rLNM7IaNnIjHYf7KJlCAmYUCzBwUYOahAPt2KMBMQwHmNzB/5RPoiCvN QRiPuVnCLmrf7PrAu6wqCQxKM2VLHDpVtmsTUQ3gKgnifwUdUl1ivbw6FmZX yKbBWMEe+aP2jeFxj8qiz+TQInc1zXIUp2OP4IT3XQXldht6OtMFny2JH8nR nVJmncahDTjrccG1JkJm/gHF9/ZzJzXU8K4S/5pkbXaV3S3nUU7saJa/H1Iz /LL2lZohP5ClX8vuaVZqRWsjmL+yCRRg5qEAMwMFmHkowMxAAWYWCjB/gfkr nwChqRwLszVu9LuxvRZm68eGnIdtSvbPRcMHp7XWrh+nIl8OzMNq/7DLsHC6 neJSrx8irz7nr65F/tZp83gSTtk5SeJflYF3umtPbe9R38Vi5okvYbedjA3e Pt/WopPdsQOy8FNUxtvv/pFe50vEVbK0p5RkWa8brUFdFUpyST1wWwdqyu3h Ud/BVzD+BZn7keTv8/YJt5baSuxEM/1tdKtov7Yeev0YfqWs7y+pq7BtNPE4 zF/ZBAow81CAmYUCzBwUYCahADMJBZh/wPyVz+DEInaNTov+216FLTTDfEEW f475wnL9BBowyhAZa2XCSzL0/nZZm62qYFAn3FBwOMmqRVKDolRJXYm12b3O c9013+YedbkMvFtW98FkYVFJIK4t1PVb3jdb5n0ow3/aXpPdqJ9DO6UssVjj 0JPQGJWNtsOTukvUDe0gNTvgsiXyRxh6Odss7FFGo5j9c2RD//Zq6xF4Fm6F mPaWZCdDrVloFzMbwfyVjaAAMwcFmHkowMxAAWYWCjCTUID5Bcxf+RCVBZKy WNZGSt9rXDsUt2F2vQMqZf1uQj1rfQwWNlsbp+TvkTl/k3HPoaDZtgoq0LWd xKgn5OACbEZs0SKpgbGMNukrib5Jwi5u45lORWbMbTL2WdkyEgHcqkXS46St xmYr459HvaZtlxyrcNXhPPl1SY7DZGdpqsukJBNDL/ZBbOXTlgsVXB6lqmxY F6xPKM6w9mYuOhwy1qNPbNzDEnphG8coeNQPZfJrsuh/Fttwyl4wf2UrKMDM QgFmDgow81CAmYECzCQUYH4A81c+hLGIPXcXygTT3mrLtdkavbHvxu8lb6/r 1m+LVh8aofN1QYrsmYFF5v1uaMswriIzqQc2sKgqsnCR1EA9qr4GDVdV4cS/ ClXQVv0M1aP6Xo2Gt/rhOp9atPFCY2oqoDaTY7F2WmVhG052gzrJ1Ddl+ziX R1lZEgCnOGtx88v+eTLiMSiothIGKgnCLpHlQbJnJoqMNtjMRWOUetTGwTL/ nxJ5XZu5kw7h/h1l5GNyINHyItPiMH9lKyjAzEMBZgYKMPNQgJmCAsw0FGB2 h/krn0Mno82xsvC/0vdabCxyhkVAo+WCDl4bLFxvjLNOqkrQ9VFVQeyDSIaf qdrsIIHfR+FPP23TMDm40NoL1xtTmIJV0/M+ghv0dpVsAs7AqdBy4Wz0Vh3e Rdb2hbvqF2EDnE6oTWyF83csDO59QRtcvwSfA4Ex5ulG7VWtj1pJZ+2cbTLl DRlwF2IUqqVn5lH6CRFX41aILSMkY6NNPEpcfaFVDa6JlKGdIQ7PXGoGn4sh PPJnXLjuCzB/ZUMowMxAAWYeCjAzUICZhALMPBRgtob5K58DwqAI+5au7y9z /47odCY3zoddihW5C/4tebuRVLfwrd/fxIlqaf5+bJY64UVsv9vq6ASRea4M /rGsi8SK04p8620aexqMrXBytsquKTLjHbjEmbQ0jLhKhv5ElvXE7iSqndSj 7BTAq0ulLFt2TsDO10PPZENnV7/WuIcw9PbMkIo8qbWXR9VVSu527EM98RXp fxuGT6sNpdcsMbfKuig5tBAxCvUs28Qol0eVH5Ud42VFMDZ0hiZvrdTU+Dbi EZnyG0jxolTXXUg2MpQFYf7KhlCAmYUCzBwUYOahADMDBZh5KMDsC/NXPkp5 LvLGq8KQEo+8DuWtFvQU7QDFpVNk+BVYOjv7r7I2Cr0d7FH4a4xRBMxKlrkf IIxHXoscuwaoIJO3z3eQgO9hJba+K+YWbLe6Z5pkrLN8J4FmUamTsUGWfCWD 78NN7irL1aNMlgL1ZRq6Q86D9Bp6v0zqhmXeZUdxM77NUCmoTnVkuawIgUdF XI36XfA5LVDmgWfBsOpRfa+Vyb+WddFodWvDac4JB8jfiw2pRz8Fjwq7GEPP 5P0RcLzvuofe4Htk9JMYenqJZ6fLlgbqa+TwMtk2VsY8g7tjQi/GUDJ/x41G MzWsDr3IH0E7LfwUtzix8OcDMH9lWyjAzEABZh4KMDNQgJmFAsw0FGA2hfkr H0VjeHWplOZghfbOSTL9t9g+I+R8U6VADUoakSa8JJuHYwuG0iypLNQxbMcY Ljip+mpE8tyd2DVj6dcy5CemtsA2Sn5R18mUX6ML35GVcnSXVJcgLtkS9ai6 KjiDGmrLCJn7N/ci7WATi7RVa0Vdj3aFq/vIwUQpyUCF2maFv8bUVuByI2er 7Jsli/+H1cJRN3571caQBNG3YL26+uHB+TC1HVounIL6OlyRqUcd2y+bhsqS L9ELpe/VJmJUBzhexFUy7be4p+ZwEor4lcWu6ztbxihXcw/1hJwtuJFk/j9l 0itQ12ZKgUaT3tgHZWUv2TEBerU8DxHProayFMxf2RYKMLNQgJmDAsw8FGBm oAAzDwWYHWH+yqfRwFucAQW15AuZ8Qfc7NzvJhT1+lyGHHLvizCvhRrHRcgP 97kCKeIBd8r4F7EX7cGFaPTntH4jvm/HKZXHEJqS42RiNzQ21Ok+4mrYJPQS t6E0YhuG0p+oDSOvQRc+lRAL/yvrB0lhKnpT2N5Q6lFVxTDUqlDcwT2ok/S7 WcKvbORRF7iOC/HfMJeh+l4jg+6GSRf8R3bGy9EddhZOjSnNhoLaOFimv43T j3R5lBpKzQIvuvC4oS7CT/TnaihVWaOelFnvwQ/z9mCmsz11NdhO/UAimpbE /0qGd0WMchvqYhin94XNeNTAu/HKxV/IxiHYwKW6zA88yonWzccOYjFG4j+g iKJvdRnqckQkt0ddeNyjLsWhWlSvBEc/harftvGSukqqCu0fo6wD81c2hwLM LBRg5qAAMw8FmBkowMxCAWY3mL/ydVC1qZTyfCnJwvLO9DWo3ayPkaQvZeEn kvAnmfMX5JMX/Q9r3TcNlSPLXDuSZGABfG2FDZesnwoNv3q+1SWY9QoPSdpK 2Z8ga8JleaAs+BeaZyb8Ge0skr7AfdCbh8vuqVgFWpCCOQ7FrFpr79RsFleT zJoKVIRLs2Co1BUoK6hHLf0avQJm/wVWSvxIlnwOQ6mzHZwv2ZulJB2Gqi71 I48yylvqG2U5qHDpsNozXdb1wxYtKrn18kRlkv676FOsdV/fX/bOgtwqOORa 2F+MRcv2lwTHqSnHKcOjDsuRpbJvpqzpi228MPQ+wK7xc/6Ga0AdjFtGYqX6 0e1oIFBxfOj5iSTAdUetVBa5dNR+tGDV65FNQ2R5gCz5DBV5jefz/oFhqENv baQcmIuAX5iCmKZDj0vWfQzmr+wPBZhJKMBMQQFmGgow81CAmYECzF4wf2UZ dOBUl+Em+pQlCNSbh2HKWx4sK3ojn7xxkOyYhCS8hi97bNB8JtRVIThrdNo1 GRsxr4tGMFdbqcLcPBT3QR9ciF4EfiUGvomxuXPRYUlbDY9KjpN1MbKiF6y0 JgJlLzWUOpu7jGX9DZrPBJUHaoSszegFunW0rB8gq8JlWaCsjpANA2GovQnw NxWldr39wRROKAQVkNlbcHmi2nttP9zysCpUVoai0qemS0mSzE0oY/nF1cqp wSVMkaSvQ9DWGLVpMMadxigdehqvto6V3dNwfccNmn0Y5q/8CAow81CAmYEC zDwUYKagADMNBZj1Yf7KUriik0bymjIsQtbRp+G6yjiKIAZ0SCLD7N9xCThd O3RUwyZVJW5DVRa4DaU/QWG0yt8lgQE2x6ls5FEFrqMQ/612GQplLL/3KGNz Z/UZNZSaBV5UeNxQRfiJ/lz9zWnXJictoa4GpoChimGcysKmHmWzDW5aiSv4 YOiVIyK5ParwuEeV4qiv5tDzZZi/8jMowExCAWYaCjAzUICZhwLMFBRglof5 K0IIIYSQFsH8FSGEEEKIh2H+ihBCCCGkRTB/RQghhBDiYZi/IoQQQghpEcxf EUIIIYR4GOavCCGEEEJaBPNXhBBCCCEehvkrQgghhJAWwfwVIYQQQoiHYf6K EEIIIaRFMH9FCCGEEOJhmL8ihBBCCGkRzF8RQgghhHgY5q8IIYQQQloE81eE EEIIIR6G+StCCCGEkBbB/BUhhBBCiIdh/ooQQgghpEUwf0UIIYQQ4mGYvyKE EEIIaRHMXxFCCCGEeBjmrwghhBBCWgTzV4QQQgghHob5K8tQXyNlR3FUFkh1 iYjT238QIcTvqauWukopzZKiVDxgXCJ+A/NXfgQFGCHE16AAI/4K81eWIW+P zP4rjqUBsnWM1FV5+w8ihPg3zjrJ3y9HVsr4F2RQJ0lJkuoycdZ7+88ixBMw f+VHUIARQnwKCjDixzB/5VHqa6WyUCrypOhIC47SbHHWSnayTHpV4l+V+f+U DYOkttLbJ2M7UGDNxVFxTKqKWchonpoy1KDL82Clmgpv/zWkvXFKRQG+8Yp8 PNAIdtKTdZK7Sw4tkiE/kd4XyoFEqS7FD4nHUL1aVSRlOVKShVIs8SDMX1kM CjBfhgLMDBRg/gUFmG9DAeY9mL/yKEVpsiZK5n0k/TtK1A1mj0ndJW8fUuv6 gPKp/cjfJ/P+IXM/kKQekhxHCzfPvgRJ+JPM/5ckfQWfJPamtkJWh0vCn2XR /2RFiBQePulZyifvonpp/1zZNESm/VYmdpOcbd7+g/wL5q8sBgWYL0MBZgYK ML+CAsyXoQDzKsxfeZSiVFkdIXP/Lv1ulIirTxxhl0rAdyXwLOn1Qwm9SCKu OunZ8S9SPnmCoztk6lsy5dcy52+yth8mDvJNdkyUia/I9LehMzV0E3tTUyZL vkTkmfW+LPhEju0/6VnKJ+9SVyW7psiqMBn9tAx/UDI3ePsP8i+Yv7IYFGC+ DAWYGSjA/AoKMF+GAsyrMH/lUXRG1viTvUUOzpcD804ca/tK32sl+laZ9pYs /ET2zjzp2Yw1UlMqmRu5fL19oXwyA+WTX0H55MtQPnkV5q8sBgWYL0MBZgYK ML+CAsyXoQDzKsxfeRTjBv/iDMnbjbDTcGwfh2XqA+7CxL0qFJ0WGj9bmIJh krX5hHxaFyOlObghujjNfZRknm5bHA1oVcW4e1pfVpLu6uqQisdlOXiLlW6i d8KGeiLlasZ0HO5zycC5qMh0ax7Xy/SxWrs0220ifVlRGvbp0PfqU41vVa6v RTOB1BUyqZtMeFFmvC3LeqJfq7v9RZbUVze98dxncdbjT9WvtewovmKctetQ W+l/1XR67mqc46/G/Fie57JSuttKeGUGfqgf0rhLrb5RXW7TUBn7C/jhjN/L 1pEnmoSoAfVjrT51qleoQfQ01QKwxhHX4MpAyxRY4xu3t6taUMczmlEYR1n2 iVGpT+motOqVjhP39etZaMAZ+6xMfUPm/EUOLz3+jafCUHWVkrvzuHy6ALc2 wAI5rt1wXL5UmgnXqil3udw3QpN6V205xmPD641BrdauLGz+LT6E032ylcfw LRuHDjH3t5/uDsgYEU5X+D2GyO8+xwycYLNSU39YVehqp5B5kk3USmh/0cgm xnhMjpUln8uIR2Xo/bJ31olv58RvJ+0F81cWgwLsTKEA+zYowM4ECrATUICd Hgowf4f5K0+j/uysde152ug4vExibpMh98viz2XLCKkqOfkFrskuZ5tbPiX+ A7dCr+wt8/8tw7rIkPtk6AMy9peyLFC2j8eY+ub2ExrBto2RFcHuLHHM7dir YtQTMu13snWUpCxuNJ/6MHpe1WWSu1tW9pJ5H0rsQwjaxrmMfBznsmOiHFnu Fk7HDsqhxdgtaPLrsNLg+6BO1cLjX0TdSgN+zvYTsaXgkCR9jVuYe18kwT9w 3URwsURd725/MeFlydyEiGSJfT10ytP4uWWkzHgHOkeNM/BuGXCHDP+pjHkG eyepWfR8xSUadWbfP1cW/BtFz7iH4Uv9O8qwB+Abif+U5BGSlXzik9VoCz+F qYPPlZDzYKU+V5xoErKkh+TthZCwNHq1smsKTlMtoPNR9K3wnLhHZe5H6MiR ndz09buny/TfY0gu/Rr9KJZ8KZN/DTfTIanH9LcxTo+scHmOLyuB5lDf2DhE Ej+GEfQb73W+9LoAqxSMr1udamc8LjF0aBjySV+wcZAcWQZrTHgJPxlwp4x+ CgbZNxuvVBHVmPo690KIOR9gVML3bsdQHfU4mnus6oNSo/4NPjvo9HRW9JKZ 78qiz2RZAMKvDgE930H3wmeGd5XlgbJtLCSQaqctcRA56lSD7sE5jvwZmlrs moprnMZotFePWhsFsTr6SRl8r/S/Q4Z2lhGPybwPEMA1+hkbDGmUU7/aNh7x PFSj1jkS+H3c62R8O/qupT0xHehVNmk3mL+yHhRgrYYCzAwUYGcCBVgDFGCn hwLM72H+yidIWw35pOFaB+CptmY25NPEV2TWezL3b7L4U0l4H+MUk+Od0BL6 39V9kCVuXM5D4a8IKfpVoQhrxhTZ7xaMr9gHEeV0SG4fhwy276/W1lhaeFgO J6F/hU73eiID75R+N7vPJf5XsjlWDi5EYNHIpnFm3xzMX+NfcFtJLaz/auCa +iaEgcb8ukq3grKTfCrJkMz18ITxz0vcQ/ANnZX6q3f9BDWCxZ9BL+W7FiFD PlXInqmS8B6KnsM6wz7RN8ugHyP4T/8dJgW1khrTEPD2lk+GmFT5vSZCZv4B 4kdNF3UjvEv9Z8pvIJAOLkDNq/G1hs6PY5+F7818B9Wx+f/ESm+VT+qcg++B VdVX98xATcdyW5O0VD6pS2gw2RUvs96VMU+7fOkWxDR1mE3DUDesLDjx4WqN 6hLZn4gxO/UtGflzGXgXXj/kXon9KXxPJX36apTAfNZu6gkL/yvjnsNfm/An mfcx+kLHPeIacbfh3zl/lZUhkrEeOlCdSvUPxuMdeEp1kbqKyiQNaFhXcFxa a+TfNg7XOBNflhGP4IoPAv4eGd5Fpv4Gn6CSqSgdL6N88gGYv7IJFGBmoAAz AwVY66AAawIF2OmhAPN7mL/yCczLpxGPSb8b8EoNayt7YSrcM1M2DZZF/5XI 6zA/7pyET2tYa62T6cbBSFAP7QIhsWkIXpC6XA4tlG2jsUhbR6jGf42Eqat9 fYV27i4k21XyDf6J61yGyc7JKDdo9NaT2jlRdk1GlNbAopE5azPKmhv6y+Zh mPVSlqDGqrZaHohYp9ae1A0SqyzXtdbdRsvX9avU2dw4VoVBberXrS6xf55s nyB7E/CTwhS8Ut2sqhhTmHqdOob6klpJX6lWUqOpEfrdiPsp9iVI/l683t7L 1wuPQDvpDBV+BZSniqUtI2ArnYPWhON8dW7SK5fd03DvSQOGfBrUScIuk8m/ gU+q16l5s5PxXp1PI38EWQUb7vPeubWOFi5fD/weFLgOzDV94Wn7ZmM19az3 ZdyzqGTpz9UmTpdOMJo2qKFUe/TvKAv+Keui8HodofvnyI5JMuUNXOao76mA N3zVBzHk0+incQp60aGhKakHQtDBROiixV9A9sCXXpAZf8CCjeVBsnMKhqFe 2iz8BApK5aJK6+L0EwFfH6i/6Qv0klafUmunrsSYVZk0+XVoM5Vky4/bhMvX vQ3zVzaBAswMFGBmoABrHRRgTaEAOy0UYH4P81c+gXn5pCMo4ioEq6SvMZHp D1UnaMzREdf3WoQvFUsHEjHdGxQcxJJaDfvDu2IAaozS31V4GBNiyiJZH4Nc tEaqDQNQLPPZTLtGcrVJxjoswI59SGIflqm/RSBKXysFKZiYNMikJKHVqv4E zQdKsTpdz1RllZ6yxmp9jZ612konuAWfSPhVWEqqH6g/bCjq2aN9qE76GktV BE7/HUKrOoCeo4ZoNYhOT2ooNZH+V1wlmKoSObwcJ7t3Jnwpfz+eytoEKaVG UEE+/W18iNqtAbu2D1UZoHPftN+izqXKUweXqoKCQxCfOybgfFUgxb8m6/uj oNOAIZ8G3o3C8fTfY0Hy0Z3wwPI8OXYAc6iKLv0iVGfmbPHeuZ0B5tuHqnwa +oCMe0G2jIKnZW/FvQ9LvpBpb+KGCA1uaWvcpS79V/1QraqRJ/pGWe6AU6n7 QXZuldRVMvvPGOmTf40rncZi1adokE/9boJInvuhrImUtFUII+oGGwaioBl8 DgLv+BcRrvUn6WvgHnqya/viujX0UtxVpKG4YWG/DkmVQHrxoiZF3fAghBCk +DKZ+Z6MelKmvA7j6JWd+/VsH+pNmL+yCRRg3wIFmGkowFoHBVizUICdCgow v4f5K5/AvHwa/RSWdya8h2muuhgyCQ0PyzFwNOZrlNNJbV20a+p3SY4jy7DG e+RjkA0H5uMtdZUIX3XGGu9dKHPM+xBv1ECnAcE3qS7BKlmd0fpcBom4dwZi C9o5Hj+X2kocekbuVo1OmEUfq44yWhficHW9qCpClWf0kzLpFZdonH1CNNpD PqkkVl+a8gYUdeoKnDU6ftS5DFIBi2Hta0Mp04mfqE1QynE1+oCVqmC0fQky /2PMX6oodk878fl2lU87JrluD+mG+vKWkSiMos9qrduLVFnN+iMWFcc9jD4M DRjySUflqMdl89ATi9vV4PpAFUJ8d/TBmPcRtL0VMS+fQn6Aovme6WhQbLTn VdfK3gIn0TEbdinqWaU57nilckKv5vSyLvFfkr3Nvbzf7XtlECF64aOm7nej q6FKrS82rzDk05hf4JJ2zJOQ2ZVF8BbcPlOGvql6gtG34FpGnUojvI4yY9m5 nqxG74kvYSHH8qCmTRiqy9y+px91YjyWypGl0FojH8cFr8owA8onr8L8lU2g ADs9FGDmoQBrHRRgzUIBdioowPwe5q98AvPyaewvZNTPsXa9idTRIKZReuob WCq5MtTVRLQOg+7QAul7DW4G3zgIJZ5jB06sN9ZDf2/Sl8gnj3oCEV7HtW9S WQgduLYvbvkf3rVp2a55mt0ox3VoPB/9c5nwAlo7apRusLY95NPGwVgZq56w zAFxWJqNNa4QSOXNLy83FrGX57l27kg/vl3OEUz9iR9hEfuEF7GktgG7yqfk ETLiZzg1HUc60TdBFcKi/0E+9e+ICmADhnwa/zxmQ7VYE1R2Tv8dPnDmuxh9 VsS8fOp1vqzrh7qeOpv72XqMUx2tw7pgc5yd8eilqf6mcUYjmF7CqBfpAxVI jYNSwUFELZUKGrX6XI4P980mooZ8MsSzSu6crSc9qxd3enYD7pJRrgaqR7ef 9KyOuMmvScytaDmrLqSDtAH3lkCuHamM7W8MmxyYI8nDIUQjr8NlkaEnKZ+8 CvNXNoEC7PRQgJmHAqx1UIA1CwXYqaAA83uYv/IJWtQ+NOFPuMG5yeSuYW3h JzL7Tyg3aLjTuVKP/H2IQr0uQGe58Mux7r2h2aNxRP4Irec0RkXfItPecu3+ 7JPozJ4cJ4s+xSnoVKWxpe60+/UY21Uf3Yl+hvM+QM+KoQ+gEV+/m1FQiLha gs7BKati3DDoxO669pBPOhktD0I8D78CXQc1Ss//F+4HxzJ+o93EyWRtxnLZ xH9gOtPg3P92WEZ9I/xK6X0hLDa0MwqvDdhVPq2NRMPYya+j02NDeaUBHX06 zek8GPwDmNfYk1eOy6epbzZvjf3z4GMzfo+2HirUrYh5+aQCac90d4Hv+NOQ UoWHUMjTC5/kWKiIGld5Sz8w4koJuxg+1tCStOHQEarPBn0fn6nyoDij+ZDo XRrah6o81iHWxDIqEed9iKLwrD+iZ6/qn5OeLUHpU1WlmnRdDHpcGNTXYjwe SMRieNXqw38qA3/sHo9qEx3RGsl7XwS/qnK1VqB88irMX9kECrDTQwFmHgqw 1kEB1iwUYKeCAszvYf7KJzAvnzTs6KhcH/MN+XQA5YnZf8bUtvhzDG09cnfi ZnwNXBqIVHhEXNM0TLn3SuiIngY6YH22/Gc0QdUY1fcaCAMNv6dHFVHRESz4 nPt3bBuhUcho1tfvJsgn/ZDgcyEMVCat7283+ZS9BZWs6b/HNzvsATRvnPkH uMTGIQjL6hJYHOtyMGOd7cH5svQrVKkatm+OvtkVrq+CtI66EZu/qPEbsKt8 Wh2BGUq//Q0DmpmGjG2dp7whgWehrtpEPk37LXpsqiWboAZP/BjfxYTmioOW oEXyaf9sVOob3xyhOkFHYtwjKA6qSs/eiiCjHqgGCbsEm0zpoRdxzcYlPQb9 GM0zfVw+6fUsBNKBk55FidN1SaLBZGlP94bpjZ/Vd8V2xS4/emljyKf6Olzz HpiL+yD0YlbfO+R+1w44N58sKS9Ev53KIrye8smrMH9lEyjATg8FmHkowFoH BVizUICdCgowv4f5K5/AvHyKfxUBef3AE5O+AeTTZ5BPGpf0gSGfjm5Hyl3l 06BOsnsKbhDWUdx4pWjDUZzu3gjGN4F8GoQZyqR8MpqmJvwZ28Lq1LZ3Fvrv qXI45moiqpPa+OdQRtTItq6RTLKHfDI2xi3YL2krsex8ZW+E8cmvQSFHXY/l 6Mlx6Ouo5O7Grd/qM6oqdXLU+T1lCWqmOkWqS2hYXvQpqsk6hW0YeOLzbS6f foOTbal8Umss+A92WWqCWz659hBvfAuAhWiBfLrwRFnq+NPQCUWpbvm0aSgc T51Tx68aJPwKtJFZGYJ+ts0GJfcmO+U+3X5h3POS8Bf0iW1a4DOW6D+FYbI8 sOkmPpBPf8am8yqf1kRhmbqzHuvVNQRNeFFiXCsT1CF1POrVkEYtHc6HFmJP HDSqvUD2zXLtNlVL+eRdmL+yCRRgp4cCzDwUYK2DAqxZKMBOBQWY38P8lU/Q IvmkKqLxomuDZuRTOXaO2DoSy9eH3I/Rp/rBort5Yvl6LLaoNrl8XU981nvo pzrqSYSprE3QjRqyDKPlbMWGFBrAm8gktc80Qz793cLyyUBnqOI09KzYMhJ7 4Ko1dAoLuxRnvTwIS9yVLFdHVlWMKkoTP8KL8/dh+jOKg/pf1QmTX0fEbiyf dk7CYmz7ySf38vVfYxMTndCbcPrl63aWT+WQT6qiE/707fJJz1dHmRn5pC4U dR2q82v64gVWpEE+zf4rTHRK+fQhHOZ08ikSFkCrihTJWAtbhV6CZs4qlnQ8 qkxC09FqVyPWOSjQq3zS68HyfJTv9VpJ5dPqMOwWRPnkcZi/sgkUYKeHAqyl UIC1FAqwZqEAOxUUYH4P81c+QXvIJw1iOkg1pKvkGHinLAuU7RN9cRWoGVra PjRzo8Q9ilarK0JQrsJGOTWu+a4eOfO0VaiHqrXVXCfJp11YaTz1TZn9l2Ya XFgLZ/3xrVtKpPIY9tfQMDvvQ/hP7EOyORavMSb3Ge+gnKcyoLbyeJ3FJQx2 T0PlVON/U/k0GbsYz/i9zPkrWjrYhob2oTP/0EyrT3f70FdkwO2y4RvtQ+0t nxq2qvlmk4HWySejNcHIx2TsL5tprWkV2lY+qRY6skK2jYetom+BSdVQ6Pfr inIqk/bNwV1L/Ts2lU+7pyNY6S8a1oXyycMwf2UTKMBODwVYS6EAaykUYM1C AXYqKMD8HuavfIJ2kU9O1HFUdWjAH3IvRvG6GEyjOnJ1lsRR496KwrghutpX 926W49s3bxyCApaZ7ZtVPg1/CEs6V4VBNjTYU0+5PA83OPe/FauyE06WT0bR UDWDTqCrervCVx3Clw9uvXEa9HQqjrmqeJXu/W6M7/rAPEzx8b+Cm20aglfu nwevQHPL7hCZOFOXcKp3be29ZRQmfQ3mTeTT7qlQmCoJZr2LdpHu7aGPv9e6 GNs3T+qGto167s1v3/yKjHhEtnxj+2Z7yye98pr2lsx8B9U6FUsN37i0Vj6h pNgDxhz1cxn7DGJURYHUVLh3cIbNKxDB9KJJf36imYOP0eby6fAy2TrWLZ/0 MbqwGuOxFhuZbR+HNg79bmoqn/YmQFaNfgJfQeoq/MT97Th9cc2/vWD+yiZQ gJ0eCjDzUIC1DgqwZqEAOxUUYH4P81c+QXvIJ+B0NSLogQAedT3qPjoP7p6G VZH6+tzdkrEB8+a2MZgR9s9ptG+Fr+GETTLWyZinEXM0wkz7HWb/9LWoA+q5 pK9BED6YiNdotDm6HQULnQenvYl7n6uKEGr05wUpuKk54U+ujV1uxIPG8gm2 +kIS3sd2Oaqj8vZg2TykSLE4ay2z8v9wkiSpF42GTfJ24RT0rPW71hMf/ST6 TkAsubZiUa9b3QcbPfe/HQ8q8jGvicuXdB5UO6C14/VN5ZOKBI35U16HO2nk L06TsmxYSd/bMKtaEZUBu6egjWqvC+A8ybFyeAlmvdSVGCM6gsIuQ91zfX9I 9wZsL580zui3n/gPnKaOvoPz8Y2X50LYGMuqWyGfdCTCquNRsQq9BAJ+mQN+ qx+lvqfHkeWoLK8OR/leB7hv0sbL1+tw+aYuF9sVjVXV/TTgw5j1aM6QnYxQ pld8IeedJJ+wmGG17JyIDskR12DuyN0ppVkuOVqOy2erX9T4Nsxf2QQKsG+B Asw0FGCtgwKsWSjATgUFmN/D/JVP0F7yyWi8OQT3TUffjMKZfv66GElZhHGq 0X73dPTPXBcpG2JQyvFd+eRCo+vMP6C93pD78a9G453x6Imq56IT3654/Ksi SqNK/l40B9CwHN8d85qqiKIUxGG1c9KXqGXoVBhza1P5pEFMI/b8jxHWpryB SUEViAauY3ultszXjdOAyuC5f5M1fWAN/YohLJfigX71I38mk3+FphP75uCV emrJcRCKA+6E4srZAl1dnI6pTRW1Cok+V2D7mybyKXUFNsWe9lvsiK3vSlsF OaEfVZLmrsZalKIjOLV5H2HfavWu5QGydRRMp7P8mgiUTSOuhq306kOnuQZs L580FqkzLO2JpeYjHkU7F/3GszYhHKlm1nGRu6PF8sldNFwso5/Bxi6TuuGG Ef0t+vaUJTh2TJDk4dAkGsqOHTjdn+dF2qX9wjrcdKNXdhqIDszFFZw+peNL LTP5Ndx08035hLYMc3HnReR1uMHHuITM2Yo9g4zXkHaD+SubQAFmBgowM1CA tQ4KsGahADsVFGB+D/NXPkH7ySeNVFVFCFar+8r8fyMA6m/pdzNWSOq0OLQz Aru+a9dkLH308blPdc6xg4jVGmzVDoM7yYA7cC7970ApKv513EGvz2rE0OiU vRUL19Uao55AlWHYT9GQQWOdTnOqrHTKm/hS0/ahRsNVlZHjnkUSXr8RbJx6 PXqNZm5yhzjfZ+9MmfmujHsBX65+xfpFqzoafI+MeQrLznUKO7YPlT5xnW95 HgSAaiEN2joDwkoPQzzMeAdeMe9DuEcT+VSeixLDqlDcPj/0J1ChGu01dC/+ UvL24gMtirFS/chyiCVV6SoO1W6qHtW7Bt+LprJLv4ZA0pHVeEqyvXwypvXM DdiqZuSjMvAueJRqHjWOXrzo/K5Do6XySY5v0qRv3DZaZvwOhWn46s1oL6Cf HPsw+mEu/kzWRbu3NvZB2lY+KXU1WGmwZQQ+TU9fA/VwHY8PIRypwlT3W9kL xm8sn8QVFcuPurYo6oa3xNwOp9XYpYpXXa4021Pm8EeYv7IJFGBmoAAzAwVY 66AAaxYKsFNBAeb3MH/lE2RtxkbDGmNXR8iuac1XmjRS6TjVY2lAMxKr6Iis DJXFn6NCoQ+atL4sOyo74vHh+itGPIblysakoHPllN9AVmnkz97i6/IJUrAE +wsvD0TvyuFdZfB9CBeDOrm2aX7bVQ1c6W4rUZyBqVCtMbE7VIEhnya8IsuC EIgS/oQ9bpJ6oP7VIETV7CVZeNfMd2SCSzZoFFJtMOFlK8knLC//D8qXer46 qcV0lIE/hhiA8P6X7JkplQVu91B76ikfmC8LP8XCbH2NmlRdIr47hOXiL1CD UDtrJFd/a0BnPexGNAKKYtTjULDGfKrK39LyycDYplkNoqpbJ6PoW+FjOvvr JKiukp3c9PW7p6PlrBp2eTDaPzbh8DLsPaTPqqD9ZktSa+DEd3rMdWfH5F/B STR6qM7R0WHIp+xtcmQlAosOw5QkdHFpPEz0v8XpKF3p67eNw/bEja/sVIcf XgoJYXyy+pK+TD9HVcfEbmiBsm2M7woAvfpY0QvfrI6UVeFNN/FB24oArDRI +horLpqIQDWCXuRqaFK32RyLAWWg43H/PCzYmNQd4gq36jwM2+pYW98fXUFU Zw76McZ4Q2MKNaN+2voBro6sj8vAThiPGt7d8inHI7bwU5i/sgkUYGagADMD BdiZQAHWFAqwU0AB5vcwf+UTYLpPx22zVYXIDDeLqgJVQXro9IcU+sn31epQ 0gFVkY8X6IMmz2IrnBK041N5UJIBraWHjmh9XJaD8FhbbpGdcZywg55geS7+ fj1wLqnuE6kpPaEboQ0qYQ21Kl6ZjtfoY7WesR1MWa67tUJjW+m79BM05ugr 9WNxHMHj+mpf15YNoH1oPqyh59tgH32gZ6FfdE1Z01afxut1kjLsCT/Mdt1f nw+H0Qca26tKGv0Cp0vHFuNl8KXjVlJjGt1KLY3OX2oiPfcG6xnDpCLP1a72 G9c1OlrV1PosRGll02drXK1c1ezqbzU+3J739KCrRi1ORL/xojT3163/GpVQ jRs60Izxog+aRp56vB1vTIWtGlpiGuh/Ub3Ka/TJrqM0Ez/RYY7OJz7rUU7Y RL99jSc6UprGB6fbYvpvZVFzz+a7n60qOukcVXdVuwaXjkTjUNtqRNIPwZWL a8Sp0U66lHPiWTVj4yFvTBO+az07wPyVTaAAMwsF2LdBAXYmUIB9Ewqw5qEA 83eYvyKEEEIIaRHMXxFCCCGEeBjmrwghhBBCWgTzV4QQQgghHob5K0IIIYSQ FsH8FSGEEEKIh2H+ihBCCCGkRTB/RQghhBDiYZi/IoQQQghpEcxfEUIIIYR4 GOavCCGEEEJaBPNXhBBCCCEehvkrQgghhJAWwfwVIYQQQoiHYf6KEEIIIaRF MH9FCCGEEOJhmL8ihBBCCGkRzF8RQgghhHgY5q8IIYQQQloE81eEEEIIIR6m IX919tlnd+zYkfKJEEIIIeT0qF7q6OLM81dt8jmEEEIIIbbHEE5nn312586d X3rppaysLG//RYQQQgghPo3qpZdcZGRk1NfXt+5D9I36duNzKMAIIYQQQk6P IcA6d+780EMPde/ePS0traamxtt/FCGEEEKIj6JKSfVSdxeZmZlnkr/Stxuf QwFGCCGEEHIaGgTYQw89dPfddz/xxBOrV68+fPhwbW2tt/80QgghhBCfQzXS oUOHVq5c+dRTTz377LOZmZln8mn69meeeYYCjBBCCCHkNDQWYJ06derSpcsv f/nL+fPnb9u2rbq62tt/HSGEEEKIz6EaKTk5ee7cuc8//3y3bt3O8L4/ffvL L79MAUYIIYQQchoaC7CuXbt+/fXXPXv2DAoK6t+/f1FRkbf/OkIIIYQQn0M1 UkxMjOqlwMDA6OjoM5RMhYWFkZGRFGCEEEIIIaehsQBThgwZMmDAgK+++qp3 796pqanFxcVVVVVsxUAIIYQQoopIdZFqpyNHjvTq1atHjx4qnCZNmlRaWnom H6tvHzduHAUYIYQQQsg3aVaAxcbGql5KS0sLCwvTn4SGhqqU2rp1a0pKClsx EEIIIcSfMVouJCcnx8TEqHbq2bNn3759VTWVlJQ4nc4z+WR9u34IBRghhBBC SBNOJcCKi4v1Wf13yJAhqp2+/vrrkJCQxMTEVatWpaenZ2dnHyGEEEII8T9U BalSUkWkuiggIODLL79U7RQXF2dopzaBAowQQgghpDHfKsDq6+v1seqlQYMG ORyO55577sknn3z55Zeff/75jh073kAIIYQQ4k+o/nnxxRe7d++uikh1UWBg 4ODBg1UpqV5S1dRW+SsKMEIIIYSQBswLsNLS0gkTJkRFRb3yyiu/+MUv9F/K J0IIIYT4IQ3y6dlnn+3WrVt0dPSZ97w6FRRghBBCCCE3tESAGa0YCgsLs7Oz M4/j7bVjhBBCCCFeoEELqS5SdXTmPa9OBQUYIYQQQohB6wRYvYv20GmEEEII IT6Ot4QQBRghhBBC/BYKIUIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBBCCCGE EEIIIcTf+H/dwxca "], {{0, 330.}, {800., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{800., 330.}, PlotRange->{{0, 800.}, {0, 330.}}]], "PastedInput", CellChangeTimes->{ 3.759812843067431*^9, {3.759812887535778*^9, 3.75981293857047*^9}, 3.759812986730946*^9, 3.759813232589273*^9, {3.7645859872562532`*^9, 3.764586019074163*^9}}, CellLabel->"Out[51]="] }, Open ]], Cell[CellGroupData[{ Cell["Sentence analogies", "Subsection", CellChangeTimes->{{3.728629012318664*^9, 3.7286290160147057`*^9}, { 3.7634672641360903`*^9, 3.763467268782531*^9}, 3.764089413780208*^9}], Cell["\<\ Define a sentence embedding that takes the last feature vector from BERT \ subword embeddings (as an arbitrary choice):\ \>", "Text", CellChangeTimes->{{3.764589997489499*^9, 3.764590024921913*^9}, { 3.7645905794140873`*^9, 3.764590595494514*^9}, {3.764593871620772*^9, 3.7645938887088337`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"sentenceembedding", "=", RowBox[{"NetAppend", "[", RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], ",", RowBox[{"\"\\"", "\[Rule]", RowBox[{"SequenceLastLayer", "[", "]"}]}]}], "]"}]}]], "Input", CellChangeTimes->{{3.764590027180747*^9, 3.764590090058592*^9}, { 3.7645902192347403`*^9, 3.764590222156015*^9}, {3.764593776076727*^9, 3.7645938026689653`*^9}, {3.764593835225231*^9, 3.76459384076499*^9}, 3.797848643260078*^9}, CellLabel->"In[29]:="], Cell[BoxData[ TagBox[ TemplateBox[{ RowBox[{ StyleBox[ TagBox["NetChain", "SummaryHead"], "NonInterpretableSummary"], StyleBox["[", "NonInterpretableSummary"], DynamicModuleBox[{Typeset`open = False}, PanelBox[ PaneSelectorBox[{False -> GridBox[{{ PaneBox[ ButtonBox[ DynamicBox[ FEPrivate`FrontEndResource[ "FEBitmaps", "SquarePlusIconMedium"]], ButtonFunction :> (Typeset`open = True), Appearance -> None, Evaluator -> Automatic, Method -> "Preemptive"], Alignment -> {Center, Center}, ImageSize -> {Automatic, 24}], StyleBox[ GraphicsBox[{ Thickness[0.014925373134328358`], { RGBColor[0.729, 0.729, 0.729], Thickness[0.018686567164179105`], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{0, 2, 0}}}, {{{23., 91.5}, {30., 91.5}}}, CurveClosed -> {0}]}, { RGBColor[0.729, 0.729, 0.729], Thickness[0.018686567164179105`], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{0, 2, 0}}}, {{{37., 91.5}, {44., 91.5}}}, CurveClosed -> {0}]}, { FaceForm[{ RGBColor[0.631, 0.8160000000000001, 0.498], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{15.5, 103.5}, {23.5, 103.5}, {23.5, 79.5}, {15.5, 79.5}}}]}, { FaceForm[{ RGBColor[0.40800000000000003`, 0.627, 0.196], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{24., 104.}, {15., 104.}, { 15., 79.}, {24., 79.}}, {{23., 80.}, {16., 80.}, {16., 103.}, {23., 103.}}}]}, { FaceForm[{ RGBColor[0.38, 0.753, 0.89], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{29.5, 103.5}, {37.5, 103.5}, {37.5, 79.5}, {29.5, 79.5}}}]}, { FaceForm[{ RGBColor[0.24300000000000002`, 0.608, 0.71], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{38., 104.}, {29., 104.}, { 29., 79.}, {38., 79.}}, {{37., 80.}, {30., 80.}, {30., 103.}, {37., 103.}}}]}, { FaceForm[{ RGBColor[0.584, 0.51, 0.8390000000000001], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{43.5, 103.5}, {51.5, 103.5}, {51.5, 79.5}, {43.5, 79.5}}}]}, { FaceForm[{ RGBColor[0.4, 0.396, 0.659], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{52., 104.}, {43., 104.}, { 43., 79.}, {52., 79.}}, {{51., 80.}, {44., 80.}, {44., 103.}, {51., 103.}}}]}, { FaceForm[{ RGBColor[0.549, 0.549, 0.549], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}}}, {{{28., 91.5}, { 25., 95.}, {25., 88.}}}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}}}, {{{42., 91.5}, { 39., 95.}, {39., 88.}}}]}}, ImageMargins -> 0, ImageSize -> {36, Automatic}, PlotRangePadding -> 0], CacheGraphics -> False], GridBox[{{ StyleBox[ TemplateBox[{ TemplateBox[{ StyleBox["\"Input\"", GrayLevel[0], StripOnInput -> False], "\" \"", "\"port\""}, "RowDefault"], "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox["\"string\"", "SummaryItem"]}, { StyleBox[ TemplateBox[{ TemplateBox[{ StyleBox["\"Output\"", GrayLevel[0], StripOnInput -> False], "\" \"", "\"port\""}, "RowDefault"], "\":\""}, "RowDefault"], "SummaryItemAnnotation"], StyleBox[ TemplateBox[{"\"vector\"", "\" \"", StyleBox[ TemplateBox[{ "\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", "\"768\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], "SummaryItem"]}}, BaselinePosition -> Automatic, GridBoxAlignment -> { "Columns" -> {{Left}}, "Rows" -> {{Automatic}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> { "Columns" -> {{2}}, "Rows" -> {{Automatic}}}]}}, GridBoxAlignment -> {"Rows" -> {{Top}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridFrameMargins -> {{0, 0}, {0, 0}}, BaselinePosition -> {1, 3}], True -> GridBox[{{ PaneBox[ ButtonBox[ DynamicBox[ FEPrivate`FrontEndResource[ "FEBitmaps", "SquareMinusIconMedium"]], ButtonFunction :> (Typeset`open = False), Appearance -> None, Evaluator -> Automatic, Method -> "Preemptive"], Alignment -> {Center, Center}, ImageSize -> {Automatic, 24}], StyleBox[ GraphicsBox[{ Thickness[0.014925373134328358`], { RGBColor[0.729, 0.729, 0.729], Thickness[0.018686567164179105`], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{0, 2, 0}}}, {{{23., 91.5}, {30., 91.5}}}, CurveClosed -> {0}]}, { RGBColor[0.729, 0.729, 0.729], Thickness[0.018686567164179105`], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{0, 2, 0}}}, {{{37., 91.5}, {44., 91.5}}}, CurveClosed -> {0}]}, { FaceForm[{ RGBColor[0.631, 0.8160000000000001, 0.498], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{15.5, 103.5}, {23.5, 103.5}, {23.5, 79.5}, {15.5, 79.5}}}]}, { FaceForm[{ RGBColor[0.40800000000000003`, 0.627, 0.196], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{24., 104.}, {15., 104.}, { 15., 79.}, {24., 79.}}, {{23., 80.}, {16., 80.}, {16., 103.}, {23., 103.}}}]}, { FaceForm[{ RGBColor[0.38, 0.753, 0.89], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{29.5, 103.5}, {37.5, 103.5}, {37.5, 79.5}, {29.5, 79.5}}}]}, { FaceForm[{ RGBColor[0.24300000000000002`, 0.608, 0.71], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{38., 104.}, {29., 104.}, { 29., 79.}, {38., 79.}}, {{37., 80.}, {30., 80.}, {30., 103.}, {37., 103.}}}]}, { FaceForm[{ RGBColor[0.584, 0.51, 0.8390000000000001], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{43.5, 103.5}, {51.5, 103.5}, {51.5, 79.5}, {43.5, 79.5}}}]}, { FaceForm[{ RGBColor[0.4, 0.396, 0.659], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{52., 104.}, {43., 104.}, { 43., 79.}, {52., 79.}}, {{51., 80.}, {44., 80.}, {44., 103.}, {51., 103.}}}]}, { FaceForm[{ RGBColor[0.549, 0.549, 0.549], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}}}, {{{28., 91.5}, { 25., 95.}, {25., 88.}}}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}}}, {{{42., 91.5}, { 39., 95.}, {39., 88.}}}]}}, ImageMargins -> 0, ImageSize -> {36, Automatic}, PlotRangePadding -> 0], CacheGraphics -> False], GridBox[{{ DynamicModuleBox[{ NeuralNetworks`Private`NetChain`assoc3 = Association[ "Type" -> "Chain", "Nodes" -> Association[ "embedding" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 2}, NeuralNetworks`IndexIntegerT[30522]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association[ "part1" -> Association[ "Type" -> "Part", "Arrays" -> Association[], "Parameters" -> Association[ "Specification" -> NeuralNetworks`ValidatedParameter[{ Span[1, -1], 1}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 2}, NeuralNetworks`IndexIntegerT[30522]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[30522]]]], "embeddingwords" -> Association[ "Type" -> "Embedding", "Arrays" -> Association[ "Weights" -> NeuralNetworks`Private`DummyArray[{30522, 768}]], "Parameters" -> Association[ "OutputDimension" -> 768, "ClassCount" -> 30522, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939]}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[30522]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "posembed" -> Association[ "Type" -> "SequenceIndices", "Arrays" -> Association[], "Parameters" -> Association[ "Maximum" -> 512, "$Length" -> NeuralNetworks`LengthVar[2137958939]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[30522]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[512]]]], "embeddingpos" -> Association[ "Type" -> "Embedding", "Arrays" -> Association[ "Weights" -> NeuralNetworks`Private`DummyArray[{512, 768}]], "Parameters" -> Association[ "OutputDimension" -> 768, "ClassCount" -> 512, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939]}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[512]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "part2" -> Association[ "Type" -> "Part", "Arrays" -> Association[], "Parameters" -> Association[ "Specification" -> NeuralNetworks`ValidatedParameter[{ Span[1, -1], 2}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 2}, NeuralNetworks`IndexIntegerT[30522]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[30522]]]], "Clip" -> Association[ "Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association[ "Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Min, 2., NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939]}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[30522]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[2]]]], "embeddingtokens" -> Association[ "Type" -> "Embedding", "Arrays" -> Association[ "Weights" -> NeuralNetworks`Private`DummyArray[{2, 768}]], "Parameters" -> Association[ "OutputDimension" -> 768, "ClassCount" -> 2, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939]}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939]}, NeuralNetworks`IndexIntegerT[2]]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "inputCombine" -> Association[ "Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association[ "Function" -> NeuralNetworks`ValidatedParameter[#3 + (# + #2)& ]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "3" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "normalize" -> Association[ "Type" -> "Normalization", "Arrays" -> Association[ "Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association[ "AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association[ "Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association[ "DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> { NeuralNetworks`NetPath[ "Nodes", "part1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "part2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "embeddingwords", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "part1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "posembed", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "part1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "inputCombine", "Inputs", "1"] -> NeuralNetworks`NetPath[ "Nodes", "embeddingwords", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "embeddingpos", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "posembed", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "inputCombine", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "embeddingpos", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "Clip", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "part2", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "embeddingtokens", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "Clip", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "inputCombine", "Inputs", "3"] -> NeuralNetworks`NetPath[ "Nodes", "embeddingtokens", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "normalize", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "inputCombine", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "normalize", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"]}], "encoder" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association[ "1" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "2" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "3" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "4" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "5" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "6" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "7" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "8" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "9" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "10" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "11" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "12" -> Association[ "Type" -> "Chain", "Nodes" -> Association[ "1" -> Association[ "Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["attention" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "3" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[2], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> {Times, 0.125, NeuralNetworks`Private`ScalarSymbol[1]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "4" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {12, 64}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "5" -> Association["Type" -> "Attention", "Arrays" -> Association[], "Parameters" -> Association["ScoringNet" -> Association["Type" -> "Graph", "Inputs" -> Association["Query" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "Input" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]], "Nodes" -> Association["1" -> Association[ "Type" -> "Dot", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association[ "1" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "1", "Inputs", "2"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"]}], "Mask" -> None, "ScoreRescaling" -> None, "MultiHead" -> True, "$InputPorts" -> "KeyValueQuery", "$KeyAndValueShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryShape" -> { NeuralNetworks`LengthVar[2137958939], 12}, "$QueryChannels" -> {64}, "$KeyChannels" -> {64}, "$ValueChannels" -> {64}], "Inputs" -> Association["Key" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Value" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT], "Query" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]]], "6" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association["Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 768, "$InputDimensions" -> {12, 64}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{12, 64}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 12, 64}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Key"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Query"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Query"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Value"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "6", "Outputs", "Output"]}], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "attention", "Inputs", "Query"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "attention", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}], "2" -> Association["Type" -> "Graph", "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Nodes" -> Association["linear1" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{3072, 768}], "Biases" -> NeuralNetworks`Private`DummyArray[{3072}]], "Parameters" -> Association["OutputDimensions" -> {3072}, "$OutputSize" -> 3072, "$InputSize" -> 768, "$InputDimensions" -> {768}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "gelu" -> Association["Type" -> "Elementwise", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[ NeuralNetworks`Private`ScalarFunctionObject[{ NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[6], Association[ NeuralNetworks`Private`ScalarSymbol[2] -> { Times, 0.7071067811865475, NeuralNetworks`Private`ScalarSymbol[1]}, NeuralNetworks`Private`ScalarSymbol[3] -> {Erf, NeuralNetworks`Private`ScalarSymbol[2]}, NeuralNetworks`Private`ScalarSymbol[4] -> {Plus, 1., NeuralNetworks`Private`ScalarSymbol[3]}, NeuralNetworks`Private`ScalarSymbol[5] -> {Times, NeuralNetworks`Private`ScalarSymbol[1], NeuralNetworks`Private`ScalarSymbol[4]}, NeuralNetworks`Private`ScalarSymbol[6] -> {Times, 0.5, NeuralNetworks`Private`ScalarSymbol[5]}]]], "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 3072}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]]], "linear2" -> Association["Type" -> "NetMap", "Arrays" -> Association[], "Parameters" -> Association[ "Net" -> Association[ "Type" -> "Linear", "Arrays" -> Association["Weights" -> NeuralNetworks`Private`DummyArray[{768, 3072}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["OutputDimensions" -> {768}, "$OutputSize" -> 768, "$InputSize" -> 3072, "$InputDimensions" -> {3072}], "Inputs" -> Association[ "Input" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], "$SequenceLength" -> NeuralNetworks`LengthVar[2137958939], "$InputShape" -> NeuralNetworks`TensorT[{3072}, NeuralNetworks`RealT], "$OutputShape" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 3072}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "dropout" -> Association["Type" -> "Dropout", "Arrays" -> Association[], "Parameters" -> Association["DropoutProbability" -> 0.1, "Method" -> "Dropout", "OutputPorts" -> NeuralNetworks`ValidatedParameter[{"Output"}]], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "add" -> Association["Type" -> "Threading", "Arrays" -> Association[], "Parameters" -> Association["Function" -> NeuralNetworks`ValidatedParameter[Plus]], "Inputs" -> Association["1" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT], "2" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]], "norm" -> Association["Type" -> "Normalization", "Arrays" -> Association["Scaling" -> NeuralNetworks`Private`DummyArray[{768}], "Biases" -> NeuralNetworks`Private`DummyArray[{768}]], "Parameters" -> Association["AggregationLevels" -> NeuralNetworks`ValidatedParameter[ Span[2, All]], "ScalingLevels" -> NeuralNetworks`ValidatedParameter["Same"], "Epsilon" -> 1.*^-12, "$Dimensions" -> { NeuralNetworks`LengthVar[2137958939], 768}, "$StatsDimensions" -> {768}], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> {NeuralNetworks`NetPath[ "Nodes", "add", "Inputs", "1"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "linear1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "gelu", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear1", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "linear2", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "gelu", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "dropout", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "linear2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "add", "Inputs", "2"] -> NeuralNetworks`NetPath[ "Nodes", "dropout", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "norm", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "add", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "norm", "Outputs", "Output"]}]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "2", "Outputs", "Output"]}, "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association["Output" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]]]], "Edges" -> { NeuralNetworks`NetPath["Nodes", "1", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath["Nodes", "2", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "1", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "3", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "2", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "4", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "3", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "5", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "4", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "6", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "5", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "7", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "6", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "8", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "7", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "9", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "8", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "10", "Inputs", "Input"] -> NeuralNetworks`NetPath["Nodes", "9", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "11", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "10", "Outputs", "Output"], NeuralNetworks`NetPath["Nodes", "12", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "11", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "12", "Outputs", "Output"]}], "pooling" -> Association[ "Type" -> "SequenceLast", "Arrays" -> Association[], "Parameters" -> Association[], "Inputs" -> Association["Input" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 768}, NeuralNetworks`RealT]], "Outputs" -> Association[ "Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]]], "Edges" -> { NeuralNetworks`NetPath[ "Nodes", "embedding", "Inputs", "Input"] -> NeuralNetworks`NetPath["Inputs", "Input"], NeuralNetworks`NetPath[ "Nodes", "encoder", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "embedding", "Outputs", "Output"], NeuralNetworks`NetPath[ "Nodes", "pooling", "Inputs", "Input"] -> NeuralNetworks`NetPath[ "Nodes", "encoder", "Outputs", "Output"], NeuralNetworks`NetPath["Outputs", "Output"] -> NeuralNetworks`NetPath[ "Nodes", "pooling", "Outputs", "Output"]}, "Inputs" -> Association[ "Input" -> Association["Form" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[2137958939], 2}, NeuralNetworks`IndexIntegerT[30522]], "Type" -> "Function", "Function" -> NeuralNetworks`ValidatedParameter[ Composition[ Row[{Function, "[", Row[{ Style["x", Italic], ",", "\[Ellipsis]"}], "]"}], Map[List]]], "OutputType" -> NeuralNetworks`TensorT[{ NeuralNetworks`LengthVar[1051709669], 2}, NeuralNetworks`IndexIntegerT[30522]], "Pattern" -> NeuralNetworks`ValidatedParameter[ Blank[String]], "Batched" -> True, "$Version" -> "12.1.4"]], "Outputs" -> Association[ "Output" -> NeuralNetworks`TensorT[{768}, NeuralNetworks`RealT]]], NeuralNetworks`Private`NetChain`opart, NeuralNetworks`Private`NetChain`part, NeuralNetworks`Private`NetChain`selected = Null}, DynamicBox[ GridBox[{{ TagBox[ TagBox[ GridBox[{{ TagBox[ TagBox[ "\"\"", Annotation[#, {"Inputs", "Input"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ "\"Input\"", Annotation[#, {"Inputs", "Input"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ TagBox[ GridBox[{{"\"string\""}, { TemplateBox[{ TemplateBox[{"\"matrix\"", "\" \"", StyleBox[ TemplateBox[{"\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"2\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], "\" of \"", TemplateBox[{"\"indices\"", "\" \"", StyleBox[ TemplateBox[{"\"(\"", "\"\[VeryThinSpace]\"", "\"range\"", "\":\"", "\" \"", TemplateBox[{"1", "\"..\"", "30522"}, "RowDefault"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"]}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, BaselinePosition -> 2, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"], Annotation[#, {"Inputs", "Input"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]]}, { TagBox[ TagBox[ StyleBox["\"embedding\"", GrayLevel[0.5], StripOnInput -> False], Annotation[#, {"Nodes", "embedding"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ StyleBox[ TemplateBox[{"NetGraph", "\" \"", TemplateBox[{"\"(\"", "10", "\" nodes)\""}, "Row", DisplayFunction -> (RowBox[{ TemplateSlotSequence[1, "\[InvisibleSpace]"]}]& ), InterpretationFunction -> (RowBox[{"Row", "[", RowBox[{ RowBox[{"{", TemplateSlotSequence[1, ","], "}"}], ",", RowBox[{"BaseStyle", "\[Rule]", InterpretationBox[ ButtonBox[ TooltipBox[ GraphicsBox[{{ GrayLevel[0], RectangleBox[{0, 0}]}, { GrayLevel[0], RectangleBox[{1, -1}]}, { GrayLevel[0.5], RectangleBox[{0, -1}, {2, 1}]}}, DefaultBaseStyle -> "ColorSwatchGraphics", AspectRatio -> 1, Frame -> True, FrameStyle -> GrayLevel[0.33333333333333337`], FrameTicks -> None, PlotRangePadding -> None, ImageSize -> Dynamic[{Automatic, 1.35 (CurrentValue["FontCapHeight"]/ AbsoluteCurrentValue[Magnification])}]], StyleBox[ RowBox[{"GrayLevel", "[", "0.5`", "]"}], NumberMarks -> False]], Appearance -> None, BaseStyle -> {}, BaselinePosition -> Baseline, DefaultBaseStyle -> {}, ButtonFunction :> With[{Typeset`box$ = EvaluationBox[]}, If[ Not[ AbsoluteCurrentValue["Deployed"]], SelectionMove[Typeset`box$, All, Expression]; FrontEnd`Private`$ColorSelectorInitialAlpha = 1; FrontEnd`Private`$ColorSelectorInitialColor = GrayLevel[0.5]; FrontEnd`Private`$ColorSelectorUseMakeBoxes = True; MathLink`CallFrontEnd[ FrontEnd`AttachCell[Typeset`box$, FrontEndResource["GrayLevelColorValueSelector"], { 0, {Left, Bottom}}, {Left, Top}, "ClosingActions" -> {"SelectionDeparture", "ParentChanged", "EvaluatorQuit"}]]]], BaseStyle -> Inherited, Evaluator -> Automatic, Method -> "Preemptive"], GrayLevel[0.5], Editable -> False, Selectable -> False]}]}], "]"}]& ), BaseStyle -> GrayLevel[0.5]]}, "RowDefault"], GrayLevel[0], StripOnInput -> False], Annotation[#, {"Nodes", "embedding"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ TemplateBox[{"\"matrix\"", "\" \"", StyleBox[ TemplateBox[{"\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"768\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], Annotation[#, {"Nodes", "embedding"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]]}, { TagBox[ TagBox[ StyleBox["\"encoder\"", GrayLevel[0.5], StripOnInput -> False], Annotation[#, {"Nodes", "encoder"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ StyleBox[ TemplateBox[{"NetGraph", "\" \"", TemplateBox[{"\"(\"", "12", "\" nodes)\""}, "Row", DisplayFunction -> (RowBox[{ TemplateSlotSequence[1, "\[InvisibleSpace]"]}]& ), InterpretationFunction -> (RowBox[{"Row", "[", RowBox[{ RowBox[{"{", TemplateSlotSequence[1, ","], "}"}], ",", RowBox[{"BaseStyle", "\[Rule]", InterpretationBox[ ButtonBox[ TooltipBox[ GraphicsBox[{{ GrayLevel[0], RectangleBox[{0, 0}]}, { GrayLevel[0], RectangleBox[{1, -1}]}, { GrayLevel[0.5], RectangleBox[{0, -1}, {2, 1}]}}, DefaultBaseStyle -> "ColorSwatchGraphics", AspectRatio -> 1, Frame -> True, FrameStyle -> GrayLevel[0.33333333333333337`], FrameTicks -> None, PlotRangePadding -> None, ImageSize -> Dynamic[{Automatic, 1.35 (CurrentValue["FontCapHeight"]/ AbsoluteCurrentValue[Magnification])}]], StyleBox[ RowBox[{"GrayLevel", "[", "0.5`", "]"}], NumberMarks -> False]], Appearance -> None, BaseStyle -> {}, BaselinePosition -> Baseline, DefaultBaseStyle -> {}, ButtonFunction :> With[{Typeset`box$ = EvaluationBox[]}, If[ Not[ AbsoluteCurrentValue["Deployed"]], SelectionMove[Typeset`box$, All, Expression]; FrontEnd`Private`$ColorSelectorInitialAlpha = 1; FrontEnd`Private`$ColorSelectorInitialColor = GrayLevel[0.5]; FrontEnd`Private`$ColorSelectorUseMakeBoxes = True; MathLink`CallFrontEnd[ FrontEnd`AttachCell[Typeset`box$, FrontEndResource["GrayLevelColorValueSelector"], { 0, {Left, Bottom}}, {Left, Top}, "ClosingActions" -> {"SelectionDeparture", "ParentChanged", "EvaluatorQuit"}]]]], BaseStyle -> Inherited, Evaluator -> Automatic, Method -> "Preemptive"], GrayLevel[0.5], Editable -> False, Selectable -> False]}]}], "]"}]& ), BaseStyle -> GrayLevel[0.5]]}, "RowDefault"], GrayLevel[0], StripOnInput -> False], Annotation[#, {"Nodes", "encoder"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ TemplateBox[{"\"matrix\"", "\" \"", StyleBox[ TemplateBox[{"\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"768\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], Annotation[#, {"Nodes", "encoder"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]]}, { TagBox[ TagBox[ StyleBox["\"pooling\"", GrayLevel[0.5], StripOnInput -> False], Annotation[#, {"Nodes", "pooling"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ StyleBox["SequenceLastLayer", GrayLevel[0], StripOnInput -> False], Annotation[#, {"Nodes", "pooling"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ TemplateBox[{"\"vector\"", "\" \"", StyleBox[ TemplateBox[{"\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", "\"768\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], Annotation[#, {"Nodes", "pooling"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]]}, { TagBox[ TagBox[ "\"\"", Annotation[#, {"Outputs", "Output"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ "\"Output\"", Annotation[#, {"Outputs", "Output"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]], TagBox[ TagBox[ TemplateBox[{"\"vector\"", "\" \"", StyleBox[ TemplateBox[{"\"(\"", "\"\[VeryThinSpace]\"", "\"size\"", "\":\"", "\" \"", TemplateBox[{"\[Times]", "\"\[Times]\"", "\"768\""}, "RowWithSeparators"], "\"\[VeryThinSpace]\"", "\")\""}, "RowDefault"], GrayLevel[0.5], StripOnInput -> False]}, "RowDefault"], Annotation[#, {"Outputs", "Output"}, "Mouse"]& ], MouseAppearanceTag["LinkHand"]]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, AutoDelete -> False, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> {"Columns" -> {{1.1}}}], "Grid"], EventHandlerTag[{"MouseClicked" :> If[ ListQ[ NeuralNetworks`Private`NetChain`part = MouseAnnotation[]], If[NeuralNetworks`Private`NetChain`opart === NeuralNetworks`Private`NetChain`part, NeuralNetworks`Private`NetChain`selected = Null; NeuralNetworks`Private`NetChain`opart = Null, NeuralNetworks`Private`NetChain`selected = Part[NeuralNetworks`Private`NetChain`assoc3, Apply[Sequence, NeuralNetworks`Private`NetChain`part]]; NeuralNetworks`Private`NetChain`opart = NeuralNetworks`Private`NetChain`part; Null]; Null], Method -> "Preemptive", PassEventsDown -> Automatic, PassEventsUp -> True}]]}, NeuralNetworks`FormatSelectedParameterChain[ NeuralNetworks`Private`NetChain`selected, NeuralNetworks`Private`NetChain`part, Association[], {}]}, GridBoxSpacings -> {"Columns" -> {{1}}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, GridFrameMargins -> {{0, 0}, {0, 0}}], TrackedSymbols :> { NeuralNetworks`Private`NetChain`selected}], Initialization :> {NetChain}]}}, BaselinePosition -> Automatic, GridBoxAlignment -> { "Columns" -> {{Left}}, "Rows" -> {{Automatic}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings -> { "Columns" -> {{2}}, "Rows" -> {{Automatic}}}]}}, GridBoxAlignment -> {"Rows" -> {{Top}}}, GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridFrameMargins -> {{0, 0}, {0, 0}}, BaselinePosition -> {1, 1}]}, Dynamic[Typeset`open], ImageSize -> Automatic], BaselinePosition -> Baseline, BaseStyle -> { ShowStringCharacters -> False, NumberMarks -> False, PrintPrecision -> 3, ShowSyntaxStyles -> False}]], StyleBox["]", "NonInterpretableSummary"]}]}, "CopyTag", DisplayFunction->(#& ), InterpretationFunction->("NetChain[<>]"& )], False, Editable->False, SelectWithContents->True, Selectable->False]], "Output", CellChangeTimes->{{3.7645900646178017`*^9, 3.764590091811685*^9}, 3.7645902240562162`*^9, {3.7645937859310417`*^9, 3.764593804691498*^9}, 3.764593842886517*^9, 3.772297913801148*^9, 3.772297992553561*^9, 3.798186156173333*^9, 3.798186225368273*^9}, CellLabel->"Out[29]="] }, Open ]], Cell["\<\ Define a list of sentences in two broad categories (food and music):\ \>", "Text", CellChangeTimes->{{3.765212068907337*^9, 3.765212086313847*^9}, { 3.773419716900482*^9, 3.773419717243661*^9}, 3.7997698231697083`*^9}], Cell[BoxData[ RowBox[{ RowBox[{"sentences", "=", RowBox[{"{", RowBox[{ "\"\\"", ",", "\"\\"", ",", "\"\\"", ",", "\"\\"", ",", "\"\\"", ",", "\"\\""}], "}"}]}], ";"}]], "Input", CellChangeTimes->{{3.7634778365208073`*^9, 3.763477879907297*^9}, { 3.763477912848937*^9, 3.7634779304226923`*^9}, {3.7634779647837133`*^9, 3.7634779746824207`*^9}, {3.763478015699883*^9, 3.7634780853638487`*^9}, { 3.763479572500502*^9, 3.763479637653124*^9}, {3.763482619434083*^9, 3.7634828944543667`*^9}, {3.763482947903274*^9, 3.763483028512207*^9}, { 3.763483059824832*^9, 3.763483142837903*^9}, {3.7722978954153557`*^9, 3.772297897220112*^9}, {3.772297931550557*^9, 3.772297948500822*^9}, { 3.7981862280464697`*^9, 3.798186336583632*^9}}, CellLabel->"In[45]:="], Cell["Precompute the embeddings for a list of sentences:", "Text", CellChangeTimes->{ 3.724173319401992*^9, {3.724173357185844*^9, 3.7241733934539223`*^9}, { 3.724173571560605*^9, 3.724173573334279*^9}, {3.724174457986969*^9, 3.724174460679566*^9}, {3.724769976037303*^9, 3.724769978244512*^9}, { 3.732555807175282*^9, 3.7325558179403677`*^9}, {3.763741315741302*^9, 3.763741326381304*^9}, 3.7638085768718843`*^9, 3.76459011388244*^9}], Cell[BoxData[ RowBox[{ RowBox[{"assoc", "=", RowBox[{"AssociationThread", "[", RowBox[{"sentences", "\[Rule]", RowBox[{"sentenceembedding", "[", "sentences", "]"}]}], "]"}]}], ";"}]], "Input", CellChangeTimes->{{3.763479030961587*^9, 3.763479035623824*^9}, { 3.763479105413314*^9, 3.7634791069068823`*^9}, 3.76347958454652*^9, { 3.7638085885699673`*^9, 3.76380859003957*^9}, {3.764005815321741*^9, 3.764005822252388*^9}, {3.764590101005357*^9, 3.7645901414109507`*^9}}, CellLabel->"In[46]:="], Cell["\<\ Visualize the similarity between the sentences using the net as a feature \ extractor:\ \>", "Text", CellChangeTimes->{{3.728647979692671*^9, 3.728647995358663*^9}, { 3.728648334421508*^9, 3.7286483687652073`*^9}, {3.763386273780385*^9, 3.7633862770603848`*^9}, {3.763741877476347*^9, 3.763741877923944*^9}, { 3.763742359654976*^9, 3.7637423684226103`*^9}, {3.763742415097239*^9, 3.763742416713023*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"FeatureSpacePlot", "[", RowBox[{"assoc", ",", RowBox[{"LabelingFunction", "\[Rule]", "Callout"}]}], "]"}]], "Input", CellChangeTimes->{{3.7645901638066387`*^9, 3.764590180803294*^9}}, CellLabel->"In[47]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzs3Qm8TfX+//FdylCGEg1KqVDJDRlCxnKUFEURucTPcKhr7KpbMqauDJmK hFSGIqGIMiUyzzqG40SZxxAZz7H/7//+PM56rPbsOMdxT6/nw+2x9nd911rf 9V377s9nrfVd69zZrF2dlld6PJ5OWfWfOk1frdqxY9P/1L1OH55r26l1q7Yt mtdo+0qLVi06lm2WSYVrr7T//f9pLwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAQEaUlJT0xRdfdO7c+TEfTYwbN+7MmTPuOoMGDerU qdOePXvSq5EpM2TIEDV79+7d6d2QkNavX68WqsPtY79+/fTx0KFD6dsq718b 5tdIAAAulGJx5cqVPQHKlSvnzi5KliypwtWrV6djU1OgTJkyavbKlSvTuyEh KdNTC59//nn7eN999+nj5s2b07dV3r82zK+RAABcqOeee06hpECBAmPGjFm3 bp3OZD/99NNChQqpULHPqUa+kUbINwAAGd6ZM2dy5MihUDJr1ix3+ZIlS+wq x4YNG6yEfCONkG8AADK8HTt2WF6xc+dOv1n169cvXbr0V199ZR8t31Ae0rt3 78cee6xs2bLt2rVbtWqVe5GkpKRhw4ZpweLFiz/99NN9+/b1GwQS1Nq1a996 661HH330oYceevHFF9esWeOeq48tWrQo7xMbG7t06VK/xU+fPj1gwIDatWtr o88888y777576tQpZ65fvqHdbNWqlVb4zTffWINHjx6tpbRsTEyM9mjjxo1h 2qkFp0+f3q9fP622ZcuWzu5v2rSpbdu2lStXLleunBq5YsUKv2XVjSqvUqVK 0aJFq1Wr1qdPHzXbZoXKNxISErQ5NclvVSNGjFD54sWL/cq//vprlU+YMMEp UR2VdO/e3Sk5dOiQSt588037GP54kW8AAFJRwYIFFUoU7gOjpJvlGzfccIP+ e8UVV1iWkjNnTmepo0ePVqpUycrz5s1rdRTI9u3bF2a1P/zwQ5YsWVQzW7Zs mTNn1kSmTJmciyr9+/e/8sorVZg1a9ZrrrnG5r799tvO4gcOHLBbP5I7d26b UMjevn27VXDnG7t377bKTzzxhOUk7du3t0WyZ8+uNVszZs6cGbSpyhlU4c47 73SGuIwfP17lY8eO1VL6qBbaxaKrrrpqyJAhzoJ16tSx+toLzbLphx9+2OaG yjfUwuuvv17TP/30k7Mq5QPaTfXtb7/95te8GTNmqHKpUqWcktatW1tj/vjj Dyv59NNPVdK4ceNojhf5BgAgFU2dOtXCpdxyyy0KRopKgc+hWL6hoKzz6+PH jyvNsFDerFkzq9ChQwd91Pl7fHy8Pu7fv18n8ipp0qRJqE1v3bo1V65cinQD BgxITExUBLQEICYmxus7Pc/kM3jw4HPnzqnChx9+ePXVV6u+c5VDmYPqly5d OiEhQR9//fXX8uXLW0ZhFZx8Y+/evffcc4+mdSJvZ/HanFaubEcb0pn+77// 3rRpU0u9grbW8g155JFHJk6c2KVLl5MnTyqHUa6iwj59+qiF58+fnzBhgvpT 7dyyZYs3+c6UUqk5c+acPXv2yJEjo0aNsvVYqhbmfkqbNm00rbTBacPkyZOt AYHNU35y7bXXKj07ePCgldx99922oenTp1tJvXr19NGuWUU8XuQbAIDUpfBd sWJFu5LgePLJJ91PZVq+8eqrrzolFjcV3zWtMKoI68Qys337drsYEuop2kGD BqlC3bp1nRIFzfvvv18n6du2bWvRooXmNm/e3L1Iu3btLGfQ9K5duzSt7aqy U0EJgF0JWbt2rTc539C5v8VxBVylLlZTMV0lefLkce4laX+7d+8e6qlPyzey Zs3q7pZXXnlFhQ888IC75osvvqjC2NhYTSv9UALTq1cvdwXVV4Wvv/7aGzbf UEJi15SUqNjc2rVrq0QJYdAW2oUUu6WiPtG0HdOOHTuqRDuu7E650J9//hnN 8SLfAACkBZ0Xjx07tlmzZrfffrulHHfccYedpHuT8w13ePrll1+sjqaXLVtm iwwePPgDFwU4Fc6bNy/oFi0ua6NB5957772aO3/+fHehheBChQpp+vPPP9d0 5cqV/RasWbOmyi1tsHxDSYLHd/fn5MmT7poFChSwuVpk6NChzl2YoCzfKFq0 qLuwRo0adkHGvdcNGzZUYaVKldw1lQipH4YPH648ym79TJw40RtpvKg2p4/T pk3z+tIhJQk5cuRQwhC0hWPGjFFl9aqmtSFNd+3a1e6SqERb13StWrW80R0v 8g0AQFqbOnXqLbfc4pwae5PzDfeDHnZ5QcmJpsePH+8JbdSoUUG3Yuv0yygc NnrBSXjMvn377JqGN/nySGAQjI2NVXnv3r29yfmGJ3nMSZcuXdw1ExISihUr 5m5qqVKlfv7556DtsXzjqaeechcWLlw41F7nz5/f6owcOdI96kMtscsO0eQb /fv318fnnntO08qINN20adOgzfP6MsZMmTLly5dP03Xr1lXlbdu2FSxYUFtU rtKpUyeVjB492hvd8SLfAACkFqUTOqXt06dP4Kxx48YpxBQpUsQ+Bj4P6843 pk+frmmFtjnBqGbQrVetWtWTfPLuOH/+vE3Y2E6/R1k3btzo8Q0NdVr45JNP +q3WRinozN2bnG80b9581qxZHt/gSb9nakQJhnqgSpUqNpgz8DKIsXzDffdH lJ+osGfPnoF7vXDhQlX4+OOPLYg/+uijXbt2nTx58oEDB2JiYqLMN/bv369W Zc2a9fjx42XLltWsBQsWBO1MU7FiRY/vXpKyNSU53uTs68svv7znnnuUjdjo jmiOF/kGACC16GzX47venpiY6DfLormisH0Mn2/Ex8drOkeOHCdOnHAqKHPo 3bt3v379Qt2nsBEaCtbuwgoVKiidUBJiY0Hff//9wAaXKFFC08uXL9f0rbfe 6n6KUzty1113OWmM+/kUu31TtGhRq680o1evXp988omz7LZt26699lrV+f77 7wNbGzTfsFsnfrughnXp0sXSCUsA/Crkz5/fk/x4S8T3b9SqVUslgwYNuuKK K7RrTj4WVN++fVXZWmVDX5RpaNrGgjq3eKI5XuQbAICIzp0799tvv61bt06x b8mSJYuTLV269NChQ86YSZ0+27X9qlWrul/B8dNPP91xxx0qHzBggJWEzzeS kpJspEGbNm2cgKhUQSX58uU7evRo0EYqJVCFG2+80RlQqhbaHYe9e/eOGDFC 07fddptzeWTfvn2WS1j4Pn36tN2n6NGjh7NOC7jOIAd3vnH48OGbbrpJH19/ /XWv74UVHt+TsM4DHWqnDWBwHsh1C5pvWCDOmTOnc99HEdxapZZ4k0eh2LR5 55137IrHmDFjvFHkG1OmTPH4ngzy29Og1Azntsjnn39ue+2MBHaOZjTHi3wD ABCGwofOXhWkJgajs90dO3a4Lwio0B7o8PiGHJQvX14h3rkF4KQK4fMNr+9N GhYTy5Yt++qrr8bExNjtialTp4ZpbfXq1VVHaUDbtm3r1Kljr+Do1KmT1xcT LVvImzfvyy+/rAo333yz3UBx2q+cwQZm6PxdWYSN3syWLducOXOsgt/7vrSz dldlxYoVWkmRIkU8vgdAGjVq1KJFCxso636FhVvQfEPs0sG1117bsmVLhW+7 dqGV2CWjf/3rX5aQNGvWTNmCVTb2IpGI+cbZs2eVklkaFn5Eq7EMR5UPHDhg JXbTx+MbzuFUi3i8yDcAAKEoRi9evNidXUyfPn2my3fffbd3717n+UqzatWq qlWr2qm9ufvuuzt37uxcBpHSpUt7kh8yNbt37/YkP59ili5d6h5+mTt3bvd5 fVCnTp1q166dPT9iqUK3bt2cmzsnTpxo3bq1vRDMrnXExsY6b+Y08+bNsxdr eHzjSCtUqDB79mxnro15cI/ZePrpp1Xyj3/8w+t7AlRphkVz23qDBg2OHz8e tKkKxJ7koZtuao/6ysnZLPlxLndobfYXaow6+Z133rHwbY8ST5o0yeO7A2L1 77//fn20d2I47Clg595WeGqM56+P6L722muegId2vZGOl7thfo0EAPydKdlY uHChZRozZszYtWuXSqJf/Pz581u3blUM+v333y+mGceOHVuxYkVCQoJfVhOG EoyNGzf+/PPPQd9/bnPVtjBrOHz4sDYa6kHR8NRLq1evjouLCxzEckEr0S6v XLlSLQnVPKU34UdfhNK4cWOFe/dQk1SUguMFAPjbcicbc+fOjeYPl+B/Qnx8 fNasWa+77rqgT80AAHDJuJONefPmue+D4H9X//79S5YsaY/M2OtEAABILyQb GdXAgQM9vneSN2rUKGU3YgAASC2///775MmTSTYyHmWSmzdvdv+5FgAA0tH+ /fsXLFhAsgEAAAAAAAAAAAAAAJB2EpOSlsXtGTcrTv80kXghL/ICAACI6OCR k28M+7FxjxnOP31UYXq3CwAAZBCJSUl+yYaTcnCVAwAApIplcXsCkw37p1np 3ToAAJARjJsVFyrf0Kz0bh0AAMgIyDcAAEBa434KAABIa4wXBQAAlwDPwwIA gEuA930BAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAuAwNHz68U6dOK1eudEo2bNjQoUOHihUrPvHEE0uWLEnHtl1imzZtUleM GTMmvRuSPtavX6/dHzduXHo3JG0dPHgwvZsAAH9HlStX9ng8H3/8sX08dOhQ vnz5PMmWLVuWrq27pL755hvtcq1atdK7Ienjiy++0O4///zz6d2QNNS3b9/s 2bOfP38+vRsCAH87fvnGzJkz9fHGG29cvHjxjh07zp07l66tu6TINzJ8vpE5 c2btI/kGAFx6fvnGmDFj9LFevXrp2qj0Qb5BvgEASCPufMOGbehj4cKFW7Ro 8cknnwTWT0hI0Kx27dr5lY8YMULlixcvto9r1qzRx/I+sbGxS5cudWquWrVK s0aOHOlefNiwYSpcv359mKZ+9dVXWlWVKlWKFi1arVq1Pn36nD592matXbtW i0+fPr1fv35lypRp2bKltqLypKQkrbl+/frFixd/+umn+/bte+bMmVDrd/IN 7UWzZs1KlixZp06dgQMHJiYmRtnyDRs2aHratGmLFi1q1apVqVKlatasqSb5 XSaaN2/eK6+8Urp06bp1644bN2727NlaauHChUF3ZOrUqdF0eETWNvWhVtig QQO1rX379nPnznUqBOYbEXtv79693bp1e/bZZ1VB3dWkSRP3QCAtPnr06Gee eUZzY2JitAsbN250L35BR8ds2rSpbdu2+tKWK1dOX4YVK1b4VQj1Jdm8ebN2 3+4SauLdd9+Nst8AAKnCnW/kyZPHGblxxRVXKOYG1j916tT111+vCj/99JNT qDCRO3duLfLbb7/pY//+/a+88krVyZo16zXXXKOJTJkyvf3221Y56Hm0wo0K v/7661DtVOi3hmmdV111lU0//PDDNldRRh/vvPNOp/3jx48/evRopUqV7GPe vHnVPE0otO3bty/oJizfyJkz59VXX209YMs2bNgwypbbGhTp7Dza8eSTTzr1 hw8fboXakLpFEzfddJP+++GHHwbdkTFjxkTs8GhY22699VYtlS1btgIFCtj6 FfGD7l3E3lNOlSNHDquQPXt2m9AezZw50yoon3Hm2p5qu87cCz06MnbsWK1B 1fSlsk3rmzBkyBCnQpgviXI854BKhQoVouw3AECq8Lufogk7AQyzSJs2bVSn devWTsnkyZNV8sgjj2haZ9yZfAYPHqzz+sTEREVSRXD92ttVjhTkG0uWLNFc BfE5c+acPXv2yJEjo0aNssBhZ7gWpq0NEydO7NKly8mTJzt06GDRPz4+XnX2 79+vE16V6DQ86FYsIluEiouLO3bs2MCBA61kx44d0bTcWYM2NHv2bLXho48+ sjD3888/q4J6QHuhoKlGnj9/Xjtii/vlG347Er7Do+S0rXr16to1lcyaNUuB W+1Zvnx54N5F7L3HHnvM6u/evVv7smrVqipVqqikZs2aXl86oe9AlixZ9H1I Skr6/fffmzZtqrmPPvpolOv3o61YVtOnTx99qbTFCRMmqCf11dqyZUs0XxIv 91MAIP2kIN/Qr7fq3HDDDfpVt5LatWur5NNPP9W0XbVu3ry5e5F27dqpULHV m6J8Q5FFcapXr17uwgceeMBZxMK0zmoPHTpkcxVu7DKFZjmLbN++3eO7cLFn z57ArVhEVghTcHQKb7/9dhV+//330bTc1qBTb/caHnroIRVOnTpV0y+//LKm //Of/zhzT5w4EXh9w70jETs8StY2JRiK7E5h586dVajQ77d3EXvv1KlT9erV K1u27N69e50K1vgHH3zQ67t/oek8efLs3LnT5mqPunfvbs/bpuDovPLKK5qr g+4ufPHFF1UYGxvrjeJL4iXfAID0k4J8Q3RaqmrTpk3z+uKIYoeC7J9//qmP 9957r2bNnz/fXd8iZqFChbwpvZ9idJI7b9684cOHq4W5c+fWIhMnTvQmRzq1 yqm5bNkyO7cdPHjwBy65cuVSoVYSuHKLyH4R7Z///KfTP1HmG6VKlXJXqF+/ vgrHjh2r6fLly2varic4LGi68w33jpgwHR4la9tTTz3lLly4cKEKq1Sp4rd3 0ffe6dOn165dO2nSpJ49e9rgnyJFitgsu2Wj3KlmzZpDhw5VOuEslYKjU6NG Dc2KiYlx12/YsKEKK1Wq5K4Z6kviJd8AgPSTsnyjf//+qvbcc89pWqFE002b NrVZNtjALnE79u3bp0JFSW9K842RI0e6RzXoLNiGiLjzDXcwHT9+vCe0UaNG BW7CIrJ7rIU0b95chaNHj46m5UHX0KRJExV+9tlnmrZ4+uuvv7orvP766375 hl9W4A3b4VGytrVq1cpdmJCQ4PENIPHbu2h6b8eOHUqlnGESYkfEyTe08mLF irkXVCZm95VScHQKFy4cqn7+/PmtTvgviZd8AwDST8ryjf379yvQ6NT1+PHj ZcuW1SILFiywWYUKFdJH93MKsnHjRhXqZNObHNf8HrmtUKGCJ3S+Ya3y+O7+ d+3adfLkyQcOHNCpruev+UbdunWdRaZPn66SggULzglm165dgVsJ+jxsYL4R puVB1+DON+6++25Nr1692l1BOYBfvuHekYgdHiVrm9/oCLvuVKZMGe9f842I vXfq1Cm7kHXLLbc0a9Zs+PDhS5Ys+eGHH9z5hlGC0adPnypVqlhmkjNnzpMn T6bg6ChX0SI9e/YMrG+P9kT8knjJNwAg/aQs3xBFVdUcNGiQziLvuusu5zf8 iSeeUPn777/vrqx4rcISJUpoetq0aZ7kUYUOG8MQKt+wC/WKNe5CndV6fM+h eIOF6fj4eI9vKMWJEyecQjWyd+/e/fr1c1/bd0TMNyK2PGK+oWU1PWDAAGdu UlLSfffdFzHf8Ibu8ChZ24oVK+Yu1GHyJI9EdecbEXtv7ty5Ht/wjMOHDzsV bHxm4cKFvb40o1evXu5Hqrdt23bttdd6fINhUnB07NaJ33dg+fLlXbp0sXQi 4pfES74BAGlAZ5G//fabfvbXrFmz2kU/+3/88YdTLcX5xpQpUzy+5x/13x49 ejjlI0aMUMltt93mnKXu27dP8dGJBYsXL9a0Qo8TVpzHQELlG3Yq3bdvX6fk nXfesUXsz50EhmnFcRvz0KZNGye+WHjNly/f0aNHA7cSMd+I2PKI+Yad1994 443r1q3z+iJst27dbA0R841QHW50lBctWuQeCxp072TChAlWosp33HGHSpRp eP+ab0Tsvc8//9zyjVOnTtnc3bt321EuUKCAPqpDPL4nYZ2/V6Kl7HbShg0b UnB0rHk5c+Z0btUpV7G7J/bFiPglEUt4Dhw4EKqXAADRO3PmzLJlyyZNmjQx GAVNd1RKcb5x9uxZxU2P7y65+4RUoaRMmTIe30sVXn755bZt2958880e36gG e5tTYmKiPfShcp20Pvzww1qDhdFQ+ca//vUvizXNmjVTqLUHJ4291iNomP7h hx9stWXLln311VdjYmLskr49KhIoYr4RseUR8w1RIz2+xzZLlizpHmygPC3U joTvcGMXSWwlYfbO4xtl0ahRo44dOyondB9rv9Ep4Xtv586d9iqM0qVL64ho JbfccotVUEw/d+6cjnWRIkU8vsdqtDlVsK5zBtNe6NERO+5af8uWLZWo2LUL rdBeyBbxSyLW4cqIXnjhhVBbAQBE49ixYzqJdicY9kpJx/Lly91/IrNq1ar6 BXaue9v7zP1GFYZiT7na0w1uOvFs3bp1lixZ7NdecS02NtZ5F6jX96dI7UFF j+/505deeqlr166aVkwMuqHjx48/99xzTvjQabJOXS0+li9fXhW0X57k4ZRu S5cudQ9ZzJ07t/v818+MGTM8yQ/tOuzxXicfC9/yoGuwx0/s+RSvLx9T+FOU zJo1q5KEnj17NmjQwLnsEGpHwne43H///dHkG+XKlVMgtleC3HTTTWq/8+5T Jage18vNIvaemqocw2ZphRUrVlTnWH17fFhJkXrPciTrLu2pDmWU6w+kr1Dn zp3tDXJG++Jc7oj4JfH6ciq7xKEvpw5EmG0BAMI4evTotGnTLM1YuHDhoUOH 0vRHtXHjxu5cxY/OOjdu3Lh169ZQi+/fv3/16tXOOyUiOnz48IoVKxTFLvT+ u3IwLZiQkBD9tsK70JY7tNSPP/7ojrlir9n87rvvIi4epsPPnDmTOXPmb7/9 NtSy7msv+mKsW7cuym4M03vKVXR8V65c6R6G4UffQO11XFyc81r46NcfaoWq rI26h444In5JlLTs3LkzTIMBAOG5kw2/P1SRFuLj43WGft111508eTKtt5Vh vPnmmwr67ov5NqIjU6ZMR44cCb9s+A7v2LGjVhI0BJu/+V+jAwCkCney4ffu i1TXv3//kiVL2nXp3r17p+m2Mpjly5fbezVLlCjRpk2bqlWr2l0n9xMrgaLp 8NjY2DAXN7zkGwCAi3Ypkw1v8kMZNuyQRwsv1JQpU+655x4bY5ApU6bixYsP HTo0/CKp0uEzZ87MnDlzqJEhAABEtHTp0kuWbHh9N9A3b97s/gMfuFD79+9f u3at30COUOhwAMDlIDExccGCBZcm2QAAAH9b3NcAAAAAAAAAAAB/Z4lJScvi 9oybFad/mkjkvYgAACBVHTxy8o1hPzbuMcP5p48qTO92AQCADCIxKckv2XBS Dq5yAACAVLEsbk9gsmH/NCu9WwcAADKCcbPiQuUbmpXerQMAABkB+QYAAEhr 3E8BAABpjfGiAADgEuB5WAAAcAnwvi8AAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAglNdee61TCD169Ei77e7du1ebGDx4cKgK W7ZsUYWPP/447dqQ6iLu1MWv4fz580OHDq1bt27FihXbt2+f4g2luyFDhmhP d+/enSpr69evn9Z26NChVFkbACDVZc6c2RNC3rx50267a9eu1SZKlSoVqsK3 336rCk8++WTatSHVRdypi1/D22+/7RygGjVqpHhD6a5MmTLahZUrV6bK2u67 7z6tbfPmzamyNgBAqrN8Q1FsTICJEyem3XYjBtZVq1bVrFmzV69eadeGVHcJ 8o2HHnpIFdq3b//bb78dOHAgxRtKd506ddLx/eWXX1JlbeQbAHCZs3xjy5Yt l3i7Fx+aL0OXIN8oUKCAKmzcuDHFm8iQyDcA4DIXZb6xZs2aFi1alPeJjY1d unSpX4XTp08PGDCgdu3axYsXf+aZZ959991Tp065K+zfv//999/XKW3FihXf eOONBQsWhA+scXFx2qIWsY9JSUmjR4/WmrX+mJiYdu3ahYm5GzZs0LLTpk1b tGhRq1attBVtt1+/fufOnXNX27RpU9u2bStXrlyuXDnt1IoVK/zWE7FCxJ1S s4cNG1a/fn01++mnn+7bt++ZM2dS1i3z58/XTl1zzTWq0KBBA03//vvvEXu+ V69e6gEd3zp16lSpUkU1ExMTg3ZafHy87eyDDz7YsGHDDz74QI23WatWrdLm Ro4c6a6v/VLh+vXrvb40SdPTp09XJ5cpU6Zly5Zt2rRRid/tEh0yFdq4oP/+ 97+a3rFjh9agiW7duvm1p0+fPirXF88+fvXVVzoE2oWiRYtWq1ZNc7XjTmXy DQC4zEWTb/Tv3//KK69UtaxZs1q8y5Qp09tvv+1UOHDgQKFChWxQQe7cuW1C IWD79u1W4Y8//rCIoPXkyJFDE/bf6MdvtG/f3labPXt2bV0T2bJlmzlzZtBl v/nmG1VQYPIbneIeDTJ27FitQYXaI2vMVVddNWTIkOgrRNypo0ePVqpUyRkM c8UVV2hCWcG+fftS0C3atK3BaFrBOmLPP/DAAyq5//77bVbhwoWD9tjWrVvz 5Mljq82ZM6dVrlq1qs394osv9PH55593L6L0SYVff/2115cMaPrOO+90mler Vi39t2nTpu5F7CB27NjR6xq/ceTIEX2vtF2nzXLw4MGrr75ah8+GgCpZstWq po6CTT/88MNOffINALjMWUTWmfWQv/r000+twuLFizP5DB48+Ny5czo7/vDD DxULFCCcqxxPPPGEVlK6dOmEhAR9/PXXX8uXL68SlVsFixfVq1c/fPjw+fPn Fb8UOKLPNxS41YAsWbKoMTrp1nm9ApnmPvroo0GXtXxDdCI8e/bskydPfvTR Rxasf/75Z1XYvXu38hZ91Gmy9khNmjBhgrIL7ZelXhErRLNTHTp0sLQnPj7e 67uUofaopEmTJinuljvuuEMVnMc6Iva85Rs6yu+9996oUaMmT54cdLWdOnVS tUaNGqlv1ZIlS5ZYYrlw4UJv1PmGPPLIIxMnTuzSpYvdGFLq4lxsUTfedNNN KtywYYP3r+NFGzRooOnevXs7K9c3UCV169bVtBpjuzBnzpyzZ88qP9GO2Oac K07kGwBwmQv1fIrOVa1CixYt9LF58+bupdq1a6dCRRxN79q1S9MKxNu2bXMq KCBawFLc0Vm8Yr0i6d69e50Kr7zySvT5huKIpnUCvnPnTpur097u3buPGzcu 6LKWb+TIkcPuOBgbaTl16lRn64rF7qVefPFFFcbGxkZTIeJOKSyqT/RRsdip oFN4u4awZ8+elHWLO9+I2PPe5Hzj3//+d6gVGuUSqvbaa685JZ9//vnAgQMt uYoy39C+uJ9ItcxHy9pHO6DOrrnzje+//94uyzjL2tzp06drWpmeEku/kcO2 X7Z1L/kGAFz2LN946aWXevyVc+Pg3nvvVYX58+e7l9J5pQoLFSrk9QUmTVeu XNlvzTVr1lS5UoJFixYFVrA1RH8/xcZJKqJptUOHDnVfew9k+YbfyuvXr6/C sWPHarpGjRqajomJ+cClYcOGKqxUqVI0FSLu1LJlyyxzGzx4sHsluXLlUuG8 efNS1i3ufCNiz3uT4/KXX34Zprvk448/ttbqmCqZ/O6779zjTKLMN4oWLequ MHLkSBU+9dRT9tEuYjgDctz5RlJS0u23366Pq1at8ianl7fccovfUBPttfpt +PDhyoHt5pHzCBX5BgBc5iKO37j++usDK+zbt8/OrDU9aNCgwGAksbGxdpF8 xIgRmnjuuefcc3VSf0H5RkJCQrFixdxXYLSs3RwJZPmG37s7mjRposLPPvtM 04ULFw56VUfy588fTYWIOzV+/PhQa5BRo0alrFvc+UbEnvcm5xsWx8N76623 bLyKyZEjR9euXW3IaJT5hpNamD/++OPaa6/Vl+TQoUOa1sqzZMniXHHye//G m2++6Uke2tGlSxdNd+7c2VmVUhf34JArrrjCBhSRbwDA/4qI+YYNRwx80MDj G6CoaZ1HBwZ3qVevnsd3dj99+nRNPP744+6527Ztu6B8wyjB6NOnT5UqVWzQ YM6cOU+ePBm4rOUbtWrVche68w1tV9M9e/acE8BGLESsEHGnrELBggUD1yC7 du1KWbe4842IPe9NzjdsyEREx44dmzRpUvPmzW+99VaL7EpCvMn5hlbrrlyh QgW/fMOGWwT2+QcffGDXT+rXr+/M8ss3fvnlF2URdk3D75lf59rLo48+qhRo 8uTJBw4ciImJId8AgHR3/vz5o0eP7tixQz/jW7dujU+mkhMnTrhrRsw3bESi cxncjB49WoUlSpTQ9PLlyzWtCOW+Aq+ocdddd6l82rRpdjFEocRdYcqUKdHn G0ozevXq9cknnzhzFZd17qwK33//feCyEfMNuzOidMJdQTuiM2sLYRErRNwp 9bZdJXB3uI5L7969+/Xrt3379pR1izvfiNjz3qjzjY8++qhDhw7O6Au186WX XtKC5cuX10etStM1a9Z0L2KDP8PnGz/88IPlCY899pgmZs2a5cwKfL+o0kiV 9O3bV/8tW7asU16xYsXAY5E/f34Vjh8/3j6SbwDApadYPGPGjInBzJ071+8v VkTMN+yy/2233aZTcitRoLSIZiHg9OnTdq3b/fdWLGoo2v7555/e5NsTzgMI CsElS5aMPt9QUPP4noQ9ePCgzVU2ZQMhgkbSiPmGnbDnzJnT2XE1yfZCLY+m QsSdSkpKKlq0qD62adNG4dvqKG1TSb58+dT+lHWLO9+IpuejzDcefPBBz1/H i3br1s1JIRYvXqxpJXjOsJmBAwfaNYfw+YZ23L4qlhc5L/TwBss3lE86d0yG Dx/ulNsIIqfb5Z133rFqY8aMsRLyDQC4lHRu+9NPPwXNNFKcbyhGWGjImzfv yy+/3LZt25tvvtkyAee0WkHHnjatVq3a66+/boMts2XLNmfOHKuwdu1ae7z0 qaee6ty5s0WQ6PMNbahIkSL6eMMNNzRq1KhFixY2vDDU4hHzDbFHUxVD7eVU dr6sFTpjFCNWiLhTOru3V4XobP3VV1+NiYmx20D2jEzKusXvediIPR9lvqFu 8fjGRehYd+zY8ZFHHrFHnqdMmeL1fa+sw3XoGzZs+PDDD2uW7Vr4fMPre+GY 7dR//vMfd3lgvqEEyV4/ovZbPmb+9a9/We7XrFkzZVZ2XIzzEhjyDQC4ZM6d Ozd//nzLKxRw4+Pjjxw5cvLkSZ0Fn0l29uxZ9zmm2Asf7AURoei8u3Xr1lmy ZLEf+dtuuy02Ntb9dkeZN2/ePffcYxUUqipUqDB79mx3he+++65cuXI2zC9X rlx22URBJ9RGZ86c6c4ZdGatNOPGG2+0TSgkNWjQ4Pjx40GXnTFjhif5cV2H Pc1qz6d4fRcHFOLt0VGjQObOuyJWiGanli5d6h7mmjt3bvd5egq6xcY2uB+h Dd/zxYsXV3lcXFyoFTpGjhypZa0log3ZHRmzfv16S12s81966aWuXbtqWt80 zVUG5QkY+2p27Nhh6/T7jikH8wQMZG3evLkKldK4C3WUtWanD9VL77zzjl2A sts9Yi80C/81BgBcPHeyodNqv5dmpwqd5G7cuHHr1q1h6hw+fHjFihV2JT+o AwcO6FzbL+eJnhZcvXq1omeol3KnYIUJCQk6y1bLU1bBG8VOHTt2TN2i9Sjf S9kaIorY81Hat2/fkiVLQu3s/v371f+h9iJN2Q4q7XRuTgEALjF3srFgwYLU isUAAACGZAMAAKS1DRs2kGwAAIA0pRzjxx9/JNkAAABpKtEnvVsBAAAAAAAQ XGJS0rK4PeNmxemfJhIv4pFJAACAQAePnHxj2I+Ne8xw/umjCtO7XQAAIINI TErySzaclIOrHAAAIFUsi9sTmGzYP81K79YBAICMYNysuFD5hmald+sAAEBG QL4BAADSGvdTAABAWmO8KAAAuAR4HhYAAFwCvO8LAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAADgMnfo0KFOkSxatGjTpk2aGDNmTHq3N81t3LhRe/rJJ5+EqrB+/XpV GDduXCpu9ODBgxez+N69e9WkwYMHh6pw/vz5oUOH1q1bt2LFiu3bt7+YbV08 984OGjRILd+zZ08qrt/vAA0ZMkQfd+/eHbQBfiVpcXBTxu+YbtmyRR8//vjj dG0UAKRcfHy8JxIFhW+++UYTtWrVSu/2prlp06ZpT5955plQFb744gtVeP75 51Nlc4ojlSpV6tq168WsZO3atWpSqVKlQlV4++23naNZo0aNi9nWRerbt2/2 7NmV/9jHkiVLqkmrV69OxU34HaAyZcro48qVK4M2wK//U/fgXgy/Y/rtt9/q 45NPPpm+rQKAFDt+/PgYlwYNGuhnrVixYu7CzZs3k284UjckWSaQ1vnGQw89 pArt27f/7bffDhw4cDHbukiZM2dWSy5lvtGpU6eaNWv+8ssvQRvg1/+Xbb6x atUq7UWvXr3St1UAkFoGDRqkX7mnnnrKr5x8w/G/mG8UKFBAFTZu3HgxW0kV lz7fCN+A/5V8AwAymIj5xuLFi5s1a6YwUadOnYEDByYmJrqrJSUlDRs2rH79 +sWLF3/66af79u175syZMJuLj49v27Zt5cqVH3zwwYYNG37wwQdag7vCmjVr WrRoUd4nNjZ26dKlzqwNGzZoltKDRYsWtWrVSr/MOgHs16/fuXPn3GuYN2/e K6+8Urp06bp1644bN2727NlaauHChaGa5OQbM2bMqFevnhrWuHHj6dOnOxUC Q9LevXu7dev27LPPaq/VM02aNHGu3lufjB49WivU3JiYmHbt2jlxX/1jAVez 1KpNmzZZuSasW8qVK6e9XrFihV8j9+/f//7772t/K1as+MYbbyxYsCBUbJo/ f77WfM0116hCgwYNNP3777/rNFk9tmXLFh3EKlWqDBgwwI5jmN72+kKzlvrj jz969+5drVo1Ldi9e/cjR45oB4cOHarvjPaladOm27dvD9qxmzdv1srtno4m 3n33XW9yvrFkyRKt87HHHitbtqz6R+fy7gUv9Evld4D++9//anM7duwIbEBg /wce3AvduvIErUpfGH0Vy5Qp07JlS9ud8F8Sb6RjGhcXp9WqglP/9OnTOnC1 a9fWCvXt0u6cOnUqTMMA4LISPt/ImTPn1VdfrYkrrrjCfreVJDh1jh49WqlS JSvPmzev1dGP4b59+4Jua+vWrXny5LG1ac22YNWqVZ0K/fv3v/LKK1WYNWtW i5iZMmVS1HM3qWjRonbG6nDf4x4+fLgVav1aVhM33XST/vvhhx+G6gHLN7Jl y6ZWaeu5cuWyNbz++utWwS8kKdvJkSOH1cmePbtNaFszZ860Cu3bt3fmWhu0 cpvrdJd1wpw5c1Q4duxYVVCJdtnWfNVVVw0ZMsRpoSL+fffdp3I1zyrYf4Pm G1rQOVi2FUXeBx54QNP333+/FRYuXDhib3uTcwNlbu7erlGjxgsvvOAuufXW W//888/Alijxc7ekQoUKzjpvuOEG95dKB8tJsS70SxV4gJzxG4ENCOx/v2VT sPWvvvpKde68805nzePHj4/4JYl4TP3Gbxw4cKBQoUK2nty5c9uE1hAq2QOA y034fEMefvhhnWodO3Zs4MCBVqL4ZXU6dOjg8SUA8fHxXt/5ms6CVaJTuaDb 6tSpk+Y2atRIZ9znz5/XSa6FObv4sHjx4kw+gwcPPnfunE7AlSQo29Fvvp13 O03SVmbPnn3y5MmPPvrIIsLPP/+sCqqmVESxe+LEiVq/zsR1fmqLRMw35NFH H7XHFj7//PMsWbJozXbZ3y8k6azcPu7evVtb0cmsTvxVohNVry9gaRe0uHZH Z8ra06ZNm9rKbXG/6/laicWjPn36aJe1wgkTJmgXtONbtmyxOnXq1FGF6tWr Hz58WBXUHmUInrDX3u+44w5VcJ7RsHxDnfPee++NGjVq8uTJEXvbm5wbKPlR AD1+/PikSZOst7VU7969df6+bt06BWWVaFaolgS9n6I1jBgxQutUmmHpQbNm zVL2pQo8QH7jRS/ofkoKtm75hjzyyCP64nXp0kXfzPBfEm8Ux9Qv33jiiSc8 vtwvISFBH3/99dfy5curROWhGgYAl5Xw+YYCnyKmU3j77ber8Pvvv9e0orld +tDvrVNBZ1t25hj0gUf9/Grua6+95pQosiuNscBql76bN2/uXqRdu3YqVNrg NEmnge4m2cDIqVOnavrll1/W9H/+8x9n7okTJ6K8vqGg7x5XaauyIOgOSadO napXr17ZsmUVbZ3KFnEefPBBr+8mgqbz5Mmzc+dOm3vo0KHu3bs7T1z6xbtX XnlFH5UPuJv04osvqjA2NtbrOxFWfyoYubdoS11ovvHvf//bqRCxt73JuUHn zp2dCv/4xz9Uoub5rcfulQQVNN949dVXnQrKf1Si6OlN0ZfKm3r5Rsq2bkdf B0gH2koifkmiOabufGPXrl2aVtu2bdvm1NfBtXR97dq1QRsGAJeV8PmGXxz8 5z//qUJ7J8CyZcvstE4nyB+42P2IefPmBW5LC9oihQoVUmj77rvv3HfG7733 Xs2aP3++exGd/1p9p0l+QbZ+/foqHDt2rKbtjG/58uXuCha7I+YbfiNj586d q0KFDG+IIYWnT5/W77zO63v27FmxYkVVKFKkiM2ysZqKJjqZHTp0qN8Vb794 V6NGDX2MiYlx92HDhg1VWKlSJa/v9o2mK1euHNgtF5pvfPnll06FiL3tTc4N tPtOBbuT4h5UoB1RSZjHKILmG+6A/ssvv6hEDfam6EvlTb18I2Vbt0SiaNGi gbNCfUmiOabufENpeWB90RdM5ZfDy0MAIKLw+Ybf4/86HVbh6NGjNT1+/HhP aDppDbq5t956y8YqmBw5cuiX34aMXn/99SpxbiKYffv22ZldqCY1adJEhZ99 9pmmLS78+uuv7gqvv/56NPmGXUxwbNq0SYX58uXzBoSzHTt2KMm56qqrnL2w URBOvpGQkFCsWDF3byiI2B0fb0C8K1y4cKg+zJ8/vyqMGDFC088995y7eTov TkG+4R6WGbG3vcm5gXsQqSWc7gDXo0ePFOQb7pGTdvJ+++23e1P6pUqtfCNl W7d8w+//QeG/JNEcU3e+Yf8nDXyIRl9alffu3TtU5wPA5eOCnod15xvTp0/X dMGCBecEoyASaovHjh3TGZ9Wdeutt9pPsZIQldtwOL8x/Bs3bvT4BsiFapI7 37j77rs9Ac9atmrVyhNFvuGMHzBLlizxJF/99rufYlcGbrnlFi0yfPhw1fzh hx88rnzDKMHo06dPlSpVLOjkzJnz5MmT3oB4p/iijzr/DexDG9Zi/fz444+7 V75t27YU5BsbNmxwKkTsbW+wZ1dTK99wr9Odb6TsS5Va+UbKtm75Rt26dZ2S iF+SaI6pO99Qh3uCvfurXr16Ht/VmFCdDwCXjxTnG/ae0hw5cpw4ccKpoF91 nW3169cv6LD5jz76qEOHDs5tblV+6aWXPMn37m1EnPtavWhbKixRokSoJrnz Dbu8PGDAAGduUlKSPQUQMd9QHHQX2uBY/Z57/xqS7D5Lnjx5Dh8+7FS2EQj2 0IfSDAVf99vRFUeuvfZaT/K4F794Z7dOlG+4t758+fIuXbpMnDjRm3zNQZHL fe9pypQpF5lvROxtb3rkGyn4UnlTL99I2dYD842IX5Jojqk739D3weN7Dshd PzEx8a677lK5vsBBGwYAl5UU5xsK5UWLFtXHNm3aOD/mil8e322Io0ePBm7r wQcf9Px1vGi3bt2c32q7yHzbbbc5J5L6WbZfVAvHEfMNO2288cYb161b5/VF Clt/NPmGTJgwwUr27Nlj115smIE7JNmddIUS59UHiunWyAIFCujj119/7fGN PnX+Qoe6wm70WLjv27evdZrNtZXnzJnTubWhYGcPV6qmldg9F+eyuSpY1L6Y fCNib3tTKd+wXMsZixs+30jBl8obKd/wa0DQ/rdlU7b1wHwj4pfEG8Uxdecb p0+ftq+EetvZiu2IsqOgDyMDwOUmxfmG/PDDD/Z+ibJly7766qsxMTF278Ce FgmkrMDjG+qviNCxY8dHHnnEHsDUmZ3X92tvkSJv3rwvv/xy27Ztb775ZvvJ tdO6iPmG6Gff43vwU7/e7lciKLyG6gEn31DjFXc6dOigE0+P6wlWd0jauXOn jT8pXbq0fvxbtGihyrbXimvnzp1TU4sUKeLxvWKiUaNGqmAP9Thx5JNPPvH4 xkg89NBDOhFWiT1xqcVbtmypSJc/f36r77xabe3atfbMrA5T586d7Vr9ReYb EXvbm0r5hh0FxdkXXngh6Drd+Yb3wr9U3kj5hl8D/Prfb9kUbD0w34j4JfFG cUz9nodVHmsPI+vb8vrrr9swY23FXuECAJe/IUOG6Ierdu3afuUzZszwuB6N NPbwo/tvVi5dutQ9NjJ37tzOWXlQI0eOrFChgo2dsyjgvhqss7zWrVtnyZLF 5ursOzY2Vid3YZpkj5/Y8yleXxh9++239aOdNWvW++67T6fq9gdinGsXgeyK hFKp6tWr20+6gpFCvw23kEmTJnlcLzpT6LGExHKnihUrrl+/3jrB7phs375d HXXjjTdaHQUFteH48eO2uCbsVQzO6a12UBHHnm00iil+Izm/++67cuXKWb/l ypXLrk4osIbaKXtGxnncsnjx4voYFxfnrhO+t8Xe9OV+3LJx48Ye3/usnBJl Gp7kEThBKaDbFQZtSEcncJ1KijzJz6eYC/1S+R0gpQoe1+BYvwb49b/fsinY ur4PnoDBnxG/JN5Ix3TmzJmev2bX8+bNu+eee2yd+orq/0ezZ88O0zAAyHiO HTu2YsWKhISEs2fPRlN/3759S5Yscd/ddtN5/caNG7du3XqhzdBZ848//uhE dmOvi9RvezRrOHTokOKUO+YGpVNUNU9n0O4b/X4U2tQehXi/N8Ab9cD+/fud i/ZWXx2odYbqFq/vJZMbNmzwewP8RUpxb0dP/alT/jB9FdSFfqkuqAGB/Z/q W4/mS+K98GOqr4caxj0UAEhHb775plILu2xubERHpkyZjhw5ko4NAwAAGcby 5cvt/ZAlSpRo06ZN1apV7WaB+4kVAACAizRlyhTnHnemTJmKFy8+dOjQ9G4U AADIgPbv37927Vq/gRwAAAAAAAAAAAAAAAAAAABARnL8+PEDBw44fx0PAAAg 1W3fvn3p0qUrVqxI74YAAIAMa/Xq1RMnTgzzh8YAAAAuEvkGAABIa+QbAAAg rZFvAACAtEa+AQAA0hr5BgAASGvkGwAAIK2RbwAAgLRGvgEAANIa+QYAAEhr 5BsAACCtkW8AAIC0Rr4BAADSGvkGAABIa+QbAAAgrZFvAACAtLZly5Y5c+Ys WLAgvRsCAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAXFKJSUnL4vaMmxWnf5rQ x/RuEQAAyFAOHjn5xrAfG/eY4fzTRxWmd7sAAEAGkZiU5JdsOCkHVzkAAECq WBa3JzDZsH+ald6tAwAAGcG4WXGh8g3NSu/WAQCAjIB8AwAApDXupwAAgLTG eFEAAHAJ8DwsAAC4BHjfFwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAID0kpiUtCxuz7hZcfqnCX1M7xYBAIAM 5eCRk28M+7FxjxnOP31UYXq3CwAAZBCJSUl+yYaTcnCVAwAApIplcXsCkw37 p1np3ToAAJARjJsVFyrf0Kz0bh0AAMgIyDcAAEBa434KAABIa4wXBQAAlwDP wwIAgEuA930BAAAAAAAAANLO+PHjO4XVtWtXVdu0aZOmx4wZE2o9W7ZsUYWP P/740jU97X3yySfaqc2bNzslBw8eTNMtDhkyRFvcvXt3GtW/HAT26oXaunWr vpCp2CQAQFpr2LChJ6zrr79e1b755htN16pVK9R6vv32W1V48sknL2Hb09zT Tz+tnZo+fbp97Nu3b/bs2c+fP592WyxTpoy2uHLlyjSqfznw69ULpWRjok98 fHzqNgwAkHYWLlw4xiVbtmyKBd26dXNKPv/8c28U+caqVatq1qzZq1evS9j2 NOcXGTNnzqyPl1W+0alTJ3X7L7/8knZNSnUXk284ycbUqVOPHDmS6m0DAFwa 119/vWKBkge/8oj5RoZ0+ecb/4tSnG+QbABAhhEx31i8eHGzZs1KlixZp06d gQMHJiYmWoW4uLgWLVq8//779jEpKWn06NHPPPNM8eLFY2Ji2rVrt3HjxjDb jVj/9OnTAwYMqF27tiqo2rvvvnvq1Clnbvfu3bX1/fv3OyW7du1SSe/eve3j hg0b9HHatGmLFi1q1apVqVKlatas2a9fv3Pnzrm3Mnfu3Pbt22vuCy+8MHPm TCcybt68WYvbDSZNaOuq3KtXL61qy5Yt6ooqVaqoedYbe/fu7dat27PPPqum qqOaNGnilz+cPHly7Nix//d//1eiRInHH39cCzr74pdv7Ny5U5vQFtX/Qfvt v//9r+bu2LHDPsbHx7dt27Zy5coPPvhgw4YNP/jgg6Swj4Wqx4YPH65eLVeu nHZ8z549Wrxz585RdrtZs2aN2lDeJzY2dunSpX4VQvVqmIYFUj+TbABAhhE+ 38iZM+fVV1+tiSuuuMKCr4KaVfAbv6H4YhWyZ8+eKVMmTWTLlk2xJtR2w9c/ cOBAoUKFrELu3Llt4r777tu+fbtV0LRK3EMQ165dqxKFXXf7ixYtatcoHO4B Jx999JEVajevvPJK7eM111xjkXHevHnOLkuFChVU/4EHHtD0/fffb4WFCxdW ofKZHDlyOPtiE9ojZ1+U4WijVn7dddfZRL169WyuO9/YvXu37fUTTzwRGOUD 6+v0P0+ePHZ0tAu25qpVq4bq82PHjt1zzz2qo521+vny5dN/b7nllii7Xfr3 76/FVZ41a1brLu3s22+/HU2vhmqYnz///HPZsmUkGwCQkYTPN+Thhx+Oi4tT qBo4cKCV2Mm1O984evSogk6WLFkWL16s8+vff/+9adOmmvvoo48G3WjE+gq4 +li6dOmEhAR9/PXXX3UqbYHYKkSZb0i1atVmz5598uRJxUFLIX7++WdVWLFi hVIRBc0vv/zS63sOJSYmxhYJdT/F8g0Vvvfee6NGjZo8ebIKH3vsMRU+//zz yhZUUz1ZpUoVldSsWdOWeumll/SxePHiSg/0cfny5Xnz5lWJEhWvK3/Yu3ev JQNPP/30mTNnQh0vd77RqVMnTTdq1EgdqE0vWbLEIvvChQuDLlurVi3NVYMP HTqkj19//bVlHU6+EbHbdbwy+QwePFh5VGJi4ocffqiMVB1rVzmi6VWH5qq7 5rroSKnaxGRqIckGAGQM4fONbNmyKZY5hbfffrsKv//+e+9f8w3FfU3rXHvn zp1WUxGte/fu48aNC7rR8PV37dqluYpi27ZtcxZRNLdgqrzCG3W+kSNHDnf7 H3roIRXqlNmbHKzbt2/vzNXZvd+ZeNB849///rezyKlTp+rVq1e2bFllC07h V1995W6JeljheN26dU6FYcOG3XnnnfbcseUPM2bMsD3S2vzu+Phx5xtKcjT9 2muvOXM///xzpYVbtmwJXFA9bB1iyYbp27evk29E0+12j6l58+buNbdr187S pCh71b3yefPmTQzhxx9/VJYYpisAAP9DwucbirDuwn/+858qtHdu+N1PKVCg gF1j13n90KFD3VfggwpTX0FTsypXruy3iGqq3HKSKPONUqVKuddQv359FY4d O1bT1apV0/T8+fMDNxE+37Azdz+nT59WAyZNmtSzZ8+KFSuqWpEiRby+SwSa LliwYKh+sPxB/WA3ICJGWHe+oQNhlw4KFSqkoP/dd9+FuTCiLNF9pcJY8yzf iKbb77333sBOW7FihbXBG12vOpT5rFmz5odkCxYsUI6xdOnSuLg4d5YIAMgA wucbfq/X0ImtCkePHu0NyDcSEhKKFSvmcVGstzsXQYWpP2jQILtD4bdIbGys ym1EaJT5hl/7mzRposLPPvtM0zZ0QaHNXaFVq1YR8w2/vtqxY4fSmKuuusrZ ERveYPmGNSMwiDssf/Akj5Dp0qVLqJru+s740rfeesueaDY5cuTo2rVr0CGj 7733niqoB9yFZ8+edfKNaLrdvi1+10/27dtnF0a80fUqAOBv6IKehw2Tbxgl DH369KlSpYrF34gn7EHr61Q6cM1Sr149lQ8ePNibnG+4H2mZM2dOYL7h1353 vqH0RtN+z1Y0btw4Yr6xYcMGp/6pU6fslF8hu1mzZsOHD1+yZIlO1Z18Q+fs mi5RooTfvjjrtPxBHTtr1ixNqB8Cj4Vb4POzx44dmzRpktZw6623WtahJCRw Qe11YIfs3bvXyTei6XYbTer39I2Ogsc3vjTKXgUA/A2lSr6htKFXr16ffPKJ U3Pbtm3XXnutJ3mwh5/w9ZcvX64JRU/33YHExMS77rpL5dOmTdNHRXBNr1ix wqkwbNiwC8o37LR9yJAh7gp2ySX6fGPu3Lke30CUw4cPO4WjRo3yJD+9snv3 brvscOLECafC7NmzteZKlSp5/5o/vPjiix7fMzVRjhf96KOPOnTo4IzHUDtt bGr58uUDF7S7HjfddJM7A5wwYYKTb0TT7Tag1HkI2uj74ORU0fQqAOBvKFXy ja+//trjexrU+WsjR48ezZUrl190doSvf/r06TvvvFPTPXr0cBaxkY0K3H/+ +ac+Vq9eXR8VXm2u8yBn9PnG+PHjNZ0/f/59+/bZXAu+7shoKZBWbh8D8w0b 86B8w3l8VQmGBegCBQpYiZrkd6PEGq/Q7P1r/qCkRfmAPr7++uuBnWbc9W3N 7vGi3bp1U0ndunWDLlu6dGnNbdOmjb02JCEhwfrZ8o1oun3EiBH6eNttt+3a tcsqqPdsf3v27BllrwIA/oZSJd/QGXGRIkX08YYbbmjUqFGLFi3sSRa/4ZqO iPWVkNh4hmrVqin41qhRw+N7WGbOnDlWwQKf6lSpUqV+/fo33nijvQQj+nzD mzz8NV++fB07dlQzbNyFOzJa/FXm8MILL3iD5Rs7d+604RMK5QrT2hHFbrs3 pFzFnjT58ccf7QUjjz/+uDZUsGBB2/Fff/3VG3B/ZOLEiXZXxX3pxs1d326R qBNUqDU/8sgj9mjqlClTgi77008/WS8pYVCDnYEf6oEouz0pKckakDdv3pdf frlt27Y333yzfQ2cqyIRexUA8DekwKdYsGbNGr/yGTNmeJIfcnTY45D2fMrM mTPdAX379u2aq7hvwUVBqkGDBsePHw+13Yj1582bZ++j8PjGIlaoUGH27NnO XJ2hd+nSxR7r8PgejlCAVqwsWbJkmPbbDQt7PsXrO6Nv2bKlvQ1Dy1avXr1Z s2aaViplFb744gu7xJElSxaF2uLFi3sCBkNOnTpVOYY1QyupWLHi+vXr7Q6C cy9p0aJFNuDEKGSvXr3aZpUtW9Yv37O3cf7jH/8I2m9+9UeOHKmecWK6UiO7 8RGKGl+7dm0lA7ly5XrsscfUeC2lfo6y2+XEiROtW7dWh1gdpS6xsbHqSadC xF4FAOAiKSgrkiqoOa89v8j6hw8fViJhF/MDnT17dt26dRf5x9nVhrVr17oH YLgpeu7cudM9+iLQuXPntm7dunLlyvDVDh06tHz5cudGQyrSOpcsWRJqF4z6 SqmI36Mlyi6UCZQrV86vcvhu9/ryvY0bN9obzIIK36sAACBDOnPmzFVXXZUp UybnIWKlBPYCtI4dO6Zv2wAAQIZRu3Ztj+9vuDz77LMvvvjiHXfc4fE9R+N+ OSoAAMDFOHLkSNOmTe0P8NljNcpAnCdNAAAAUsvp06e3bNlif5ENAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AADg4h06dKhTJIsWLdq0aZMmxowZk97tvShDhgzRXuzevTsFy27ZskXLfvzx x6ndqFRz8OBBZ3rQoEFq7Z49e0JVvmS7427Vxa8h4n4BAC5P8fHxnkj0I//N N99oolatWund3otSpkwZ7cXKlStTsOy3336rZZ988slUb1Wq6Nu3b/bs2c+f P28fS5YsqdauXr06VP1Lszt+rbr4NUTcLwDA5en48eNjXBo0aKDf82LFirkL N2/enDHyDZ0a16xZ85dffknBsqtWrdKyvXr1SvVWpYrMmTPrAEUfly/N7vi1 6uLXQL4BABnDoEGD9Hv+1FNP+ZVnjHwjA7s84zL5BgAgqIj5xuLFi5s1a6af /Tp16gwcODAxMdFdbdOmTW3/X3t3AltFtcdx/CoYcSlFFHcWFxa1cUsrixUK pVQlsagoivgsBEjdqIABw2sUFKxY1AQUsAKJkYKCdSElBSWVPJVFNGqkKLRU tAiVp5aGpbW23vfL/YfJvLvM3BLQSL+fEDNz7pkzZxbn/5+1EyYMGDCgb9++ OTk5W7Zs8ZjXN998M27cOLVcVlY2duxYtZmdnV1aWmo/qZ3k5OQhQ4YUFhY6 4Wb69Oma5Oeff3Ya2b17t0pmzZplo83NzUuWLLn99tuvvfbajIyM3Nzcbdu2 OZWfe+45Vf7xxx+dkq+++mrmzJnp6em9e/fW3L/88stYvS0vL9e0r7zySjwz irRjxw5bM9dff/3IkSPnz5+vFtwVNGu13y9Eq27Tpk1hLcSq8N1336nc7nxp 4Pnnnw8eicsbN27UmsnMzOzTp496+MUXX8RanGeffXb8+PGHDh3Kz88fOnRo SkqKtkjY2tBqX7hwYVZWljbuY489tmfPHi3RlClToi5v1F5JQ0PDiy++qEa0 3rT2VF5fXx9/C77LZZtmwYIFI0aM0CyGDRtWUFDw+++/e2wa47vrvvPOOypP S0tLSkoaPHjw7NmztSz2k/Yi9bCkpGTOnDk33HCD1qR1yXejA0Cr5Z1vtG/f /pRTTtHASSedZIFAR1GnztKlS0877TQVnn766QkJCRpo27btvHnzYs3L2uzV q1ebNm2cZ0XUvuJRhw4d3A+QKLGxSa644gqNKhI5jehQrxIdz21UcdAmOfPM M61ZdclymGDE8xvr168/9dRTrY6dR2sSpTpRexv2wIP3jMJUVFScc845tt60 Dm3CgQMHOhVeeOGFk08+WYXt2rXT2rOeKAeIp4KyNWdzSGpqavBIXD777LPd G0uzdsJo2OJY/RtvvNG92jWvTz75xCrU1dX17NlTheqGLcKFF16o/15wwQVR Fzlqr/bt29e9e3cr6dixow1om37//fdxtuC7XPv37+/fv7+Vd+rUyeoo8aip qYnaT+O76yq7dtaJfrJhrS77VamIRi+55BKnt8uWLfPd6ADQmnnnG3aM1amx oo9yACuxywU//fSTIq9Gdd7X1NT0559/Ll++XMdw5Q/bt2+POi+nzYyMjJ07 dyro3HLLLVai89a1a9cePHhw4sSJGr3qqqtsEu98Q7FGUVgpxIYNG3Qi+dtv v40ePVq/pqenW2V3vqFwkJiYqFig9EYd1rSWQqgzUXvrDtC+MwozefJk/Tpq 1CjV1JrR6bnlDB9//LF+VSNtQubOnfvHH3+oM6+++qrWm/pmFzF8KwRj3HfQ JIWFhQcOHFA4tmUfM2ZM5OI49bW9dIa+d+/eXbt2JScnq2T48OFW4bbbbtNo ZmbmL7/8otFVq1ZZDI2Vb0Tt1a233qrRlJSUyspKjWou/fr1U4nK42zBd7ls h0lKStqxY0cwdE1m8ODBKnnggQdizcJ319X20q/qybp16xobG2traxcvXmw7 quU5lm/IoEGDVqxYkZeXd/jwYe+NDgCtnHe+oYOwDp5OYZcuXVT4wQcfaPjx xx/X8NVXX+2eKjs7W4U5OTlR52Vt6oTx119/tZLi4mI7bjsnvIoFdtZpo975 hso1rJPK6upq+1XBcfr06UVFRTbqzjdsSe+8806nqfr6eiU2irNVVVWRvXUH aN8ZhbnnnntU/4knnnBK3nzzTSVsFs7srsHYsWPdk+Tm5qpw2LBh8VQIxojL U6dOdepbiFR8j1wcp/60adOc+qWlpXZlwJZOwzrxt2TDFBQUtCjf2L17dyB0 /cq9erV9LQprO/q24LtcygTs+psSAKeC9iW7yBDrLVrfXVfphzLJsGdrVV8V lHcFj+Qb2pPd68d7owNAK+edb4Qdk++//34V2jcc7NJERkbGfJeRI0eqsH// /lHnZW0qhXBKKioqAqHL4E5Jc3Oznc/aqO/9lG7dutmRf+jQoS+//HLYhXp3 vmEBZenSpXGumbAA7T2jMFpFlkd1795decLatWvdTxT06tVLP3300UfuSXTi bPXjqRCMEZfdYXfnzp0q6dq1a9TFsfolJSVO/W+//VYll19+uYaVUkZehdi1 a1eL8g1FWw0PGDAgrI5WoMpjpWotWq7Nmzfbep47d657P0xMTFRhWVlZ1FnE v+sqO1IjCxcuVAZo94NWrFgRPJJvJCUluSt7b3QAaOW8842wzzXojFuFS5Ys 0XCPHj0CMXTu3DnqvKzNm2++2Smx2OHEUKMz0/jzjcrKymuuucY99+Tk5K1b t9qv7nzDIldYEPcQFqC9ZxRp5syZ9oSASUhIePLJJ+3pwbPOOkslYae9NTU1 djUgngrBGHHZ/aURu7zQpUuXqItj9d0PXtplgcsuu0zDL730UiDilkRjY2OL 8g3btXTWH1YnJydH5c4Tvx4t+C7XsmXLYu2Esnjx4qiziGfXXbRokfvxDO2T 9jiNO9+I/L/GY6MDQCvXovdh3fmG3e5/+umn10WIdcM6ss048w33myBq351v GMX92bNnp6Wl2aN97du3P3z4cPD/842BAwdq+P3333dP6PHmZtQPZMWaUVR1 dXUrV67USrvooossACkeqdweoQz7CpmWMRB6qDKeCsE43huNJ99w13fnG2+8 8Ubk1t+7d2+L8o2ioqLIFSh33313IHRFwrcF3+UqKSkJhK7JRO6EoppRZ+G7 6zpXKtLT05UwFBcX79u3LyMjIyzfcN+bc8Ta6ADQyh11vmHXn3XQdlf47LPP 8vLy7Jgc6Sjyjeuuuy5w5CE9s2DBAiffUPR/5plnXn/9defXqqqqM844I3Dk IRN3vmEPRYR1ODU1VRE8LAkx7gDtO6Mwr7322sSJE52b+4qeDz/8cODIUwf2 FKXzaqrRWlWhljeeCsHjnG/YvZvzzjvPnU0tX768RfmGdgYNK+y6bys0NTVd eumlkYlf1BZ8l8s+lpuQkHDw4EGngqadNWvWnDlzYt3z8t11b7rppsgKnTt3 DoTeQwnGyDe8NzoAnDB0fKuurtZhU6dppaWlq102bNjg/oSF21HnG2+99Zad 4DuX/XXMt0vQBQUFUed1FPnGkCFDVEHHbRt13q+0fGPVqlWB0Auqzl/c2L9/ v927t7dc3fmGApyGzz33XOcxQq0Wu1SuM/fI3roDtO+Mwqh7gf9/dPCpp55y IlRhYaGGL774YucEvKamxqKwxTjfCmLZjlaIjR7bfENSUlI0+tBDD9kXVyor K23jeucb7l41NDTYJDNmzHAq2EOnyhAOHTrk24LvcjU3NyclJVk/nRRFeVog 9PautlHUWfjuuvb8jHs3zs/Pt4sV9heFouYb3hsdAE4MtbW1a9asWRFDWVlZ rL9ZdtT5htiLhwoQ48eP1wHfTgCTk5PDvgnm0aZvvmGRVyVpaWkjRoxQtmBf S7B8QyfOV155ZSD0fYZRo0aNGzfO3qBRH2zysO9vWPai0/YJEybccccddio9 efLkqL11B2jfGYWx+xHqtjowadKkQYMG2dus7777bjAUJa1jnTp1euSRR9SZ 888/3+ZllwJ8K4jFx27dut13333B45BvfPrpp7aqlfYo93AeS1Acj7rIUXul PM0+iKFdZdq0afagpppSShxnC77LtX79evscSp8+faZOnZqRkWG3ut577z2P fnrvuo8++qglJGPGjFGyZJWNfQIlar7hvdEB4ASg0/Pi4mJLLXRw0xF448aN m122bdumhCTqtPPmzdNBMisrK6x89erVAdfbl8ZuSTh/Y1QnsFOmTLHXG42O zB5v/0W2WVVVpZIePXq4qyl8KGTYsI7/eXl57dq1s/aVmWzZskXHcIUhq6Ao qV4pD7EKimX33nvvgQMH7FfFoIDrqcj6+vrc3FynNVXWGWis7MjeD3WyI+8Z RVq0aFFqaqo9ZGgB1H0HQSfUDz74oH18zGJ6Tk6O8/nKeCroJN0uBaiO8hO7 HOF+ydTeLHbeTwlbnMj69vqJO/crLy/XjqEEIzExMTMzUxFcFXr27BlrkSN7 FQx9xcu+GxYIPeyqdfLhhx/G34LvcsmmTZvcj/J27Ngx1hU2h/euq8161113 OT9p8fPz8+2qiN0csVWhOmHNem90APhHU7Lx9ttvK9NYuXKlAkSs6Hn8KChU VlZ+/vnnzlc1jrnGxsavv/7a48/Kqw86BY5z8VVHCdjWrVuP4nXFFs0oGLoP otwv1pqxnlRUVHh3NVYFBc3q6mr3owvHila4AmVY6qjMQTG0b9++3tNG7ZXW gBLFWPdQ4mnBV11dnWahXVGdj3MS713X+qw8s6V/DsZ7owPAP5E72Yj1ND7Q UsrE2rZt26ZNG+dNZIXm3r17K9+YNGnS39s3AMBfjGQDx09WVpayiw4dOgwf Pjw7O7tr16525yvqs7UAgBMVyQaOq9ra2tGjR9vXwgOhb7krA2FPA4DW5ocf flgZQgjA8dPQ0LB9+3b7a2sAgNZJKQfJBgAAAAAAAAAAaM2amps3l+8pWlOu fxpo4q9MAgCAY+q/tYf/veA//5qx2vmnURX+3f0CAAAniKbm5rBkw0k5uMoB AACOic3leyKTDfunn/7u3gEAgBNB0ZryWPmGfvq7ewcAAE4E5BsAAOB4434K AAA43nheFAAA/AV4HxYAAPwF+N4XAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA0Br8D42FW3o= "], {{0, 360.}, {360., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{360., 360.}, PlotRange->{{0, 360.}, {0, 360.}}]], "Output", CellChangeTimes->{ 3.764590182223329*^9, 3.7645902311951*^9, {3.764593788459964*^9, 3.764593812183195*^9}, 3.7645938446200447`*^9, {3.772297922618548*^9, 3.772297961508634*^9}, 3.772297995314396*^9, 3.798186164969049*^9, { 3.798186245667493*^9, 3.79818634013667*^9}}, CellLabel->"Out[47]="] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell["Train a classifier model with the subword embeddings", "Subsection", CellChangeTimes->{{3.728644832191423*^9, 3.7286448555810347`*^9}, { 3.7286721153422413`*^9, 3.7286721189663763`*^9}, {3.763326351143921*^9, 3.7633263530075293`*^9}, {3.76459060814266*^9, 3.7645906083343678`*^9}}], Cell["Get a text-processing dataset:", "Text", CellChangeTimes->{{3.728651425224104*^9, 3.7286514412308683`*^9}, { 3.728665385186487*^9, 3.728665392834448*^9}, {3.732972143493802*^9, 3.7329721500516033`*^9}, {3.763741406385029*^9, 3.7637414070563707`*^9}}], Cell[BoxData[{ RowBox[{ RowBox[{"train", "=", RowBox[{"ResourceData", "[", RowBox[{ "\"\\"", ",", "\"\\""}], "]"}]}], ";"}], "\[IndentingNewLine]", RowBox[{ RowBox[{"valid", "=", RowBox[{"ResourceData", "[", RowBox[{ "\"\\"", ",", "\"\\""}], "]"}]}], ";"}]}], "Input", CellChangeTimes->{ 3.718024202169945*^9, {3.731323108382778*^9, 3.731323110187052*^9}, { 3.731323215591506*^9, 3.731323234750684*^9}, 3.731323330603273*^9, 3.751385579355688*^9, 3.7513867294381657`*^9, {3.754054059484207*^9, 3.754054060952406*^9}, {3.754054105068007*^9, 3.754054190565796*^9}, { 3.754054229925753*^9, 3.754054230353718*^9}, 3.754054287920835*^9, { 3.7633202551101913`*^9, 3.763320258603861*^9}, {3.763321824597167*^9, 3.76332182907216*^9}, {3.7633341715137987`*^9, 3.7633341955511503`*^9}, { 3.763334789758251*^9, 3.763334800324493*^9}, {3.7634677584029827`*^9, 3.7634677598328876`*^9}, {3.763467802319062*^9, 3.7634678203430634`*^9}, { 3.763486081509118*^9, 3.763486091483197*^9}, {3.763486212528408*^9, 3.763486232036531*^9}, {3.7634877208173237`*^9, 3.763487729621838*^9}, { 3.7722832483744907`*^9, 3.7722832805032797`*^9}, {3.772283314425034*^9, 3.772283341721057*^9}}, CellLabel->"In[501]:="], Cell["View a random sample of the dataset:", "Text", CellChangeTimes->{{3.763741410851025*^9, 3.7637414429457397`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"RandomSample", "[", RowBox[{"train", ",", "1"}], "]"}]], "Input", CellChangeTimes->{{3.7603044131707277`*^9, 3.76030444561474*^9}}, CellLabel->"In[117]:=", CellID->1153613343], Cell[BoxData[ RowBox[{"{", RowBox[{"\<\"looks more like a travel-agency video targeted at people who \ like to ride bikes topless and roll in the mud than a worthwhile glimpse of \ independent-community guiding lights . \"\>", "\[Rule]", "\<\"negative\"\>"}], "}"}]], "Output", CellChangeTimes->{{3.760304421753704*^9, 3.760304451984828*^9}, 3.763320269133627*^9, 3.763334199232924*^9, 3.7633342958775682`*^9, 3.763362864176012*^9, {3.763467822748149*^9, 3.7634678520747843`*^9}, 3.763486238196076*^9, 3.763486736188501*^9, 3.7634872049942713`*^9, 3.7634873193905888`*^9, 3.763746936226151*^9, 3.763815821346582*^9, 3.763914848677979*^9, 3.76408112040168*^9}, CellLabel->"Out[117]="] }, Open ]], Cell["\<\ Precompute the BERT vectors for the training and the validation datasets (if \ available, GPU is highly recommended):\ \>", "Text", CellChangeTimes->{{3.7286654345369997`*^9, 3.72866546818818*^9}, { 3.728665565970962*^9, 3.72866556673906*^9}, {3.7286656407092237`*^9, 3.728665647036696*^9}, {3.731229937302918*^9, 3.731229959199244*^9}, { 3.763321012198453*^9, 3.763321015574923*^9}, {3.763468098936027*^9, 3.763468130326207*^9}, {3.7634872175324306`*^9, 3.763487218651146*^9}, { 3.763487558606895*^9, 3.763487562404523*^9}, 3.763746665472373*^9, 3.763916465255411*^9, {3.763916501371235*^9, 3.763916504856986*^9}, { 3.76397946650238*^9, 3.763979484502833*^9}, {3.764593249607634*^9, 3.7645932505360413`*^9}, {3.773419750119244*^9, 3.773419752566256*^9}}], Cell[BoxData[{ RowBox[{ RowBox[{"trainembeddings", "=", RowBox[{ RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], "[", RowBox[{ RowBox[{"train", "[", RowBox[{"[", RowBox[{"All", ",", "1"}], "]"}], "]"}], ",", RowBox[{"TargetDevice", "\[Rule]", "\"\\""}]}], "]"}], "\[Rule]", RowBox[{"train", "[", RowBox[{"[", RowBox[{"All", ",", "2"}], "]"}], "]"}]}]}], ";"}], "\[IndentingNewLine]", RowBox[{ RowBox[{"validembeddings", "=", RowBox[{ RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], "[", RowBox[{ RowBox[{"valid", "[", RowBox[{"[", RowBox[{"All", ",", "1"}], "]"}], "]"}], ",", RowBox[{"TargetDevice", "\[Rule]", "\"\\""}]}], "]"}], "\[Rule]", RowBox[{"valid", "[", RowBox[{"[", RowBox[{"All", ",", "2"}], "]"}], "]"}]}]}], ";"}]}], "Input", CellChangeTimes->{{3.763321001930324*^9, 3.7633210232701283`*^9}, { 3.763321345701984*^9, 3.763321349467573*^9}, {3.76332189934901*^9, 3.7633219009472218`*^9}, {3.763327164559267*^9, 3.763327166391221*^9}, { 3.763362866249284*^9, 3.763362871579873*^9}, {3.7634677622719917`*^9, 3.7634677983838377`*^9}, {3.7634678284722643`*^9, 3.763467829846356*^9}, { 3.763467925333419*^9, 3.763467926827427*^9}, {3.7634680841753807`*^9, 3.763468090582642*^9}, {3.7634681452310133`*^9, 3.763468174580349*^9}, { 3.763487209094948*^9, 3.763487211476808*^9}, {3.7645916563591423`*^9, 3.7645916571771*^9}, {3.764591692154367*^9, 3.7645916973086147`*^9}, { 3.7645927999430037`*^9, 3.764592828476617*^9}, {3.764593240626279*^9, 3.764593245208138*^9}, {3.797848643564291*^9, 3.7978486435653057`*^9}}], Cell["\<\ Define a network to classify the sequences of subword embeddings, using a \ max-pooling strategy:\ \>", "Text", CellChangeTimes->{ 3.763321076457397*^9, {3.763468189619194*^9, 3.763468201308799*^9}, { 3.763487223834408*^9, 3.763487231762141*^9}, {3.764089467238929*^9, 3.764089479447352*^9}, {3.764593901317069*^9, 3.764593916444992*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"classifierhead", "=", RowBox[{"NetChain", "[", RowBox[{ RowBox[{"{", RowBox[{ RowBox[{"DropoutLayer", "[", "]"}], ",", RowBox[{"NetMapOperator", "[", "2", "]"}], ",", RowBox[{"AggregationLayer", "[", RowBox[{"Max", ",", "1"}], "]"}], ",", RowBox[{"SoftmaxLayer", "[", "]"}]}], "}"}], ",", RowBox[{"\"\\"", "\[Rule]", RowBox[{"NetDecoder", "[", RowBox[{"{", RowBox[{"\"\\"", ",", RowBox[{"{", RowBox[{"\"\\"", ",", "\"\\""}], "}"}]}], "}"}], "]"}]}]}], "]"}]}]], "Input", CellChangeTimes->{{3.7634872532320557`*^9, 3.7634872735993547`*^9}, { 3.764590804887718*^9, 3.764590806902957*^9}, {3.7645914792435093`*^9, 3.764591502803321*^9}, {3.7645915966649323`*^9, 3.764591617691963*^9}, 3.7645916825264606`*^9, {3.7645927758103456`*^9, 3.764592788670253*^9}, { 3.7645928600687933`*^9, 3.76459286062899*^9}}, CellLabel->"In[6]:="], Cell[BoxData[ GraphicsBox[ TagBox[RasterBox[CompressedData[" 1:eJzsnQlcVdX693Eer1qOmZo5ZqWZqTeHrMwGLbPUvNfx5pBDg+VQ5oioqDgL piY4IIoooILiAIKIOKAioCiIgKggk0AQ3G691X1/nOey/su9z7CBcwROz/eD fvZ+9tprr73OGn5r3M9O+Hro5Mo2NjazauK/oePnvDlz5vi5wxrg5JPps6ZN mf7ZpIHTZ3825bOZr06oAuNV/PsIB4XH/2UYhmEYhmEYhmEYhmEYhmEeC7/9 9lt2dnZqampKSkoywzAMwzBWDar7tLQ0VP0QAGWtQZjHwZ9//omf+8GDB5mZ mT/99NPPDMMwDMP8BUClj6ofAiArKwtioKz1CGNB8PtmZGQ8fPgQv3t+fn4B wzAMwzB/GVD1QwBABkAMsOSzYiDps7Oz8VuXdYpjGIZhGKZsgAzI1lHWqoSx CL/99ltmZiaLPYZhGIb5iwMxAEnAc/mskpycnNzcXB7GZRiGYZi/OBADkAQQ BmWtTRjzQ9P2yjqJMQzDMAxT9tBEvrLWJoz54Zl7DMMwDMMQNIuvrLUJY35o 95WyTl8MwzAMw5Q9tENLWWsTxvyw3mMYhmEYhmC9Z62w3mMYhmEYhmC9Z62w 3mMYhmEYhmC9Z62w3mMYhmEYhmC9Z62w3mMYhmEYhmC9Z62w3mMYhmEYhmC9 Z62w3mMYhmEYhmC9Z62w3mMYhmEYhmC9Z62w3mMYhmEYhmC9V6HJysrKlPjj jz/EJdZ7DMMwDMMQCr0Hi6wfcFoWKobRxK+//vrloyQmJoqrFVfv3blz52Ix SUtLM+Tb/fv3i+sbbjHkGx5UXN/wOpaJJ4uQkJDg7u4eEhJS1gFhGIZhzIlC 77m5ucn6YdeuXWUhZBhN/Oc//8FvdOzYsftF/P777+JqxdV70Eg/FpO4uDhD vkVFRRXXN9xiyDc8qLi+4XUsE08WYd++fTY2Nh999FGZPD0iImLRokU+Pj5l 8nSGYRgrRqH3cnNzhXhYvnz5zp07y0DHMNogvXf58mW9Vyu63oOODdOAt7e3 Fr3n6+urxTc406L38FAtvuEVWO8Viz179uDpkydPLpOnMwzDWDFG5u9t2LCB 9V55xrr1HvTSAw2cPHlSi947c+aMFt/gTIvew0O1+IZXYL1XLFjvMQzDWAjW exUX1nus98yOQu8hFfXp02fChAnXr18fP358u3btnn766YEDB8ov5eTk1Lt3 79OnTx86dKhfv35Nmzbt27fv0qVL7969K9wkJibCzWeffSY/69q1azDOmDFD PKh9+/Z4OnyAHU95LG/MMBWSW7durV69GtmkcePGPXv2nDp1anJysrhKuTIw MNDW1hZX69evHxoaasQOvL29p02b1rFjx1atWg0ePHjv3r3Ct8OHD+Ou6dOn K8IwcuRI2FEHWf51GTPAeq/iwnqP9Z7ZUei9rKwsnD7zzDNt27bFQaNGjWx0 1KhR48aNG+Rm5syZsHz88cf4v1KlSsLNq6++mpOTQ27gGJYePXrIz0JFA+M7 77yDY7hELfPEE0/AUrt27ZYtW7700kuP7a0ZpmKBxhRaXsgsVatWbdGiBeU4 SLV79+6RA8qVL7/8Ml2qWbPm2bNnjdiXLFlCliZNmtSpU4eO161bR77duXOn SpUqyPXyyjjK1FCbubm5jz0CmJLAeq/iwnqP9Z7Z0av3QNeuXRMSEgp0Ig3H sEyZMoXcUA0C0NinPj28OFVGSJ/kxqTeI3g8l2G0MHbsWOSUIUOGIIfiNCYm 5sUXX4TF2dmZHIhcaWtrK284oNceHR1duXJlyDl/f/8CXeNr/fr1cNO6dWtx 48CBA2GBJBCWZcuWwfL5558/hvdlzALrvYoL6z3We2bHkN4LDw8XbtasWQPL G2+8QadUg3Tq1Ck7O1u4CQoKoo5BOmW9xzBmJDU1NTExEYW8sKC+Rt6ZNGkS nVKu/OSTTxQ36rXn5eXBt5SUFNnSrFkzuBTbSbm5ueH0gw8+EG66desGixgO Zso/rPcqLqz3WO+ZHb16r379+vn5+cINtB/1+NEp1SBfffWVwivcBTvioYD1 HsNYgIyMDB8fn127dm3cuHHcuHHIO6NHj6ZLlCthV9xiyE5cuXLFw8Nj69at Dg4OzZs3h8ubN2/SJbTmGjRoIIZ0YadWnsVejjE/rPcqLqz3WO+ZHb16r2XL lrIbRIJa7zk5OSm8ouZ/cHBwAes9hjEryIOjRo0SE+0ECr0H5aa40ZB9y5Yt nTt3tlEh9B6YOHGiGNKlwdylS5da8i0ZM/PX1HtI27Nnz960adNjeNYff/yx Y8cOPA5ZDArNjD6z3mO9Z3ZKrPe+//57hVdPPfUU7PHx8QVFeu/ll1+WHQQG BrLeY5gSMGTIEMqDdnZ2vr6+aDpt3LixxHrvwoULlSpVqlWr1rhx41xdXQMC AlCtdOrUSaH3aJIGDemiNYdbKHczFYUS673ff//91KlTx48fv3Llivoqakxc ioiIKJGQKRWZmZlIvQcOHFi/fj1Stb+/f0JCgsKNg4MDlNK6deseQ3iio6PF 90oQMDP6bAm9d8RMlCZNst4rQ0qs9/r166d207hxYzpNSkrCafXq1R8+fCjc 0Iwjtd4bP368Bd6MYayE2NhYZJNq1arJWx4tXry4xHqPxoLF6qoC3ZIN6jyU 9R5o27ZtjRo1ULLh0ltvvWXmF2MsTIn13r///W/SMF9//TUqBcVVe3t7XNqx Y0dJtcz/ERkZ6e7ufvXqVS2Og4ODv/nmmy9V7N69+9dffxXOHqfeQwtIBMOQ NisZrPdY75mdEus9sGbNGrIg7fXu3ZvWD4q7aPr39u3b6RT54tlnn1XoPS8v L54UxDDGSUtLq1q1KnLKlStXyBIREUEr4kum98g4bNgwYZkxY4Z6PBcsWrTI pmg7F5GXmYpC6fUe2Lt3r+KqGfXe/v374dWWLVuMO/vjjz9Qt4og2draIjU6 OTnNnj2bLHv27BGOH6feA6dOnULY/Pz85O/blh7We6z3zE6J9V7jxo3x/3PP PTdo0KCGDRviGBWQWNwHJk+eTDXI888/D41Xv379pk2bKvQeKpdKlSrB2LZt 24kTJ1r+dRmmQjJgwABkkwYNGqBJNXDgwLp169K+lyXTe4GBgZTvIORGjhyJ HIrjJ598Uq33YmJiKBfjiZmZmRZ9R8bsmEXvTZ8+PSMjQ75qRr0H2aZF74WG hlJgvv322+joaDmczs7O8+fPz87OFsbHrPcsBOs9i+q9tKtXM6KjFb6VXu9l xsamXrpUmvixKB4eHja6zZPpFLkGp61atZLdxMfHy5PxxIq/CRMmVKlSBce1 atVCHXT+/Hn5rtzcXFQlYoZ57969jx8/joN3331XdoY22jPPPEPVjSVflGEq MKmpqYMHD65WrRrlpg8++ODQoUM4GDNmDDmYNWsWTlH+KG40ZHd3d6cJtza6 LZddXFxowz0IPIXLvn37yg9iKhBm0XvA1dVVvmpc7/3+++/379+/dOkS2g6G nv5fXZfdr7/+un79eni1adMmdeCFy19++WXu3Llw9s033yAjqL1SLJRQ6D2E 5/bt2xBOd+/elYd99YYcbiIiIhD4xMREOQwyhmL7zz//VDv+7bffcAn/Fzcw ltN7RyWM67qjKiyt96L37AkaOvReXJxxvZd48ODZf/4zNy1NrfcUPhjSez6t Wvn36YODrISEkE8+uX/qVIE59N6ZoUMP1KhRmvgpb8g9BtCHyB3yRnwK8vLy 4EDe6UtNfn4+Yk+e6ccwjJqcnBwUVvI3L0pJQkJCbGyskYoDl1544QXk9xMn Tpjrocxjo/R6D6IO/3/11Vey0DKk96BqfHx8FFPsli9ffu/ePdlZaGio3ml4 MnPmzBHug4KCyHjgwAEjAklAes/JyQmvv337dlRYwlvoRqhQ9S2ogPbu3fvt t9/KYcBb43WEVCMgwAyFuUCfPlyzZg0unTp1CtWcxsAQf029Fzp1qkflygmh ocb13uWvv4az9KhCtk+Zcqhv36TISL0+mNR7ycHBHjY2UUuWFLDe04ehkSOG YawJNNZQWSOzv/LKK2UdFqYklF7vQW/Y2triAFpFXNWr93Jzc+3s7Oiur7/+ eunSpfPmzaPT6dOny4OwwcHBxsUe+P7774X7PXv2kFEetDUC6b2FCxdSyCk8 wucZM2YgqLL7s2fPQtoJB7NmzZJVmWL64oULF4ql9ygwW7Zs0RgYgaX1npZx W18d6ru0POv/GSAzM/P27dsPHz78tz5wY25qqjiFrIJjFES4EYJZeAIR/u+H D3GAJBEyYwYEW8a1a3p9kElPT4dvuAU3Qu+deu018q0gM5MO8CA4wEP13q4A rwDHmUX3EmeHD4fe+5+3VgHrPYaxelxdXevVq2ej+2Kvn59fWQeHKQml13uQ N+fPn6fj+/fv01W9eg81Ajnz8PAoKFI+t27dWrBgwZe6FRZi+BJKJqeIRYsW 4aqjo2POo8gqaN26dV8+2uNnHJJYhLOzM2p5GFHL4ylkRHqW3SclJcE4f/58 CFE4/lNHTEwM6VVIQQgA4RhvoQiqt7e3Sb2nPTCCiq73fv3ll3OjRt12dhZa KHLu3Mh58yCQIg8eDP7kk5Tg4Mtffx3w2mtXZs5Mj4ggEXXb1TX0n//MS0/H cfTatSHjx0d6ep4dNSqof/8we/uclBTyKnHPHngOnXZmxAifjh2h904PGnTN 3l7hQ6HGCw8P+/xzPCVs6tRYb++4W7dI7x1q0eJk7944+Ck+Hl4lnziB42tr 1vgPGRI0dCh8oL+LU6YIGXl79+6zI0acevPN8O++y8vMFHoPgTkzeHDQW29d X7YsZMgQK9N7+LmR9RSz9RiGsSY8PT179OgxYsQIHsmtuJRe7505c+b333+n jjsoOrqq1nsRERF6e8MAhBNd0ruV3+LFi780tV4D1Q3crFq1yogbGSGxkHRl Oyr66dOnw+7i4qK4JS4uTjFuC65evUr+hISEGHkcBJsWvac9MERFX6/xn/x8 yLCwSZOE3jvWufPxl16CQLrk6IhL0EU+LVsG9O3rUbmyZ506+T/9hFR3+Ztv cCn7zh0chwwdiuP9NWoc797dTyfq/D/88JdffoFXV2fNwmne3bunP/rocLt2 OA56++3IxYsVPiQHBXnWrn2wYcMzH3+MZxWGZ+NGofeO6r4HkX7+POwxGzbA GD5vnt+rr57o0+fU668fad8e9iNt25LeC58zB6e+rVsfe+klHOD/TF1v4VV7 e5z6PP108KBBeBC9lzXpPYZhGKb8U3q9FxwcjNNLly7RaVJS0n/16b0DBw5Q V1h+fr7aN+ooO3r0qPqSFr1H0+oQYCNuZEhirV69Wn2JhKveS4I///wzKyvr 5s2bp0+fprf29vY24l6L3itBYCyq9/BbBAQEnNIMHMtdglqeZVLvnX7//YL8 fKS0Gxs24DTJx0ev3ru4ahWN5x579VXPevXuxMf/+uuvpPd+Tk5Wj+fKPlz6 6iufFi0yrl/HcX5urveTTx7p1k3We4mJiamhoULvifHc/Lw8/1dfhRC9HxBA uhFuQkeNogBHLFyI02hn59jISK/69fEIGlxGeHxbtWK9xzAMwzxmzKX3/vjj j2XLln2pW0grNIys99auXUvu1+mDLuldz6tF7y1dupRGhI24kTGyHwstB16x YoX6Unp6OpQbXmTGjBlfPoq601JGi94rbmD+a2G95+/vn5aWlqMZOMYt5tV7 iV5epNAyIiNxGrFggVrv7a9RQ8zfC505E5euHT9+//79cN2xSb1HZN64cdfP L2H//mPduh2oWVPWe/D8hq+vWu9d0fkZtWQJ+RBpZ4fTW87OqRcv4i/R0xOn 57/4IlJ3cOHTT614/h7DMAxT/jGX3vuvNLgZHx9PQkXWbyTbjCOv+FDcaFzv 0U7L06dP17ihsRGJhbdWS6w///zz6NGjIpzffPPN6tWr8XY+Pj4W1Xt6AyOw qN47deoUVNx/NAPHuMVCei/z5k2cXp03z7jeO//994V679gxWEKnTdOi93If PPDv3Run+DvYuPH+atXUei/S21uh967v3l04QKzbOI78pJ5GxV/wsGH0InFb trDeYxiGYcoQM+o9iCIoE+ri27x5s97+vdmzZ981jPq7bP8t0nvw0FBIwOHD hykwemcAqimuxDp48CD5v3LlyvDw8F9++UVcor4+1nsl1nsXJ04svd5LTExM Sko6+913Qu+dnjSpUGhdvoxLRvTeuTFjcBy7detPyck4DXz/feg98u3g00/L eu+8rS2MuHTjzBmvBg0OP/30T/fvix5C8ifxwAHq36O/5KiocJ0yjFq4kPUe wzAMU4aYUe+B69evk5G2FpH1Hn0WDdD6U+2QhjS+FiMuLo48X7Zsmd49jRUU V2LNnz8fxjVr1qiXbFir3nsM47kQPD5PP32yZ08SP//Jy8NpyfReamrqw4cP Q1V6L+XmzZSUFNJ7YoWv7ANk26GnniJ7fm6uX5cu0Hu4Bb4p9N4Ve3sYU+7d O9K1q0eVKslBQfLWK9dXroSbSDs7YUkJCcGL3AgN9ahUKaBPH3pHpB8cs95j GIZhHjPm1Xtg9erVYtxT7/pcqKZifUaWdnGZOXNmgYGPWcjOSHop/L937x50 oLy/X7EkFt6UNt87ePCgwvGlS5fokvXpvcewXgOC59LUqZBDEd99d9vF5dTr r0MymWs8N2z6dDGee3nr1sKOxM8+u3/qlMKHoPfeK5xfN2nSjQ0bChcC68Zh 0+/cMTSee1k3LfBYz563XFzEX25aWl56+uHmzSHkzo0bl7BvXxjey8YmeuvW QuWpG+rFmybu3Rs6YgSvz2UYhmEeP2bXe2JzlS9V6y+cnZ3FqGhsbCzqOxIt UVFREIG4qvdB/v7+dBcUnQgqJMG5c+f++OMP4Sw9PZ12LyHtFBYWlpqampCQ cOzYMeqCw/8FRXKrWBLrzz//pPW/33//fXx8/H91XwmBz1u2bBFvan16r5Ro eRYSQNb16yFDhuyvWrVwK5VXXz3Sps3xrl2F3rtz8KCs92i9xhWd4sq5e7dQ 7w0b9ojemzeP9N7Dhw+vfvstjvMfPIDei42KOt6jR+EjevVS+PAwJgaeeNaq VSg1X3454O23C3Wmv3+h3mvZ8mjPnvA8/uTJwjHfjRthDOjXTz1PLyMqqnBR SVSUf58+kK+wHGraNNLOjvbfS01MPPP++2T369Qp8PXXD9SsyXqPYRiGeZyY Xe/9t2hVqVrv5efnQ+kJjQR5RvvmEV999ZXeD0ngQd99992XEmJ5bGJiouzy 1q1b4oMdCmbOnBkeHi5clnj+ngI8jr7ba5V6T/uuyyXWe/+TPenptF0JYfz7 Ggrk72skJSXhOC0t7f9JQO/BiOZAdnx8ngE/8zIzc+7d+/ej39e4f/8+jlOK NnAmTH5fA15lJyTQsfx9DZz8/KhXrPcYhmGYx0aJ9R7EBg1lqvUe6jhDvV5/ /PHHmTNnaDqcLMZcXV3lz+8qyMjIoI/MykAFxcXFKVzidXbt2vX9998LZ7Nm zXJyckKtLTsj32jrGPVbqyUWgn348GF5G5YFCxZ4e3ujTndzc8Opl5eXocD/ 15TeK25gBFaj9xSURu8lJyf/9ttvevWeFt8Ueu/evXsK30r/PTXWewzDMMzj p8R6r5Tk5+dDrcXExGRlZWlZYfFf3ed34R51OmphvZs2y6DKjo6ORu0vj/mW kl9++eXOnTvwVuNXei1N2eo9+UtqJdN7hrh48eKPP/4YFhb2QAMnT56EY6Ql 3Hj8+HG8tcK3qKgoOEArQ4tvcAbHuAU3BgQEQLApfMOD4AAP1eIbXgGO8Tql iQ2GYRhGkJOTU9ZBqKiUld5jSg/rPYXeU4u9glLoPb2+la3eu3379tmzZ/GL m8tDNQkJCe6PgpdFO8tyTyxD/Pz88ILp6ellHRCGYUwAmWdvb9+tWzcbG5sp U6aUdXAqJKz3Ki6W03tHVRjSdeVK7+mlxHpPL2Wl9/BT+vv729nZLV68eP36 9aX30BD79u2z0UfXrl29vb0t91xL4+TktGjRoqysLNn44osv4tUsqp8ZpkKj N+OUCR4eHsitVatWHTBgwLZt28o6OBUS1nsVF9Z7fxG9B6+cnZ0XF/EY9F7H jh0X6Jg5c2afPn2qV68OY7Vq1Q4dOmS5R1uUTp064RXu3r0rG1nvMYxx9Gac MmH48OEICTRJWQekAsN6r+JS0fdjMQTrPQUODg6QeciPISEhj0fvffTRR7Ix Ozt7woQJsNesWdPf399yT7ccrPcYpgSUH71HIYmPjy/rgFRgWO9VXFjv/UX0 3tKlSw8cOJCTk4Onl4neA3l5eR9++CEu9e/fnyxIYH369IEOPHv27IgRI5o3 bz506FDhPjo6esGCBf369WvatOnAgQNXrFhx//59cVXcm5yc/MUXX3To0KFV q1affvqpr6+vOki7du0aO3Zs69atn3/++alTpx49elS+igjp3bs3wiAbZ8+e DWNERESBbkAKxzVq1EDgu3fvjmMPDw9ypkXv3bp1a/Xq1XiFxo0b9+zZEwFA mOnS4cOH4dv06dMVt4wcORJ22VvExvz582Fs1qwZYmnTpk1y3qQQBgYG2tra 4hH169cPDQ01EiTmLw4lPPoGgczw4cNhv3r1qrAg74wePbpFixbdunWbNWuW OqnfvHlz0aJFyKdNmjR544035s6dKxapGck4BZozuN7CQS9GsjkyoCIkxsdz KeTe3t729vZ9+/ZFphsyZAjKduMBMJLTCwxnUiOZFwGYNm1ax44dUbgNHjx4 7969wjeNRUdsbCxKxc6dOyMCR40atWPHDvUSwmLBeq/iYgm9Vx5gvafgxo0b 4ullpfcKdFUDLjVo0IBOs7KycNqwYUMUcThAaYyiUoQTBRRN/EM9Qgc9evTI zMyU723fvj0pLvhZtWpVHFSqVMnT01N+6Lx58+j2evXq1axZk9zIJSeKQRhR fsp3ISQwUgm/bNmyli1bVq5cGRaU/Dh2dnYmZyb13t27d59++mmaNYRKk0KC AvzevXu4eufOnSpVquDF09LSxC34seAGVUZubi5Zbt++TbFRvXp1BIA8od2Z CFTcsLz88st0Ca+pkK8MI0MJr3bt2iJDAcg8SuFompEFTQxKUZBklHeQW69f vy5uiY+PF6kal+gA6ZB684xkHI0ZXG/hoBfj2RxXFSFZs2aNEd/effdduGzb ti3lROGzrIQVGM/pBYYzqSH7kiVLRPzUqVOHjtetWyf/gsaLDlQfTz75JCyQ i5B8iBAcv/XWW0Ze3CSs9yourPf+InpPfnoZ6j1A5c+1a9cKiop0KhXRpBWJ LT8/Hy102NGiR7FWoOsKoPLwk08+ITfiXpSEfn5+uDc9PX3OnDmw/O1vf0ND m5zt3r2bilAfHx/UYrjL0dGRhJOIVZN6jyjZeO7YsWPhYMiQITRfPSYmhm4R FR89SJ5ThFoSls8//1zERpcuXWCZPHkyeYKQUx3k6upKbqjKALa2tnInCcMY ghKe3PBZuHAhLN988w2duru747RRo0bU3ZSdnY00SbmVHCBlQkXA8sEHHyQk JBToBMZLL70EC3Ki8FadcYqVwRWFg160ZHO9ITEEaVcEEgHD6ZUrV5599llY oBsN3WIypxvKpHrteC7UKeQczX7JyclBoQ03rVu3FjeaLDqmTp2K0xkzZiDC C3SFP0lrkx2VRmC9V3Fhvcd6z+wY13tvvvmmECqiSI+MjJTdREREUMkmdz6g /ETpXatWLep8EPe6uLjI96L1CqMYrKEZ2hs3bpTdTJw4EUY7Ozs6tajeS01N TUxMlHfmQamIWyZNmkSnbm5uVGMKB7RfhBjTodhAxUqFNrFr1y4YR48eTadU ZYi6kmFMsmfPHqSZYcOGCYsiMVPeQXYWDpACO3ToAOPNmzcLilJmixYt5C4m FFnIO/JuJ+qMU6wMrigc9KIlm+sNiV6gtaib7vbt28JIYvjjjz82dJfJnG4o k+q1IxLgW0pKimyhvn1SyAUaio7BgwfjdPPmzcIBPEQVAOluPAaMwHqv4sJ6 j/We2TGu97p3746rtEqXinQ0ORVutm/fDvuYMWMUduo6uHTpUoFUHaCYld2s WrXKRhrrbNOmDU4Vu/8dOHCAWuJ0alG9R2RkZPj4+ECkoUoaN26cLNVQ9jZo 0ECMy9CQN56liI3XXnvNWWLp0qUwvvDCC+SGqgxFfccwRqCEV7t2bZrQRRqs a9euwgHlHVTicsIjReHu7g4HaGrJKdkQ6oyjPYOrCwe9aMnmekOiF2pMQSzJ xnXr1sE4bdo04/cayemGMqnxzHvlyhUPD4+tW7c6ODjQCDiJ7QINRQd1e1ap UgWlMYJkloUqrPcqLqz3WO+ZHSN6Lzc3l6ZM05wWKtJbtmypcDZr1iwa3VDY hw4dCjsKLnEvGrwKN97e3rC//fbbwk21atUUbsLDw210c//o1KJ6DxEO/8X0 G4FcS1JHBI3L0IgM5JwiNvQiXo2qDFQKhoLBMGoo4ZF4W7RoEY5Xr15Nl0R7 Si8oQwoM51MF6oyjPYOrCwc1GrO53pDoBXUinKHlKBtpuFYMzqoxmdMNZVJD 9i1bttBwuQKh9wpMFR2owe3s7KAJxb0o1kq5kov1XsWF9R7rPbNjRO+dPn3a RpqCYqhIR1PXRjddTWHv1asX7PBE3Iumq2JsYtOmTbCPHz+eTps2baou4Y8e PQrjgAED6JT0nmKVR//+/c2i94YMGWKj6zZBwevr64vClt5O1ntBQUE2ReMy 3bp1q1SpktwSp1GhYcOGBakIDg4mN6z3mBJA+XH48OE4hrSoWrWqnLxpjuiR I0fUCY/Sp5OTExxMmDDB+FPUGUd7Btei9wq0ZXO9IdHL3//+dxEMwXPPPQcj LdjXi8mcXiy9d+HCBZQDtWrVGjdunKura0BAAAoZCr+s94wXHURaWhoKt+nT p9NwMPxEzWI8BozAeq/iwnqP9Z7ZMaT3UBA988wzuDRy5EiyGCrSz549CzuK L9mYmZlZr149CDwafhL9D4hJ2dm//vUvG90gFJ2+9957OPXy8pLd2Nrawvjt t9/S6WeffYbT77//XnZDlZ1a7ymKU+N6LzY2lnoe5CoG8W+jGgVr27ZtjRo1 8HPbqBbQIQwwoirU+wiC9R5TMtq1a1enTh2oC6Sf999/X770zjvvwKjYvEjm 3LlzNrpJBXI1kZKS4ujoKC8iUGcc7Rlco97Tks0LtOm93NzcmjVrQvrKHwRJ TU2FlKpfv748h1ZGS04vlt6jsWB5DX5OTg51Hsp6r8Bo0aEAPrzxxhty8VgC WO9VXEjvQU78XoR8lfUewXqvWKj1XlJSEoyoF2x0S97EGjRDRTrKJRqDkMcm /vGPf8DSs2dP+V7w7LPPik2uDh48WEmHiDEa42jRooUoh1G1oYULo4+PD1mo mwINczHtnO5S6D1aP7hnzx45qMb1HjykXWKuXLlCloiICNq0QaH3aDSNHrF9 +3b5Et6UtqSQZ/jMmzevdu3a1DNTwHqPKSmU8ORZeQLaD6Rjx44if0HVtGnT BgkP5VKBbriwUaNGcPPdd9+Ju0aMGAHLqFGjhEWdcbRncI16T0s2L9Cm90jE Isyy0c/Pz0baOFSNlpxeLL1HRnk1zYwZM6hQUug9Q0UH6m5IwSZNmiAqyAKx Cklvo1rjViwUeu+PP/4Q4gHVCuu98gzpPRlUzeIq6z2C9V6xIL1Xt27dzjpo KrXQZrRvA2GkSPf396eZfh06dBg8eDBNVG7WrJmYkk33okivo+Odd9555ZVX aIMpmlxE5OXlDRo0CEZUUmj89unTh4rl2bNnCzfR0dF/+9vfyLeBAweShKMR IlnvUYu7evXq7777rtgPnByjXO38KN27dycHAwYMsNFtDzhkyBB4jmihKlKh 9/BeFEVwIC9aJI4dO1atWjVcxTuOGTOGprXjRURJznqPKRnUMQWeeOIJyDD5 Um5uLnXxNWzYcOjQoUi9JNLkfY+Dg4NJViGbo4n31FNPURo+f/68cKM342jM 4Br1npZsXqBN79Eg7GeffSYb7ezsFLJWjcmcXiy9FxgYSKUZhNzIkSNp+xra zEqh94wUHV999RXFyfjx4/EUaGkb3e58ijVuxUKh99BGkPUD673yzJ9//onU ckPi119/FVdZ7xHWpPfi4+Mhhyy6kJM+Si6D+gKlFtqhYicBIjs720a3F6he fw4ePIgGNdUmqIxQKQh5UyBVB2h6t2vXjh7Uvn179QZZDx8+nDx5spCdXbp0 WbBggcKN7AnU4/Lly2lcOCQkRLjBD/fhhx9WqVIF9rVr15JR74RqG91WYOQA RStCTmrNRjfT5tChQzb6Fif27dtXr53Yv39/v3796DPEqMt69epFfSwETYBH UtF7L8MY4fXXX7fRN5sOpKenT5gwATmUUi8EzMKFC8U24ISvry+kDuVTSDik zPDwcNmB3oxTYCqDGy8c1GjJ5qSaxAbIeqHPPiqy0scffwzjgQMHjNxoMqcb yqSG7JBSpJ9tdOuUXVxcaBGZYhlygeGiAzLY1taWZCcJP6jiUq7SVei9tLQ0 WT8gEspCyDBmoKLrvb1793ppwNXVVYve2717txbf4EyL3sNDtfiGVzCv3qtw 5OTkoBhRp0NF8z8hISE2Nta4V4mJicb3Ik5KSrp+/br4uIBeUA0lJycbmsZj CLwFkoS8TZkCvCCNd584ccKIP6jUIiMj5T2+GOYxAI2haKwpQApHylRIQRlD GcdQBi8xJrO5RTGZ04sLlWxG4kdL0YHfDiWbWSLZyPw9pkJT0fVesTCp94qF Sb1XLP7Kes8QxRruKedAYS5fvpyGa8s6LAzDVBgef9HBes9aqbh6D42sjGJi pFcHjdbi+maknYsHFdc3xbwapsCK9J6rq2u9evVolNbPz6+sg8MwTMWgTIoO 1nvWSsXVe4zVA9k8Z86cFStWlHVASounp2ePHj1GjBhhfCSXYRhGpkyKDtZ7 1grrPYZhGIZhCNZ71grrPYZhGIZhCNZ71grrPYZhGIZhCNZ71grrPYZhGIZh CNZ71grrPYZhGIZhCNZ71grrPYZhGIZhCNZ71grrPYZhGIZhCNZ71grrPYZh GIZhCNZ71grrPYZh9PJzTk5qWJj4yzT18WKZ3MzMvKwsy4XNavgpJSUlNFT8 ZSUklHWIzM+9e/fM5RVqKyMfNlKQmpqanp5urkeXfxAzxr8MrhHWe9YK6z0r A7XH2bNnPTw83NzcAgMDjX8AvcSgVop3d7/n76/36t1jx3A1NyPDLE/BX77h JJp66RIc3D1+vJTPMkLmzZtR9vYXP/ssdMyY8DlzEr28jITHmki9eNHDxgZ/ p/r3Pzdu3PXVq8lOv0tySIihG6EMvf72t4MNG0LMPK7AGsNkgMuQ5ODgCxMm hI4ahRhDVF/59tuyDpGZ+eKLL2xsbBYuXFh6r6DfnJ2dg4ODtThOTExcvHjx kiVLHjx4UPpHlx+MlPAXL17csmVLcnJyKR/Bes9aYb1nTcTExNjb2y+WQHF3 5coVsz8oft8+1E37q1RJPnNGffXYSy/halp4uHYP0yMiIhYtuuPjo34K/hK9 vQ3deKxLFzg4/vLL2p9VLC7PmOFRqRIFQ/wdadfuYVychZ5YYvTGYWkgvXf0 uecUdvpdznz0kaEboWEoojKuXTNXYEqDyQCXB67On19R9J4uoS3y0ZbS3n77 bei9f/7zn6V8KGSbg4PDvn37NFZYN2/epDIQwq+Ujy4/GC/h8/Pzvb29ly9f XsoOVdZ71grrPasB5eHSpUtRAri7u9+4cSM6Onr37t1UJpS+xadAKDEon9zM TMXVEui923v24JaLkyfrfUrQwIF670o+fZocWEjvXVu5slDWVq0a+Pbb11ev jnN1vTBx4uHmzWH0e+GFn1JTLfHQEqM3DksD6b2Tr76qsGuRT4kHD949dsxc ISklFULvIYFVFL23Z88eSLjJ2lJaQkLCjz/+WEoFgnpqsw7ttRXET1RUFErC 0jy3XKGlhMdbu7i4bNy4MTs7u8QPYr1nrbDesxrOnz+PjL99+3ZkebLk5eU5 OTnBGGLukSyhxPAHCaS4ana951GpUmZMjPqu0FGjLKf34j08qGcv1tlZtuck J/s++2zhKOebb4qoLg+UK71XrqgQAbZWvWcWDhw4gHIsRl8hQJhl3pqFPDTk VWRkpI+Pj6EyJCAg4Ny5c7JFYwkPgQ2Lm5tbiUsn1nvWCus9q+H48eNr165V FBEHDx5E3sf/5n0WVaDHunQ5ULOmerxVr96L27UrdPRonxYtTnTrdmXWrLTL l8mOksW/T58j7dvjlkNNm/r37n26qDfvf6PG1arprQdz7t07UKPGgerV1XoP 4QmbNu1ox44+rVoFDx58e+9e+epNJyc8Jfn06TuHDgX064eHBvTtG7V0afbd u7KzwAED4HPYF1+oXz/10iXP2rVxNT0iolh+gozo6Kvz58PxoWbNzgwdenPT Jnk2IPlzPzAw0tb2ZM+eXvXrp4SG0qWHt25BEgQNHHiwcWNcujh1ak5Ro95I HIrIPzd2rG/r1n7PP48bk44eVUeI+qGl0XsBr712+v335eCdnzAh4/r18+PH H2nX7vDTT+NF4H+xIsd4JMgPKpzhNGLE4ebN4YnGABv3/M7hwwjVpenTFbeE jhwJu0jMBYbTOWHk97WE3oMY6I3EcPr0ihUrXn311SZNmrz99tsXddHu7Ozc vXv3Ro0awY6rihu9vb2nTZvWsWPHVq1aDR48eG9RDkJ90adPn/bt20PvNW3a FJ4PLEpp9KzAwEBbW9uePXvWr18/VPdq27Ztg33//v2F0RUa2r9///feey9F mti5detWGGfPnm3oLWgm3ubNm9WX4uPjd+3atXLlyiVLlkAIBQUFZRRNG87N zd2wYcMPP/xApzdv3txgALkr7Pbt23j3devW2dnZIeQnT57EKysCk5SUZDza /f394S3kGeTc+vXr4RXeMaKorBDs2LED7+Xp6akWZidOnMAleCIbtZfwiAoY 8crGw2kI1nvWCus968bFxQUZX+MMZ+1QBRryySfRa9eSxpCFjVrv0dwkckkS 8WDDhqj9C3SLQKHKvJ94AkaIKJ+WLXG7/BTUntB1B5s0+fnRgvfa8uXUu6jQ e1FLltCzcItnnTp0HL1unXBweebMwtr/44+p5/Bgo0bkBsIGgRHOEEIYMw2U mRBUuHprx45i+fnw9m0aDoZMhaQhN5e+/FIRNrwOXUJcQbrAjuiFRqLBZWgJ ugpBm6MbJjMSh4WRP28eufeqV48iHyGUNbChh5ZG7xXGf6NGdJyXlYVTn2ee OdK2Ldn/96AaNTKl4TaTkWM8EsSD8MNBR5H/pHu1BNi459l37uyvUgUe5qal iVsQ+MLHNW78c9GKUSPp3HhUF1hG782cORPC7PXXX8f/TzzxhI2OBg0afP31 1zioU6dOzZo1yQhNIu6CdiIj9CHc0PE6XQ7KycmBAiSvateu3bJly5eKUho9 6+WXXyb38Pms7tXmz5+PU8geckbT+T799FM6jYuLq1u3btWqVaGODL3FkSNH UIgdV63Jgp6h8U2oIOrmAps2baL+NAQVp3BAjm/cuLHuUZYtW0a3ZBUtJL9z 5w7NjkMMQEPS1T179gg9dv36dTJCFhqJ9h9//FHMr4OWFhPtbt26JTuDzoQO xCUvLy9Z8kFkkthLkxKbIfSW8NC95K3J2/XCes9aYb1nxaBFiVyPYs3sK9SE 3kMxRf1gwYMHi6sKvRfv7k61PHVl5GVnh33+OVWm4hYj47mXvvoKkg8Hca6u 4hKe69umDXTLg3PnZL2XER3tUbky6mVaOwwhFL1+PRz4tm4t7qU6t1BJjhxJ MjU1LIzqeqEuMmNiSCAZioHwOXMobNr9RJhpdQlek/YqgaCCWpBfTfgTaWub c/++eNy5sWMLFcuQIXQjguf34ouKsWa9cXh7927SFXd8fPJREWZl3XB0JE0l utcMPdS8eq/wZ+ralfYbQUrAcWFop0zRHjkmI0E8CEkLjxB9g1oCbNLzIJ3C j9u5U9wStWxZYQ/w55//7yka0rmhqC6wpN6rUqWKu7t7gW62v9Bjc+fOzc3N TU9Pnz17Nk6ff/55uiU6Orpy5co1atTw1+UgqCZINThoLeUgveO59Cxga2t7 X3o1hd6Lj48nuXjixAmcvv/++3SLkbdwdHREOaaeibdv3z7YDx8+TGIpOTmZ xBVeoUCl9xTgKknEwMBAsqAqXLNmDckk6tNDUMnDM0Wr0i5fvkzi7ZrhhUjI ZKRCIdho1iJkJPmsXuECqbllyxZZ8hVL7Bkq4RFy2FcXLagvLqz3rBXWe9aK aKueOnXK7J4LvYfjrPh47yefxGnMli10VaH3QoYPxyluEbejZDvaoYPce2Zc 7yWfOYODgL59xaUkX19YTg8ahEpT1nuQNNmJifI2ILBQZ1F20cYFVOce7dQp TxrHuR8URH1Q/3u0hwdO/Xv1MhQDt93cCoPUr592P9MjInB6rHNnuS0ft2uX h64PU/aHIlbmp9RUvJfcw3ljwwa4vDBp0v8FSV8cUuTf2LhRNlKnaGRRl46h h5pd78ldvtFr1sBy6o036FRL5JiMBPGg9MjI4gbYpOf0iwd/8IFwcKJbN1jE gKyWdG4oqgssqffGjRsnLC4uLrC0aNFCWJKSkirpIJEDuZKYmCiPt8LSrFkz 3CW2/jCi9z5RvZpC74Hdu3fD0q5dux07duCgZ8+exifLUTmmHkXdtWsX7BBI wgL5BJlEG/QZ13ukFeW+u7i4OBJacgqkKXNubm4iKi5evBgWFmZkahxNn1Os kvPz84MFAVa7F5LP29u7WGLPSAmfmppKYShZ5c56z1phvWeVoLlHLUoUa5ZY UyDrPXB7716cetatS8NzCr1X2BEHybFhQ6yzs/ijujJe1+1QYErv4Zi6g9KK dh4IHjwYp3cOHVLoPQFcQrPFbt163cGBRgkVda7omhPQIOBPupby3WPHSH4Y ioGYH36Qa38tft7avr1QIr72mhwPUUuXeuhW+8r+KOSZIDcj446PD1QQHJwb N07WQobikCJfsdol4cAB6ssy/lDz6j1EhZwUkTyox49OtUSOyUj433hukyYl CLBpz7OzvRs0EEO6SE6k8MW9WtK5kd/XcnrP0dFRWIKDg2EZ+Oj0Tsg/GLMe 3R/7ypUraPRs3brVwcGhefPmcCDmgxnRextVr6bWe2DEiBHUGVirVq2oqCgj r5CdnU3SRb1z8oULF+gSAnn69On4+Hg5gRnRe4gEXFIsYg0JCSFjkATNjkNZ aiSECkJDQ3HLTqkfGECSGRlgRcxv3ryZ3kWj2DNewqNaJ98ySrQJKus9a4X1 nvWBPI5SC5kdbXnte9EXC4XeAzTk6t+rV/7PP8t6T3S56P2LXLyYbjep90hf XZw6FccP4+I8Klf2ffZZFHRqvRezZQt0mvpZCr1308lJ8VJUNSfrZsL8lJLi oZvKpZg0KEBI4CCiaBdZLX5emTXLUDzsr1ZN9gcyVeEPXjl01CgxHVH8Gdd7 FPnCcwFprSPt2xt/qJnn77VsqXgjWe9piRyTkaD3QRoDrCWGqV+UhnRpMDeq SE5oTOeGorrAknpvq/Q4vXqvZcuWst7bsmVL586dbVRo0XtbVa+mV+/dv3+/ atWqsH/99dfGX0HoPbV0QeHm7++/fPlyMVnuhx9+EOrRkN6LiYmxs7Ozt7e/ ++haqsOHDy82jPaC1NPTU93n5ubmBmNoUVewGl9fX3qQlha6yRIeRkMiWQus 96wV1ntWBopHaiqi0C7NFkzGUeu9n1JTfVq18tCNEir692gWVtKRI/eDghR/ WfHx5Mak3svNzPSqX9+rXr3cjAxagBBlbw+7Qu89uHDBo1Ilz1q1zo0bF+fq ei8gIO3y5aOdOqn1Xvj33yte6vBTT8EugkSbrqQYKKJP9uyJq4lFy+K0+Enj gyHDhqnjIblourUhPXBmyBBSR4jeJF9fhOrGxo1a+vcONW0Ko2KZcNLRozAG Dhhg/KGPU+9piRyTkVAavaclhmmAnjp1C5V8pUoitRRoS+flX+9duHChUqVK tWrVGjdunKura0BAwOXLlzt16mRevbd69WrSkE2bNjW5Nd+SJUtQoBlyhlIu MjLyyJEjq1atIpGTqtsYU6/eS0tLI2fhqt2iaI3Dzp0741QYX52hQO9sQ7wv jAkGPpYH1YqriKJNmzYt1i22NSL5tJTwkHnFlakyrPesFdZ71gRyNy3XQjs3 y5JfL1XrPXDP3x814P5q1WgQU+i9oHfeKawHH90DRAFplfPjx6ufIgZJafY7 hMGhZs0O1KhBCycVeo/G4ORFnT/n5FCnjULvial3BMkPVNnCAu0BC6r1PFWh etPJifptFHMCjftJsxCNzAksMKAHMmNjqZtLlm2Rixfr1XuKOAx6771CXfro QFKkra0sLcqD3jMZOVoiocR6T2MMgyNt2yLtpYaFFQrmt96SL2lJ5+Vf70Hm 4fhLKQdBONEqXYXeG/9oStOu9yIiImrWrNm4ceMJEybg0lDdnjlGoIFLxeLW Ap0skSsvlH60MBYatUCf3svLyyMHEIfqp9D3ODZu3ChrLRxDU/1koJNfjeiN lItfmk2HkOidpijEHrSomMtnSPJpLOHv3r0LN/a6FnEJYL1nrbDesxrwO9Ko gZOTU6bqmxfmRa/eK3h0VE7oPdogpXBri6IJzKhefdu08axdG/UmWSBIPB6d DVWg0ns0pZ82vgsdNYqMCr33v/nww4YJTwo/iKZvPBd/0UXTcgr3bevd20Oa 0lag222DemzwrJ+Klr+hEE709qYtAaOWLfu/p2jwE2qElLA8fevqvHl4o5Dh w2V/FHogNy1tf9WqHtL0RUQFLf6V1YjeOKRhR58WLYSSeXDhgmetWjCKL6+V B71nMnK0REKJ9Z7GGAYRixZ5FG2ocmv7dvmSlnRe/vUe3TJMykEzZsxQjOd6 eXnhtNOjKU2j3oPmeeWVV2DZsWMHiqlnnnkGx67S0ns1tC7j0qVLshHKhzZU EUtlhd6jYVO13jt06BCNgeqt8vD6tPzBz8+PLMjstM/zjqJtlx4+fAgLXj9H 2mRJ5saNGyQaZePVq1dhRNjU7oXYSy36WI+8fEMh+bSX8FFRUYaeqAXWe9YK 6z2rgYqOxbqNnhweZdu2beZ9liG9h2KCBnNlvfdzbi51fRxs2PDM0KGnBw70 btCgsP6V2vWFs991H7M40rat+GCHQu+BU2+8QZ6LMT6F3rsfGEj+wBI6cqTf 88/jmJYPK/Qeabmjzz13etAg2moP9Xu29PHxAl2nE0mj/VWqQLxBeIi94xRf FdHo591jx0grnnjlldAxYyiuoDSgwWR/1HqANr1BvEE9IgI969alkMhqRG8c 5uflITCkkwPfesu/Tx8SNlek7W1LpvcQhmOdOyv+YouqmOLqPS2RYzISjOs9 4wHWEsMFRRv1kG+KLwlqSeflX+8FBgZWqlSJdtIbOXLk888/j+Mnn3xS1ns4 IDdt27adWJTSNOq9hQsX4vStoq5RaDDy39BYZ0HRCghPT0+FnRZTAMjF/fv3 r1u3DscrV66kmX4KvRcTE0OOIZZ2PIroKIN0tLOzW6zbyQR2lJw4hiWu6GPZ FBJg6KPkiD1chSaUjbQ419fXV+E4ICBAIfYIeZMW2a69hKfZgGKrmeLCes9a Yb1nNfj4+BiabKxeNFdKaLuSsyNGqC9B5h2oUaNQ7129Koy56ennJ0ygCX6k BCIWLvz50bklN52cfJ55huSZ/BT5owa0EFjeTDgnOdnj0fUa8e7uNGvOQ7dU 85aLC+2cJtaoijWSFyZMgJArrLtr1ULt/EDfjq9JR46I3XHpz6t+ffiQ/+jQ jHY/E/bvD+jXjz4LAjHj36uX6PwpKOogjVU1zH9KTQ0ePJjkkIduCtmdQ4cK 1ciYMcbjEOQ9fHhx8mRaPeqh+ypKxIIF8l2GHmpQ7+l+F71/Yv1C4abTRetk 87KzC2VYq1ayJ1nx8R6qhdXGI8dkJOh9kMYAa4xhENC3r157gYZ0biiqCyyj 92bNmgU1JffzhISEwPJ+0adPiFatWsEoJoO5u7s/9dRTYstlFxcX6EMcy58z g2qirrkni1Ka+lnEggULYKdPRVy+fLlq1ao1a9akLfKI4cOHw8EQqWtdQXp6 ur29PTSYor8rLy/vxIkTYmPkZcuWQfgJ7URrVMWYZmRkpKHiUV7UEB4ejvel GYNQengdWYgmJSXhcdBX8n41MkhouPFs0R7aBEIF4+VHv7RSoNOrarFHkORT 7N+ivYR3dHSE0DUUSJOw3rNWWO8xjxOILkUfmgzK858ePIA+Kf2DshISMmNj 8/WlbbmPBQohPSJCPUNPQeaNG3f9/BK9vaHf9AavuH7Ck/TISEOLfw3xc05O elRUrtHtGozEYXZiomKPX+MY0nuWxnjkaImEEmPSc6QovxdeQLTc1W0XbAjj 6Vwv5e37udA5sbGxRmoHpLQHDx48LE5uhW/q6XB6jQpo65ULRT29CiDYkpOT zbjxVE5Ozr179/ROt1NMGiwNNDnQ0FVESMm+4UubMJfmm0qs96wV1nvMXw0j Y2rlys8y5396r2fPsg5IeSE/L4++4nfilVfM7vn1VavKld4rb7i7uzs4OGjZ m+6vTGZm5tq1a3fu3Fka9ct6z1phvcf81WC9pxHSeweqV7+1Y0fhzjaqLSz+ UsS5unrVq0cDzXeLpvSXnpz795ODg+8eP36ie3fWe0ZAPeXr6wsxo9g3jxEk Jydv2LDBy8urlNuust6zVljvMX81ko4cCZ8zR+9svXLlZ5mTlZAQ9vnn4k+9 m/RfikRPzxM9epwdMcL4SG5xSTl7Vo5ksaMjo5ewsDBDayWY8PBwI1s6a4f1 nrXCeo9hGIZhGIL1nrXCeo9hGIZhGIL1nrXCeo9hGIZhGIL1nrXCeo9hGIZh GIL1nrXCeo9hGIZhGIL1nrXCeo9hGIZhGIL1nrXCeo9hGIZhGIL1nrXCeo9h GIZhGIL1nrXCeo9hSsm9e/fKOggM8xcCdZb2T0ikpqamp6dbNDwlAOEv2edx HwOs96wV1ntWBkq28+fPHzx4cOfOnUePHr169aqlf9/4+Hh3d/fjx49b9Cnl li+++MLGxmbhwoXm9TYhIQGxGhISYl5vGaaiA/3m7OwcHBysxXFiYuLixYuX LFny4MEDSwdMTUpKytmzZz08PNzc3AIDA+/cuSMuXbx4ccuWLcnJyY8/VCZh vWetsN6zJlC4rV69evGj7Nmzp5SfUzTOV199BcFTo0aNMilRHycRERGLFi3y 8fGRjW+//TZe/5///Kd5n7Vv3z54+9FHH5nXW4axBE5OTsgaWVlZln4QChkH BwfkDo3V1s2bN6kYRNlo6bApiImJsbe3l4tiyE7xMbj8/Hxvb+/ly5eXw9EB 1nvWCus9qwGibsWKFShVtm/ffv369du3b586dcrOzg6WE2b94qdMXl5e06ZN bXQ4Wfv3VaGc8ZqTJ0+WjQkJCT/++KPZC23We0wFolOnTkiud+/etehTUFtt 1qG9zoKsioqKunHjhiXDpQfo0qVLl6LsdXd3x9Ojo6N3795Nqk/06SFsLi4u GzduzM7OfszBMw7rPWuF9Z7VEBkZicIEpYf8gwYEBMC4ZcsWCz304MGDKOef f/55/N+rVy8LPaWcoFfvWQjWe0wF4vHovQMHDqA0i4mJMeTA7DPiSuzh+fPn qekNUSe8QosYRnmSBlqLsLi5uQln5QHWe9YK6z2rAU3II0eOXL16VTZGRESg PFm3bp2FHjps2DCU8x4eHh07dsQBwqB2Exoa+sUXX3To0KFly5ZDhw69cuXK smXLevfuHR4eLtyghYtADhw4sHHjxi+//PKaNWtQEsLNN998Qw6QUPv06TNh woSzZ8+OGDGiefPm8ErcvmvXrtGjR7do0aJbt26zZs26fPmyIgwm/Qe3bt1a vXo1uenZs+fUqVNFS5ye3r59e7xj06ZNcSOc0aVt27bhdP/+/fLjEJ6xY8e2 bt0aShj+HD16VL6KYh+3nD59GuK8f//+jRo16tq166JFi/AU4Uaj3vP29p42 bRoiv1WrVoMHD967d6+4dPjwYTxl+vTpiltGjhwJuxxFRmLPeLQz5RlKADNn zlTYhw8fDrtcUJjMPjdv3kT67NevX5MmTd544425c+c+fPiQLlFirlGjBpJr 9+7dcYzSQNyIAmHBggW4EbkGWWbFihX3798XV4uVumgm3ubNm9WX4uPj8Qor V65csmQJJFZQUFBGRgZdys3N3bBhww8//CBeZIMB5E6227dvI2ehxLCzs0MG P3nypJw3KTBJSUmGggqOHz++du3ac+fOyUa0jvEK+F82IsAwImBGfHvMsN6z VljvWTFoM0I2oDCRS2Az8uDBA5Tz9evXz8nJWbhwIQr8efPmKdxcvHixTp06 uFSpUiXoKBzUrVsX9QIOIHiEs1GjRtGgcIMGDapVq4aDN998E/8PGDCAHGRl ZeG0YcOGeBxNFxSKa/78+XQv6pSaNWuSs+vXr8vBMOn/3bt3n376aViqVq2K io8cQ0fRQC1eEILqiSeegLF27doQri+99JL89PXr14tnIRLo9nr16lF48O6y EkMVDONbb71FsUFuACo+4UaL3kPtRjeiFqZIBkLb37lzp0qVKoiotLQ0ccuN GzfgBj+EmNJpPPaMRDtTzqEEgOSamZkpjJB5+B2bNWsmeq5MZh+oKZEjcIkO 0Gii3jy03ZAdKleuTN7i2NnZmW6Mi4uDhBNJlA569OghwlOs1IXGLIoy9bow KCUaOYW+og40sGnTJnpB5FycwgE5Rvpf9ygIP90iJh8i3mjeHfIXNKSYBS26 4BA5ZIQsLNYv4uLigrsUK02gTmH08vIqllcWhfWetcJ6z/pA2xOlekhIyI4d O1CSrFq1KjU11RIPQqGKUvpf//pXga4MxHHr1q3lgQmSSbScgWqH8PDwdu3a Uckv9B5KaZxCsRw7dgypEW1zoWQUeo80WGhoqEi07u7uMDZq1AjGAl0/3uef f07ORDC0+D927FicDhkyhIr9mJiYF198ERZReRUYGM9V6L3du3fjFPWmj48P ahz45ujoCEv16tUhfckN6T1AtRL02IoVK0gWik4Dk3ovOjoalSyqSH9/f4pq hIF+AuEGtScsO3fuFBbUbrAgijTGnqFoZyoElADktga1y0S3tskEgOzcuXNn WD744IOEhARYwsLC0NiBZc6cOcJb9XgubqRpHkOHDqV1qUixUImwfPLJJ+Sm WKkL+QilmXomHjVpDx8+TCVPcnIyTWOmsQaF3lOAqyQRAwMDyYIKcc2aNSTA qE8Pcpc8PHPmDLm5fPky6b1r164ZCbACGmpBBlSsa4P/sK9evVq7V5aG9Z61 wnrP+qBZLgRKKuPjDqWhV69eKKv9/Pzo9JVXXsEpyQ8CZTgsbdu2lRfuUQ+D rPeoVlKM1KBS0Kv3IiMjZWfDhw+HEWW+sKDY79ChA4xiiESL/5DE0MnyqM2G DRvgYNKkScKiRe9ReDZu3Ci7mThxIox2dnZ0SnpPHrpCHkSdC2NAQABZTOo9 CEUEOCUlRbY0a9YMd4ltH9zc3KimFm66desGC1XuWmLPULQzFQJKscOGDRMW asWIEVuTCQAqBcctWrSQe4kh+ZCnpkyZIixqvUc3ovUh9y6iGYWGT61atajz rVipi/rc1KXZrl27YD958qSwwFuElnqwjes90opy311cXBwsyPtyu5Um4yE3 0SkCj7YbIkH7pDvRZ3jq1CnFJZQ8VFaXn4qY9Z61wnrP+oiKikLp5+np6eDg QGWdUBFmBM1nGsER6WfVqlWwfPrpp8INdayNHz9ecS8UoKz3SKWgnSu72bFj h1rvNWnSROFVmzZtYEf57CxBqsbd3V27/0RGRoaPjw9qEAi2cePGwcHo0aPF VS16j8KjmFIOBU6dh3RKeg+/juwG8g9G/Gp0qn29xpUrVzw8PLZu3QoPafhM CN3s7OwGDRqIIV3YcRVVs/bYMxTtTIWAEkDt2rVpuh1psK5duwoHJhOAi4uL IhfoRa33tm/fDsuYMWMULqlv8NKlSwXFSV14ERJF6p2TL1y4QJeQBVCkIJsr hhgM6b3g4GBa4CbP3AsJCSFjkATNu1uzZo3JcOrlwYMH1GeITK2WiKiCKfxi zmGZw3rPWmG9Z8WgbEE7lAoTeatPs0Cz1OrUqdO7iOeeew6Wv/3tb6I3j4SN ra2t4l6aukZ6j8r8atWqKdygmFXrvZYtW8puRP+AXvDW2v1Hu37UqFFiFpyg WHrP0LPCw8Nhb9++vRwtqJ5kN//617+Kq/e2bNlCY20K5Lnf1LVIQ7o0mCvq Pu2xp4h2pgJBCYDE26JFi3Ashg61JIBZs2bpzcIK1HrP0I3UrkGTqqA4qUvo PbUoys3N9ff3X758uRjU+OGHH9DmpauG9B5aZHZ2dvb29oo1xYcPH15smBJs ZIoAQz3iXihnvbfDaEjKlhWs96wV1ntWD0o/FCbnz583o59Qkq1bt7bRTbGW oWnbVJgX6AQJTseNG6e4/ZlnnrGR+vdoNvitW7dkN9u2bbMxpfcArQE5cuRI kArRoafF/yFDhtjouj5QC/j6+oaGhqKULq7eA7QboaISOXr0qPwss+i9Cxcu VKpUqVatWoheV1fXgICAy5cvU7Ur6z2StTSk261bN9wi93OajD3WexUd5DL8 gsOHD8cxWgdVq1aVE6fJBEBd9PJKIr2o9R5lH/XmRTQJRG7raUxdS5YsQTlm aKNLCMLIyEi8yKpVq0g+0aRlvXovLS2NnMlbBBC0egLtozgVxV2dQaHavHnz Yt2OWIY22YPMK7GYtBCs96wV1ntWw969e1GsKXYAAGvXrkVhot5joTRAXaCg fvbZZxV2aEvY33nnHTqFyKTyXJ78Q5P6ZL33/vvv41QxXEJGk3oPz4JdseGJ ApP+x8bGUr+cXGEh0vTqPcXwtELvvffeezhVrLaztbWF8dtvv6VTs+g9Gm7+ 8ssvhQVVG/VPKvZ2aNu2LaR4WFiYjW5RsHzJZOyx3rMC2rVrh4SBBgJ+SiR7 +ZLJBIDyBA5eeOEFuZpISUlxdHSU1wGR3pObEmfPnoUFTQzZt8zMzHr16lWp UoXGl4uVumhIVNFqK9CJEzlsUE0//vgjXNIkFrXey8vLIwcQh+qn0Pc4IFbl gVccQ60ptmQxCUJCC3JRKhr58gjKHLixt7cvlucWhfWetcJ6z2pAAYtyA8JD 3uGKmqt2dnbm/dTRp59+avPoAj0CFUH16tUrV65MS/mQumiQd9CgQdeuXUNJ i0KYtj2R9R5NpcONEFSoCHDvjBkzyI1JvUcrbTt27Cj2yoN4a9OmTe3ataFw NPoPOVq1alWciq8dRUREUDhlvQcVZ/Po/LcCld6jMdMWLVoI6Yh6tlatWjCK D7GZRe+RJ/JUfPFSCr1Ho3i0NHL79u3Fij3We1YAJQDFpFbCZAJAFqaVRN99 9524a8SIEbCMGjVKWCh1IX8JC4RWgwYNbKT5A+Af//gHLD179qTTYqUuWpdB E/8E0FS0oYpYKiv0Hi1KUuu9Q4cO0eiq3ooPQaKFFWIZGsQerYBDMUIWlCGw oDSA54ZCC8/d3Nxwl5OTk7xiRU1UVBScIcxaIuHxwHrPWmG9ZzWgVNmwYQM1 FV1dXaEcaCQXHDt2zIwPQpGIRjoKavVoCPjggw9wafny5XR6/fr1J598knQI jfY2b96cqh55/z3aBcJGtycJHXz44Yda9B6Kd+qjaNiw4dChQwcOHEi1jGLj VpP+48BGtzvfkCFD4EndunWpmpP1HnQU3d62bduJEyeSUaH3oGkhbm102/S9 9dZbffr0ISU5e/Zs4Y92vYdgdFZBVUNgYCAFBlXtyJEjae8LimqF3ouJiaFX hm+Kqsdk7LHeswKo+xo88cQTComiJfsEBwdTgwU6EA2Qp556itKSPEWEepvR pHr33XdFv5m/vz/tw9yhQ4fBgwfTYqJmzZqJ1UzFSl3QbyjKRAYR0GIKgEJv //7969atw/HKlStppp9C7+HR5BgybMejiBYxpCN9hnL16tWw06o3WOLi4uSQ ANE8VIN3JzdQ1A6Psm3bNtmlr6/vYmlDmPIA6z1rhfWeNYEiC01smuhCoMgy 78y9giIp0qVLF71XaQ8QiBBhgeRDefvmm2+iXf/FF18kJiaSIpL1HmSSu7v7 +PHjcSMqHRzTsK/QY9nZ2Tht1aqV+onp6ekTJkygjf5sdJuJLVy4UDEZxqT/ qampqJJoK2Yb3YS3Q4cO2agWGKKaoMmHUFZkWbBggY1uhaNwg+b/5MmTaeUj RRTcyJ7QVHZFix5hs5EGgj08PGwMICovvAVVvja6RY4uLi6084z6g1N9+/ZV v4uW2DMS7UwF4vXXX7cx8ClALdkHmgQ5hVQfJFyvXr0UbT1oITSgqlSpAgdr 164Vdoix/v37041Qm8hiFy5cEFeLlboQTrRkUaAplrgia584cUJsjLxs2TII P7HjKK1+FaOl9NFJvcjLJfB2yE1UkELpIavSgAWRlJSEx0G5ybshKfDx8TH0 IMVmTY6OjsjRRrx6/LDes1ZY71kfKKtROkVHR1tom+VikZycrCjKMjMz6UMV YutUCCQ4U8xnppEmeczIJNA5epcha/c/JycnKipKnm2oBtXNgwcPxPekjABl K4+tWwj81rGxsUZyMS698MILeFlUi0b8MRR7zF8EkwkAuQN6yciyAmQxZDT1 liO4EZm99BUNbb0iK0YZCDa9Ty8xCPa9e/f0fkJXMWmwxNAmzIovbpQ5rPes FdZ7jOVAgUkfYBLTxjIyMubOnUtDosIZfV2iS5cuYrQR7Wsa/RFzZkqDpf0v z6C2Wr58Od70lVdeKeuwMExpcXd3d3BwMN4iqyigOFq7du3OnTvNqFHNAus9 a4X1HmNR7O3taZyoVatWffr0ofk8+H/37t3CzYMHD2hcsmbNmj179nz22Wfp ln79+pllTypL+19ucXV1pZmWVatWFfPPGabigtrK19cXMkmx5VGFIzk5ecOG DV5eXuVnGxYB6z1rhfUeY2k8PT379u3buHHjSpUq0ZRv8SVKQUpKyqRJk557 7rnq1avXrVu3V69e8+bNM2NJaGn/yyeI+R49eowYMcL4SC7DVCzCwsKMrJWo EISHh4vPGpY3WO9ZK6z3mMeG3pkwCvJ1WC4MlvafYRimQsN6z1phvccwDMMw DMF6z1phvccwDMMwDMF6z1phvccwDMMwDMF6z1phvccwDMMwDMF6z1phvccw DMMwDMF6z1phvccwDMMwDMF6z1phvccwDMMwDMF6z1phvceUH3Jycso6CAzD lHdQZ2nfLD01NbUcfkYH4deyH2mZwHrPWmG9Z61kZ2ef1XHp0iULPSItLc3b 23vRokWjR4+2tbX18fHBQ0vgD2Sevb19t27dbGxspkyZYvZwlh4/Pz93d/dy WGswzF8N6DdnZ+fg4GAtjhMTExcvXrxkyZIHDx5YOmBqUlJSUAJ7eHi4ubkF BgbeuXNHXLp48eKWLVuSk5Mff6hMwnrPWmG9Z60cPHhwsY41a9ZYwv8rV660 bdvW5lGg2eLi4orrFcpD+sbrgAEDtm3bRsaIiAgoSWhIcwe8JLz44osI4eXL l8s6IAxTHnFyckJuzcrKsvSDINscHBz27dunsdq6efMmFYMQfpYOm4KYmBg0 YxdLQHaKz8Dl5+ejsbx8+fJ79+495oCZhPWetcJ6zyq5ceOGKGQsoffS0tLq 1q0LCfT3v//dxcUlICBg69atHTt2hKVly5bFTVHDhw/HjTt37pSNe/bsgXHy 5MlmDXgJYb3HMEbo1KkTMsjdu3ct+hTUVpt1aC9hIKuioqJQHloyXHqALl26 dCmKX3d3dzw9Ojp69+7dVCCLPj2EDYXnxo0bSzYsYjlY71krrPesD/ym69ev R8Hi4+NjIb1HYqxr167yLJrMzEyIPdiPHDlSLN+osoiPj1c/gvUew5R/Ho/e O3DgAAq0mJgYQw7MPiOuxB6eP38eQd2+fbv4Wje8cnJygjEkJEQ4S0hIgMXN za1cfdSb9Z61wnrP+jh69CjKkG3btt27d89Cem/69Oko3seNG6ew+/r64omB gYHCgobtggUL+vXr17Rp04EDB65YseL+/fvi6urVq3v37l2jRg341r17dxwj 2EiTffr0ad++PYy4C0bcSO5RYOL09OnT8OfVV19t0qTJ22+/ffHiRVxydnaG D40aNYIdVxUB8/b2njZtWseOHVu1ajV48OC9e/eKS7D379/f0dFRWNLT0wcN GgTj1atXyaJF7926dQuvg6A2bty4Z8+eU6dOFQ35w4cPI9iINMUtI0eOhF32 FtE1f/58GJs1azZ06NBNmzbJ2ZNeH9Fra2uLR9SvXz80NNRIkJi/MpTqZs6c qbAPHz4cdpG2wa5du0aPHt2iRYtu3brNmjVLnc5v3ry5aNEi5GLkuDfeeGPu 3LkPHz6kS5Qm5Szs4eEhbjSe/SmnT5gw4ezZsyNGjGjevDnSvKHXoZl4mzdv Vl9CUxGvsHLlyiVLlkBiBQUFZWRk0CU0SDds2PDDDz+IF9lgALmT7fbt2ygx 1q1bZ2dnhxLp5MmTCKoiMElJSYaCCo4fP7527dpz587JRppjg/9lIwIMIwJm xLfHDOs9a4X1npWBgghlFMo9iD0UrRbSe6tWrULxXqdOnaioKCPO4uLiUIbT 1D7UFHTQo0ePzMxMcjBv3ryWLVtWrlwZdigcHCO0OTk5UGVPPPEEjLVr14bx pZdeIveovGB8/fXX8T85AA0aNPj6668pPDVr1iQjIkEEA7EhwgA3dIzCnK6e OnWqUqVKuPHatWtk+eKLL+Dg448/Fj6Y1Ht37959+umnaRYi6k16BOQlTc65 c+dOlSpVUCempaWJW27cuAE3EIeijxS1DEVX9erVERvkyZdffiluodd/+eWX 6RLCjIrS9K/F/CWhVIccJLIbgMyjvCZ6rtC+oOQESUbZp2HDhtevXxe3QE2J JI1LdIBESL15y5YtU2RhtLzoRpPZPysri/xEywUHyCCiZafmyJEjKM2goxR2 KCUaOYW+og40gIYSvSAKE5zCATlGplv3KAg/3SImHyLeaN4dyg1oSLq6Z88e 0QWHyCEjMmyxfhEXFxfcpVhpAnUKo5eXV7G8siis96wV1nvWBEqkrVu3ovTw 9/fHqeX0HopEqCwU0dWqVfvggw/wUPUyDQTm+eefhxu02WlhGhr7pFU++eQT 2aXewSC947kkeFCLubu7F+hmRAvxM3fuXAin9PT02bNn4xSPplvwUFRGqEoo TlD+r1+/Hg5at24tvJ01axbJSBxDQcE96j55HrVJvTd27Fg4GDJkCNUaCBjd Iuo+VGSKOYqoaGD5/PPPRXR16dKFXpk8uXjxItQgLK6urvLrA1tbW7mfhGH0 QqlO7s1euHAhLN988w2dIh/htFGjRtRRnJ2djQRJTRVygGTZuXNnWJDNExIS YAkLC0P7C5Y5c+YIb9VZWEv2J71Hj0MAjNdEjo6OKM3UM/H27dsH++HDh0mP JScnr1ixAhY8rkCl9xTgKklEMSSBChEFJgkw6tOD3CUPz5w5Q25QDpDeEy1E LUREROAW5HrFSmH4D/vq1au1e2VpWO9ZK6z3rAmUmSg6Nm7cSG1by+m9Al2x L7QWAbmCslc4QPlGskruXoAQql69eq1ateSJMcXVe/I4MprMsLRo0UJYkpKS Kumg4hoPSkxMTElJEQ5god4zsT0Cin2q1FD4U13m7e0tP9ek3ktNTcVT5EGf DRs24JZJkybRqZubG1WawgHtPyMGZCm6EAx5Js+uXbtgHD16tPz6CrXMMIag TDRs2DBhUaRkWiolZ1skvw4dOsBII4yULJG/5K5p5H0oSXnrJHUW1pL9hd6L jIw0+S7U56YeRUUegf3kyZPCAm8RWuo2N673SCvKfXdot8KCzCtnQ5qMhyxM pwg82mKIBO2T7kSf4alTpxSXUHSQeiw/FTHrPWuF9Z7VkJGRsXz5cnk+s0X1 HnHhwoV58+b17t0b+oqK7n79+tG2ydu3b8fpmDFjFLeQoJJ3BSyu3pPn2gUH B8OiGAai4SfF7hBXrlzx8PDYunWrg4MDDTPJc2YQHtRE9Arjx49XhFnjeg38 BD4+PqiAILkhSmWplp2d3aBBAzGki0fjKl5c3EvR9dprrzlLoJ6C8YUXXpBf H54bDwbDEJTqateuTdPtSIN17dpVOGjTpg0skDdyqqOWCHWhU3tKJGNDqLOw luxPeq9JkyZaXoREkXoPTBRBdAlZ+/Tp0/Hx8bIMM6L3UHRQ61ieuRcSEkLG IAmad1figvTBgwfUZwh5qZaIqIIp/GLOYZnDes9aYb1nNdDitf379wvLY9B7 gqioqM8++4xUH2RnQdEgqa2trcLl0KFDYYcoEpbi6j0U7MKiV+/RMmGh97Zs 2ULddwoUc6RpEUrVqlXlzkDCpN6Li4sbNWqUmBwokCvKiRMn2hQN6dJgrlwN UXTppVq1aoZen2GMQ6mOxNuiRYtwLIYORfeaXlB0FBjOxQrUWVhL9qcAILea fAuh99SiKDc319/fn5q6xA8//CDmFRvSe2gU29nZ2dvbK4qdw4cPLzaM9o96 CBBgqEfcC+Ws93YYDUnZsoL1nrXCes86yMvLo0ID6u6HItBsF7OXZX1lOT76 6CMb3TQ2HKOUUws20KtXL9jREhcWi+o9NP+hQmvVqjVu3DhXV9eAgADINnqi rPdSU1PpLoBqURFmk3oPr0w9J6hEfH19Q0ND6fVlvRcUFGRTNKTbrVs3hEre gobGf4cNGxakQkzwZr3HFBdkNKSZ4cOH4xitHjRn5IxGE0SPHDmiTnWUOJ2c nOBgwoQJxp+izsJasr92vVegW3WFoszQ7sQQhJGRkXiRVatWUaGHHF1gQO+l paWRs/DwcIU/tHoCjbI4FcVdnUGh2rx5MzxEk9PQJnuQeSUWkxaC9Z61wnrP OhB6zxDm7eXr0qULagr14lCoHRTgY8eOLdAtfLDRfXFDdpCZmVmvXr0qVaqI /RwKjOo9xdBqCfQejavKq1xRBVBHnKz3yNnUqVMRPNSJUImyh8b1XmxsLPXC ya+AOLdRDYS1bdu2Ro0aYWFhuPTWW2/Jl86cOQMjakO9jzD0+gxjknbt2iHB I0kj8bz//vvypXfeeQfGo0ePGrr33LlzNroZBXI1kZKS4ujoKC8+Um+hqSX7 F0vv0ZDorVu3FPafdYhTqKYff/wRLtGyK9Cn91BUkgO924TS9zggVuWBVxxD rSm2ZDEJQkILctH0NvLlERQacGNvb18szy0K6z1rhfWe1YAiNPNRUDaiJEFL Fsfm3cKddizp2bOnXMJfvXr1mWeegZ0mNqOkpTW8cmH7j3/8g26UfdOr97y8 vGweneFWUCK9R7fIU9ZnzJhB/XhC73l6etKzkB2oX6Jz5840C5EwrvfS0tIg EeFAfCwpIiKCtmdR6D0aUKN1Ltu3b5cvIbS0K4U8PW/evHm1a9emzhm9r88w JqFUJ8/KE9BWRR07dhR7RaLx0qZNG6Q6tEoKdLPLGjVqBDffffeduGvEiBGw jBo1SlgoSaONJixasn+x9B6ty1B8DRyaijZUEUtlhd6jlVBqvXfo0CEaXdVb 8SFItLDCz8+PLBB7NFVmx44dZEFJCwsKKLmIUADPUQziLicnJ3nFipqoqCg4 Q5i1RMLjgfWetcJ6z4qx3H7LkZGRtMQVrfVBgwZNmzbtvffeq1WrFokZsZTP 39+fNmLt0KHD4MGDaZUEblTskK9X70GM0WzAtm3bTpw4kYwl0HuBgYHkDwI2 cuRI2iPiySefFHoPz6XNwWjpHIp31EeKCo70HkLS+VG6d+9ODgYMGGCj2wlw yJAhCEzdunWpllToPbw4SU04UNcCx44dq1atGq6+8sorY8aMoZntcmcj6z2m BFD/s41uy0qFRIE6oi6+hg0bDh06FEmXRJq87zGyGGVt6MCPPvroqaeeogR8 /vx54Ya6x6tXr/7uu++KfjOT2b9Yeo82H0DTTGEXHwp3dXXdv3//unXrcLxy 5Uqa6afQe3g0OYYM2/EoogsO0tHOzm6xbo8U2B0cHHAMi9hyikICRPtODd6d 3EBROzyK+EQ4od6jvsxhvWetsN6zYtBmX6zbhtQSnicmJqJsp34toWHGjx+v GLlAady/f3+qL1DdoNhXDJUC0mDqmTkok6nDEPKMLDQJXG4Lh4SEqEepWrVq BaPo0nR3d6dKyka3GBBNe9qXjOod6qyQZyhdunQJ71W5cmVRo+ld7mGj2/GY HKSmpuLVSK3Z6CbpHTp0yEbf+sS+ffvqtROosPr160crhRGGXr16UTeLoddn GC3QFuV6v06Ynp6OxE9Zxka3F9/ChQsVc8mgSdCioVwMCYdkqZj5Bi304Ycf VqlSBQ7kAsd49kcOhR2P1vIKCKe9vT00mGKJa15e3okTJ8TGyMuWLYPwo8l7 BUWrX8VoKRqqhma8yMsl8HYoJWjGIJQechztPUgkJSXhcVBu6oVdAvqWpV4U 6+sdHR0hR4149fhhvWetsN5jSgOaz+fOnUOpfv36dSO7UcHZjRs3SpDS4OeD Bw/kyX4lBiV2bGysOgzIAqh3FHYEWG00Ce6KioqSdypTAA9feOEF1HGooYz4 g/dFxVTc+UIMUxrQ/BE7UuoFyRvJ0siyAmQZtDHV5UCJs78C2npF3WAkINj0 Pr3EINhohOr9hK5i0mCJoU2YFV/cKHNY71krrPcY5vGAimP58uU0XFvWYWGY Com7u7uDg4ORJlUFIjMzc+3atTt37jSjRjULrPesFdZ7DPMYcHV1rVevHo3S iqngDMMUC9RWvr6+kEmK6b4VjuTk5A0bNnh5eZWfbVgErPesFdZ7DPMY8PT0 7NGjx4gRI4yP5DIMY5KwsDAjayUqBOHh4eJbiuUN1nvWCus9hmEYhmEI1nvW Cus9hmEYhmEI1nvWCus9hmEYhmEI1nvWCus9hmEYhmEI1nvWCus9hmEYhmEI 1nvWCus9hmEYhmEI1nvWCus9hmEYhmEI1nvWCus9hmEYhmEI1nvWCus9hmEY hmEI1nvWCus9pjTk5ubu27fv22+/HT169JQpUxwcHOLj48s6UAzDaMLR0bFe vXrItv+/vXsLiap74zieZakF2UHNDLWw1IoIK6k0pC5MFDqREFpIGRlSQUZh GHawKEo7eqGQXYQpQUaIoZSYXZhmWWLnFE06aEVWDnpVxP/BB4b5j5o6r77O u/1+LmT22s8sh5lZs357z957hvuBwI6Q94yKvAeb1dXVeXt7j/p/Y8aMOXLk iA291dbWHj58uLCwcNAfp8rMzJT+29rahqh/4D/nzJkzMmaPHj063A8EdoS8 Z1TkPdimubl51qxZMln4+Pjs3Lnz2rVrJ0+eXLly5ejRo6UxKytroB1KD3LH hISEoXi0Yu7cudL/f/1n1oFBRN5Dd+Q9oyLvwQY/fvxYsmSJzBTz58//+vWr 5ar09HRpd3R0LC4uHlCf5D3gX0beQ3fkPaMi78EG5eXlMk1Mnjz55cuX3dfG xsbK2ujoaF1samoKCQnZsWOHZc2zZ8+kMSkpSW7LmzA0NHTOnDlyr2nTpkl7 ZGSklmVmZsqi/Ltbt26FhYXJ2hUrVhw/ftwytvXZv3bi5OQk/UtMldvXr18f zKcDsGO5ublxcXG+vr7+/v4xMTElJSXmVT3mvZs3byYmJgYEBPj4+KxZsyYv L8+qQxlQ4eHhnp6eS5cuPXToUHV19UALYM/Ie0ZF3oMNzp07J9PE5s2be1xb UVEha/38/HRRMqEsBgcHd69ZvXp1Z9feQplZJD1Ky/jx4729vRcuXKhl+/bt k8YNGzbIXwcHBzc3Nz1KcNmyZXKvfvZ/4sQJ6VO/aJY5SG5fvnx5EJ8NwG6l pqbqkJk4ceLYsWP1CFvJY7q2e95LS0vTeg8PjwkTJuhtGe/mgvj4eGlxdnYO CgqaOnWq3JZuLTNknwWwc+Q9oyLvwQZbtmyRj/HeTuv7/v27zClS0Nra2tmP PKZ6/D5X856IiYnRfXrV1dUzZsyQlt27d2tNP/vn+1yMNHl5efKeHzdu3I0b N0wm0+fPnzXgyVaVjk2rvPfixQvZLHJycrp7925n14bY+fPnpWDmzJla8O7d OymQza5Xr17JYkdHR3JysuVA67MA9o+8Z1TkPdhgwYIF8hl++/bt3goCAwOl 4M6dO52DkfckqkmGNDfeu3dPGn19fXWRvAf0KDo6Wt7z6enplo379++PjIws Kyvr7Jb3JBM2NTV9+vTJXCwtnp6eUiNBThYfPnyoQ6+9vV0LZPqor69vbm7W xT4LYP/Ie0ZF3oMNQkND5VP9L0fBeXl5SUFVVVXnYOS9PXv2WPXv6uoq7S0t Lf3vn7yHkUbPoNddbT3q7XyNmpoaGd3Z2dmnT5/WsWzuJCAgQM/KT01NlS0v mUGs7ttnAewcec+oyHuwwa5du+QjPSUlpce1Hz58kLVOTk66jf/P8575cCOz RYsWSfv9+/f73z95DyNKW1ubHjv3l5rueS8rK0v33lsx570HDx6sWrXK3O7h 4XHq1Klv376Ze+izAHaOvGdU5D3YICcnRz7JzWfRWiksLJS1ixcv1kXNY0FB QZY1ZWVl/c97Bw8etPoX06dPl3b9LY9+9k/ew0ijX8U2Njb2VmCV96qqqhwc HFxcXOLi4q5evVpaWvr48WMdOFY7CR89epSRkREREaGh7sCBA1Y991kAu0Xe MyryHmzw9OlT/RjPz8+3WtXa2urv7y+rzBdIaW5u1oPGLbfxL1y40GPe27Zt m2VvmvfCwsIsG+vr66XR3d19QP3rtMXPvWHkiIqKkvd8bm6uZWNRUdGlS5fq 6uo6u+U9iXmWZ0J1dp2yoWfp9valcHFx8aiuyyj1No/0WQB7Q94zKvIebLN3 7175GJe5oLS01Nz4/v37yMhIPWBbD/BWup/hypUruiihS48sssxjBQUFemqG 5X8xn5+bkZGhLfKODQkJkZZ169YNqP+goCBpkVQ5aE8BYN80znl4eDQ0NGhL dXW1bBlJ4/Pnzzu75T0dbhs3bjT3kJSUZPl9rgRFNze32NjYjo4OLaisrJS1 Mtx0sc8C2D/ynlGR92Ab+TzXy+KN6rqo3aZNm4KDgx0cHGRx0qRJT548sSxO SEjQynnz5kkGc3V1le19qzwmE4re3c/Pb/v27dqoE5C7u7v8DQwMjIqK0it6 zZgxwzJP9qd/3Xchk11ERERRUdEQPz3A8JPP9vXr1+vBtOHh4bKh5OjoOKrr 6kZaYJX3ysrKdAzKxpHUyGiS21OmTDHnvaamJmdnZz1cVsbm1q1bde9fWlqa 9tBnAewfec+oyHuwWVtbmwQt80VZ1fLly8vLy60q29vbZfowV8q8U1JSIjck elmWZWZm+vr66hSjLZr3Ll68GB8fr9f0c3FxiYyMrKysHGj/9fX1a9eu1U7O nj07NE8JYF9kkCYmJs6ePVuHhpeXV0pKinnnm/764bFjx8z1+fn5enCs7hjM ycnRPfavX7/WgpqaGhnjmht1T77Vj2X3WQA7R94zKvIe/qEvX75UVFQUFBRI xHr79u1fKk0mU21treXVvbqTmailpcV8JJ7mvezs7M6uyzjL3S0vxGdD/3L3 jx8/muc7YIRoaGiw3CX+d42NjW/evPnL1CATh4w1vSCSbQWwW+Q9oyLvwZ5Z 5j0AwFAj7xkVeQ/2jLwHAP8m8p5Rkfdgz4qKipKTk62O1gMADBHynlGR9wAA gCLvGRV5DwAAKPKeUZH3AACAIu8ZFXkPAAAo8p5RkfcAAIAi7xkVeQ8AACjy nlGR9wAAgCLvGRV5DwAAKPKeUZH3AACAIu8ZFXkPAAAo8p5RkfcAAIAi7xkV eQ8AACjynlGR9wAAgCLvGZXJZCLvAQAAIamgvb19uLMJBl9HR4e8svJ3uN9i AABgOJkjwXBnEwy+X79+8ZUuAADQL3MlGAx3NsGQ0DxP5AMAYMSSGMDOPWP7 8+ePvsryly92AQAYUWTq1xhgMpkkEgx3KsEQktdXXvGfP3/q6RsAAGAk0BM0 JABI6iPsjRC/f/+W1KevOwAAMDzJezL1SwAY7gwCAAAAAAAAAAAAAAAAAAAG 5n+vIPFz "], {{0, 103.}, {424., 0}}, {0, 255}, ColorFunction->RGBColor, ImageResolution->144], BoxForm`ImageTag["Byte", ColorSpace -> "RGB", Interleaving -> True], Selectable->False], DefaultBaseStyle->"ImageGraphics", ImageSize->Automatic, ImageSizeRaw->{424., 103.}, PlotRange->{{0, 424.}, {0, 103.}}]], "Output", CellChangeTimes->{ 3.763487429938702*^9, 3.763747613201725*^9, 3.763816413894326*^9, 3.76391550665205*^9, 3.7640819786621027`*^9, 3.764591543522355*^9, { 3.764591606638081*^9, 3.764591617969303*^9}, 3.764591804634597*^9, 3.764592923131172*^9, 3.764593338183589*^9}, CellLabel->"Out[6]="] }, Open ]], Cell["Train the network on the precomputed BERT vectors:", "Text", CellChangeTimes->{ 3.763321126516569*^9, {3.7633214392298937`*^9, 3.7633214468289948`*^9}, 3.763468410163715*^9, {3.763746670671907*^9, 3.763746671534988*^9}, { 3.763810009854807*^9, 3.7638100102440243`*^9}, 3.764092501769985*^9, 3.7640931575616817`*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"bertresults", "=", RowBox[{"NetTrain", "[", RowBox[{ "classifierhead", ",", "trainembeddings", ",", "All", ",", "\[IndentingNewLine]", RowBox[{"ValidationSet", "\[Rule]", "validembeddings"}], ",", "\[IndentingNewLine]", RowBox[{"TargetDevice", "\[Rule]", "\"\\""}], ",", "\[IndentingNewLine]", RowBox[{"MaxTrainingRounds", "\[Rule]", "50"}]}], "]"}]}]], "Input", CellChangeTimes->{{3.7633209380446663`*^9, 3.763320949020236*^9}, 3.763320988762795*^9, {3.763321410608366*^9, 3.763321453212701*^9}, 3.76332170959013*^9, 3.76333481934881*^9, {3.763363037985634*^9, 3.763363038054048*^9}, {3.763468322792994*^9, 3.7634683760863247`*^9}, { 3.7634684534895773`*^9, 3.7634684547441673`*^9}, 3.763486376282708*^9, { 3.763486523269945*^9, 3.763486633923852*^9}, 3.7645915558733377`*^9, 3.76459185639326*^9, {3.76459283732694*^9, 3.764592864324844*^9}, { 3.764593135416925*^9, 3.764593135703196*^9}, 3.76459340460596*^9}, CellLabel->"In[8]:="], Cell[BoxData[ FrameBox[GridBox[{ { ItemBox[ FrameBox[ StyleBox["\<\"NetTrain Results\"\>", "SuggestionsBarText", StripOnInput->False, FontSize->12], FrameMargins->{{10, 5}, {-4, 2}}, FrameStyle->None, StripOnInput->False], Alignment->{Left, Bottom}, Background->RGBColor[0.96, 0.96, 0.96], Frame->{{False, False}, {True, False}}, FrameStyle->Opacity[0.1], ItemSize->{Automatic, 1}, StripOnInput->False]}, { ItemBox[ TagBox[ FrameBox[ TagBox[GridBox[{ { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"summary\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{"\[ThinSpace]", "\" \"", "\"batches:\"", "11700"}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", "\"rounds:\"", "50"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"time:\"", "\"2.7min\""}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"examples/s:\"", "2284"}, "RowWithSeparators"]}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"data\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{ "\[ThinSpace]", "\" \"", "\"training examples:\"", "7462"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"validation examples:\"", "3200"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"processed examples:\"", "374400"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"skipped examples:\"", "0"}, "RowWithSeparators"]}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"method\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{ "\[ThinSpace]", "\" \"", "\"ADAM\"", "\"optimizer\""}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", "\"batch size\"", "32"}, "RowWithSeparators"], "\"CPU\""}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"round\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"loss:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"loss\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "loss"}]}], "Tooltip"]& ], TagBox[ InterpretationBox[ RowBox[{"\"3.69\"", "\[Times]", SuperscriptBox["10", "\"-1\""]}], 0.36917458142864684`, AutoDelete -> True], ScientificForm[#, 3]& ]}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"error:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"ErrorRate\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "ErrorRate"}]}], "Tooltip"]& ], "\"16.2%\""}, "RowWithSeparators"]}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"validation\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"loss:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"loss\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "loss"}]}], "Tooltip"]& ], TagBox[ InterpretationBox[ RowBox[{"\"3.71\"", "\[Times]", SuperscriptBox["10", "\"-1\""]}], 0.3706750814923043, AutoDelete -> True], ScientificForm[#, 3]& ]}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"error:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"ErrorRate\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "ErrorRate"}]}], "Tooltip"]& ], "\"15.9%\""}, "RowWithSeparators"]}, "RowWithSeparators"]}, { ItemBox[ DynamicModuleBox[{NeuralNetworks`Private`NetTrain`index$$ = 2, NeuralNetworks`Private`NetTrain`plotList$$ = { GeneralUtilities`TightLabeled[ Graphics[{ AbsoluteThickness[1.25], {{ Hue[0.083, 1., 1.], Line[CompressedData[" 1:eJw1l3lcTekfx5Nka1UpRCprC5W01/3cbvcWpn4mO6GyVdbGTJKl0GgYY2wN LYpGZM9eUk+FxhAjXN17zrnXNGQJIVsY9bvznKfzz32d1/nc9/2c5/l+v8/n 2scujZynr6en59JJT++/z8mT/ruuSsZcdv+xPrWSiPdKiVd6eG5z9lWiR6+/ JbU54zOenr7BnjdKTPYc+DB48V/k2JH/riaJ/EvDisjjt0knqn8twY95G3+6 Xceet0iEJ+FBuTV3yRT6/feSqI+GVxXh94g+1bdK3n59bR3jrCTHqf6zpKDh F69gtZJMpfqvkopFha47bOoZXw8nslbvrxtwn/nphIAniZp8y3pylH5fH12t P4a1S+uZfwMkLXPWPqy4T6h8Uhdo5iwPrLBREyo/Yoiiuf57m7fVk3Z6dcWC 4I2lyWPryUSq7w5lWne/BuE+OUz1PVAbZXzxh6lq0kb1PTE5w+JuwB6OTKB6 Y/xt18W2vFpFiqjeBIWrPkS9ragnX6neFBF/BmpfPqhnPDPUlj+6nf5AxX7P HOuvG8Zc7qRmfsyxpU3Q7+vb4bcXlra1pZo/UbP3sUDjSfNbMy+q2ftaos/e X3+7Wcyx9bDEliHy83xBPVsvK/ScO/dMaC81W8/eWKSN+XHVFjXbr97Y3WI+ q+6smu2XNdZMfmGQo8+z/bLBH89OlpX4qNl+2aDpB+3bK1YC268+OPcibZP/ HRXpTPV9kWIacaguQ0NOUH1fvPpc5LH4lYpMo/p+2FAd/6bBtJ4YUL0t5ntL t6hNeHKS6m1RWxgbPcGeJ9Opvj8iQ2e1/tuuJl2ofgDiP+w5OmaOlhRTve7+ 2WK7orX3yQyqt4PD6Fcxo6p45mcg7HosyXi4n2N+B2J4a+d4i+qO9xmIobe3 DAsco2bva48rznMSLv/NsfWwR82K00GdDghsvewR72n2ncxWYOvpgBWNO0ca KzvW2wEm4c3RK47wbD8cYLDDLKsgjGP75Yi2iOsphtM79tMR+35e9eWsd8d+ O2L9sV3t0fYcqwdHrJ11yYhbzbN6GQTXN2Mm/3CPY/U0CP4DHqkvfN9Rn4MQ rXWd1lYosPocjCT9DcfXm/OsPgfD6sZ9RfoVFavPwcjI3dFrdgpPIql+CP6K EbJuDRbIIaofggC7fWX/WKnJv1Q/BLBbc+GGnCffUv1QNKdM3PboKU8OUv1Q 2CaVJdTo1vsL1Q9FzIKwC9PO82Q81Q/Dl6TLZVpdfRVS/TBIP+yPk+7kyWeq H4aEwMavvzTw5H9UPxyLDDYvGx3IkwNUPxyPjzhIiiM48onqhyNu+vPGKbr1 jKB6JyyIqo+T/8mT36neCSam362P3M+TVqp3QoQBX5J7kCPhVO+Mp8/T30f+ xZECqndG09yImHHrNOQj1TuDv9T8/eEENfmG6l1w8Bf/c+npPNlP9S5YYL1m 5pq7avKB6l3w9N2XnMYRHBlH9a6oDbj3JkfJkX1U74q5VRunq0w15D3Vu+J5 8v1DpVYcGUv1I3Drq17wyrsCyaf6EVhTd9KhwJsn76h+BBKqk++MPaMiY6h+ JE5bZ2Sud+ZIHtWPRPS5uOXKlwJ5S/UjUTJlim+ToyCun5MbXv+vtDViipqt rxu0jQW51zzY+qe64fPKZ84uz9Vsf3T3/j3GNjSqxP1TusHTcLnrWWMN2183 /Kt6X9AvShD338kdTXnFv3a35lh9uGNMvDZ03TVOrJ9Ud6z59mW/XC+O1Zc7 onqMynfPFcT6U7qjJxezu2AZx+rTHfoao2Ondb9H69fJA/+W97QPadew+vaA o2T7un66+qP1n+qBqw9t3UICBdYfHji3/IRQnyCI/aP0wPXwbfYR03jWXx7o 9d3YbMl61n9Oo5BX+PuXkyUd/TkKRh+7nfeYKoj9mzYK4098Gb/qoor19yh8 f9ex/8+EF/tfOQoRPSoVbrE8mw+eGIV1j8328uL8cPLEdO8fwjekCWy+eEJb nbwzfD+bP2meUKWUTRjXLrD55IniT1tj9xXz4vxSeqK5ov+s+fc75tto+Bdu cXtozIvz1mk0fAoXXjR4LLB5OxpKeT+HgkJenLdpo7E5N954qYuGzdvRMPgj IbUhTS3OW+VoBPR9pzROV7F564WDVUHbumwTxHnr5IURze5D/G+q2Lz1wv3+ h+fdieSJIeV7Yd06k12D4rTkFOV7IbZwUPJhSzWJonwvDF/36ZnqGy3pSvne 0H+TOmTH/HpymvK98eZmasohB57MpHzd88r2WBcjDelG+d74OTy+7lI7R85Q vjdmva9e4pbNkVmU7421GWqpKpEn3SnfB2sP9fMfvpwnZynfB7s/9bt5L4gn synfB7t8SydV1HOkB+X7wMhk00/JFgI5R/k+iBwjbVz2PU+iKd8HExXxkcNn CKQn5fsiLsJwUEGsQM5Tvi/ij67t0vOUhsRQvi8GHljsmizhiRHl+6J6Reel Nw9z5ALl+0IzfcrJK7r5FEv5vmi7flJv6AwNMaZ8P8zLlQWtWiqQEsr3w/zN M8a/W8uTOZTvh+re1enJ3QRiQvl+eK9WXmzYLJBSyvfDoPR93WonCmQu5fth Ef9Y4aSrb1PK90cf/Y+T9MGTi5Tvj755fW9ZVAtkHuX748rpBUfy0nhiRvn+ WB642GLNAZ6UUb4/7i2vfZo9kyfzKd8fBrkrExqu8cSc8gOwx0O+Q9bEk0uU H4AK87eru9hoyALKD8D7nRltvjc40ovydXqryguHIgVSTvkBuGQhHT7hg5rE UX4ANqrfeu0oVRMLyg+EWZDjlMAonlRQfiCyNWZZRmcFEk/5gfD3Ll5T7q8h lpQfCMGv2SjrpZqIDR+I8iLbPHteTRIoPxANe3ptVcRyxIrygxB1L+34pxKO VFK+7t5m3MKt11VkIeUHQW/nXs6gUiC9KT8ILcl93ONWcKSK8oMg2/zHa8VE DVlE+UFwMJ2d+TRQQ6wpX4InS8omWSdypJryJVjfXCF9VyqQxZQvgd32zLYf vxWIDeVLsHJx7coDunxymfIlGONWKWw048gSypdgRK9QaZ5u/YzE+IzspiDD yUm8WI9WQM0r2e8GMzrqFWh1yzE7s50T6xnAZMU30b4LBbHeJwF7oi3srDfx Yj8kAImhqT1e9ujoF+CBssZ8epkg9lMmkGCxYZ7vv5zYb0eAgPA6ExKgEfux Erh13e9U3FCB9StwY2TxWVM31s9NgAnv/MnnAut3PSkuBJt+9l+sFueBlRRz m82Dr1mr2byQ4rfhrb3tj/LiPIEUmY+f3rH15sR5M0mK+V/NYofeE8R5lCDF wLigmYKfls0rKWYXf72V15UT51mmFN2MvFuG1atYvpTiSdLjpP7FWnEeVkrx 9Z5lfNvmjnkpRed4y6EFrSy/NkkxoLdzcmGWwPJtMAavObfRxoAX57FVMLpX HX55ZUzHvA6Gt+ve+pIglp8RDGnXV6V9DTUsrwYjr9ru28/l7DxICIbrlQ3b Ui53nBfBcBlSM2FvsEY8TzKDUdRXL7WopiOvBiMq8WHF4RZ2HlUGY5dkaka0 fcd5FYz484YLRhizvNoUjIOZf88LaWXnnZ4MdklndiTr1pPasZIhW7v188Eu AjsvZRiQbpWfcJDlVYkMiz5/THdbxM7bSTKYTI9bMVuXd+h5nCBDXP6lxV31 BXZey1Bd1aKqu8X+T2XKkLPWsZ9cq2Z5VYaW53VhtktYXiUyFBwve/nnLpZX lTIk7rQdEH9aLeaJJhlWr933l3eNluVVGb7xeGeVECGIecQqBKmfTrm6V7G8 6hQCw6PJgRMu82KekYSgBuVdk1s78moITJyTLB/VcGIeSgiB5RSTA/t16yHm pRD4/rRN/1SRIOapTN3z1s0ppj4alldDcLNrtePNN7yYV0kIHiwdHXPoO5WY V5UhyDnmkD/CiuXVphA8z671/tJfYHk1BC45b0saa1hetZLDctqGqzfzeDGv OskRmeyVblKuFfOqRI6iRPP8ZbU8y6ty3f+5lvjtTmoxrybIMe+15FCf7byY V1PlmKsd8mytjOXVTDlcVia22yfzLK/KkWbw6rlsgVbMq0SO6Ku+isP/8GJe Vcqx9FqVxfzJnJhXm+TIPzMueeRCjuVVOb7M+fX5pQu8mFetFNCWzTlWtlEQ 86qTAn/ekUcb6fITzasSBZ68nbrp1RI1CaP+FWjt0dw9eamG7KX+FUjpY1bi nseRFupfgcqJDzzLp/IklPpXoNsFv5Xt3R+QXOpfgd9C36TamwvkDfWvAHd0 zteSQTxRUP8KZMkvHXlUoSY51L8Cpa8Kzz004chr6l8Br3eTlCdLeSKn/kOR Yl2+oO62hmRT/6GYNfNoldFqgbyi/kOhahA+Lt3LkRDqPxT3FRnV3kU8yaL+ QzHsRX3+jX08aab+Q3Ex9Z+Js94KREb9h+Kdfk/Plec5sof6D0XZljmbTJsE 8pL6D0VS7KGmV64CCab+QxG5/WjLwSye7Kb+Q7HPpntqYiZHXlD/oVg/dfvA jJUCkVL/YbAZybkfz+HIb9R/GIzea7cGb+PJc+o/DJrEoBcZz1QE1H8Y9ANK w62KtCST+g9DiWTO5xezBdJE/YfhfFVj50V1PJFQ/2G4myXLb74tkF3Ufxi6 1Ho+vLddQ55R/2EY8Gjm+M0lPAmi/sPg9k+m5x/larKT+g/Dbvdh/U7M4slT 6j8MYcYffskeyJP/AwrNZ+k= "]]}, {}}, {{ Hue[0.59, 0.7, 0.75], Line[CompressedData[" 1:eJwd03tQzFEUB/AtPaQYVqa00SRbUlJLSak9tdvWGiWlRBERIZtK6akpjCwz 3o9ZPVDyykxGWFQHsxWrEAr9rsoWSslay1hW/H7dmTv3jzP33Pl+7r32CSkR iYYsFotHT2ZN3ZF0tODBZz4nds95V0E71kytupsm+cuPdB0S/TJ/i8f05h3s BhNYfiQy3NGoE5cOKcKGFBNg4+kCD2FfJxoPN/SqEieDVss1f7yJwt5Eztml m60gOuxlwaCSwgfvdsbVJXMgrjxh5LkFwbKoV9bO2+2gn+OuWfKTwoaNpz8r 79hD5Q3Jd+dqCq+dbwopvOcAtdJOtz4xwZIubdWCei4Uma30TgWCB2wc6COd wHdWaoXhBII50cvWV953hiRdQmabhEImzaqHLtDBf9Sa7E9wxdNrdpMUc+BV 6fXxLQ4EReNIfnPjXMhtjtnm1Ewhh2nX6w4B8jf5mcsJsgNsp8n6PMCnebX2 cBOFZvHzlmR84EGBtKoyaitBg/zFueEf50Hh2VPivGiCv2Trrrh8mg9rz+ht 964nqJZnvTXp94TssqacWHeCnzoOjVX1e0FxxB6NmxfBbibOwAK4LdDs1AcS pHFpAW+wsC7an+tLsNW9/UT64ELA7u9B+xwJNoYx4D6gGtGkzaF96pLHaJ2/ +MLrLH7vZVeCtVIGZBFcUvDYtmyCVy95RPQM+0GUGzuGZ0CwguH86g89XQcu y+j8sr74mpNqPqz6Y6reQvuK6ao8C6DoJNfG1IWgcGLwPkcdgKVl9fskT4JQ /mL68ewAMC2PCxm0JrjIbc0tg98BkFeWKv+hptC7fiA0JScQ1C/mh7eMI8ho kd+B0GXypFpUQeFcikWLCcDvS/E3XQ2FLlsOWsr/CGA4OyNqA73fSWdVzc0T wjNRabZXK4UziisEx/RCSNtsLI0xJTjdiukQBKqyi20bPlBoM/o8g2Bo4OZN ZTmFUzyZGxWB7IJUotJROEnRdk48IoLQj0p9jxXB8ZGrF97eFQwlxenKbhZB M1X/85n/gqF9dyZXOJug0egHCAFJrYa3jK6zRocYbK/X15ecoPA/PtucSg== "]]}, { AbsolutePointSize[2.735294117647059], Hue[0.59, 0.7, 0.65], Point[CompressedData[" 1:eJwd03tQzFEUB/AtPaQYVqa00SRbUlJLSak9tdvWGiWlRBERIZtK6akpjCwz 3o9ZPVDyykxGWFQHsxWrEAr9rsoWSslay1hW/H7dmTv3jzP33Pl+7r32CSkR iYYsFotHT2ZN3ZF0tODBZz4nds95V0E71kytupsm+cuPdB0S/TJ/i8f05h3s BhNYfiQy3NGoE5cOKcKGFBNg4+kCD2FfJxoPN/SqEieDVss1f7yJwt5Eztml m60gOuxlwaCSwgfvdsbVJXMgrjxh5LkFwbKoV9bO2+2gn+OuWfKTwoaNpz8r 79hD5Q3Jd+dqCq+dbwopvOcAtdJOtz4xwZIubdWCei4Uma30TgWCB2wc6COd wHdWaoXhBII50cvWV953hiRdQmabhEImzaqHLtDBf9Sa7E9wxdNrdpMUc+BV 6fXxLQ4EReNIfnPjXMhtjtnm1Ewhh2nX6w4B8jf5mcsJsgNsp8n6PMCnebX2 cBOFZvHzlmR84EGBtKoyaitBg/zFueEf50Hh2VPivGiCv2Trrrh8mg9rz+ht 964nqJZnvTXp94TssqacWHeCnzoOjVX1e0FxxB6NmxfBbibOwAK4LdDs1AcS pHFpAW+wsC7an+tLsNW9/UT64ELA7u9B+xwJNoYx4D6gGtGkzaF96pLHaJ2/ +MLrLH7vZVeCtVIGZBFcUvDYtmyCVy95RPQM+0GUGzuGZ0CwguH86g89XQcu y+j8sr74mpNqPqz6Y6reQvuK6ao8C6DoJNfG1IWgcGLwPkcdgKVl9fskT4JQ /mL68ewAMC2PCxm0JrjIbc0tg98BkFeWKv+hptC7fiA0JScQ1C/mh7eMI8ho kd+B0GXypFpUQeFcikWLCcDvS/E3XQ2FLlsOWsr/CGA4OyNqA73fSWdVzc0T wjNRabZXK4UziisEx/RCSNtsLI0xJTjdiukQBKqyi20bPlBoM/o8g2Bo4OZN ZTmFUzyZGxWB7IJUotJROEnRdk48IoLQj0p9jxXB8ZGrF97eFQwlxenKbhZB M1X/85n/gqF9dyZXOJug0egHCAFJrYa3jK6zRocYbK/X15ecoPA/PtucSg== "]]}}}, Prolog -> {{ Text[10, Offset[{-2, -2}, {10., 0.112}], {1., 1.}], Text[20, Offset[{-2, -2}, {20., 0.112}], {1., 1.}], Text[30, Offset[{-2, -2}, {30., 0.112}], {1., 1.}], Text[40, Offset[{-2, -2}, {40., 0.112}], {1., 1.}], Text[50, Offset[{-2, -2}, {50., 0.112}], {1., 1.}]}, { Text[ Superscript[10, "-1"], Offset[{3, 0}, {0, -1.}], {-1, -1.}], Text["1", Offset[{3, 0}, {0, 0.}], {-1, -1.}]}}, PlotRange -> {{0, 50.00213675213675}, {-1.112, 0.112}}, ImageSize -> {302, 182.}, GridLines -> {{10., 20., 30., 40., 50.}, {{-2., GrayLevel[0.3001]}, {-1., GrayLevel[0.3001]}, {0., GrayLevel[0.3001]}, {-1.6989700043360185`, GrayLevel[0.8501]}, {-1.3979400086720375`, GrayLevel[0.8501]}, {-1.2218487496163564`, GrayLevel[0.8501]}, {-1.0969100130080565`, GrayLevel[0.8501]}, {-0.6989700043360187, GrayLevel[0.8501]}, {-0.39794000867203755`, GrayLevel[0.8501]}, {-0.22184874961635626`, GrayLevel[0.8501]}, {-0.09691001300805638, GrayLevel[0.8501]}, {0.30102999566398114`, GrayLevel[0.8501]}, {0.6020599913279623, GrayLevel[0.8501]}, {0.7781512503836435, GrayLevel[0.8501]}, {0.9030899869919434, GrayLevel[0.8501]}}}, Frame -> True, Axes -> None, AspectRatio -> Full, BaseStyle -> { FontFamily -> "Verdana", FontSize -> 8, FontColor -> GrayLevel[0.5], ScriptSizeMultipliers -> 0.2, ScriptMinSize -> 6}, PlotRangePadding -> None, PlotRangeClipping -> True, Background -> GrayLevel[1], FrameStyle -> GrayLevel[0.5], ImagePadding -> 1, FrameTicks -> None], {None, "loss", "rounds"}], GeneralUtilities`TightLabeled[ Graphics[{ AbsoluteThickness[1.25], {{ Hue[0.083, 1., 1.], Line[CompressedData[" 1:eJxdV1dwlGUUXRCClNCSBYyICS2wJCFl08v3bV8bKBob2KKoxJI3nnDCDE/4 xIxuEDVWbFixgS0JW5NgmWFmHXxw8AVHV1QMooCibs69Of/Izj+z8+9//vPd 795zz3e3rKtnw+bJLperYpLLNf59Y+f4J2WKWvfnr2+N3GfN4QPb8tc3xoXP d2ZrV3n++lqfHzfe6I78dcS8sW/8kzN4vPWImQT8SeX7Sp+PmZK9vvz1pbkJ 7582+wH4wkwG/ozZNHB848Dxw+ZN4M+ZbQjgc3Mz8Od1/VHld1nhG9V4JtlH +4/u7D86Yl7H+5Pt7p6zfT1nRzX+KfZIaWH+GjWAd061wpc2gO8rsIUAjJh/ 8ZlG/A3AT7cIb+OoeQ34GXYHEpAx/wA/U9dLm+uBLyTfq8DPtmX4YdicB36O 3VWwJX8NK99c8sl684iXeObZE8l1+WtY451vq46NVR4bS+l+iizSXZTR/RZb 0O9Kaz6KrdRrWPPltkhfybDmc4E9iIRntF4LrNQzrfVaaIuxQELrtcjKOsNa r0UW6d+Z0npdQr6LgCux34+nb+CQeQv4Eu73FuAvtePZ6+sZMVOAX2zXY8Np 8zbwizUfSXMr8Jfx+VTgl2g94+Yd4JeQbyPwlzNfEk+p1mci3lLyy35Kreg3 o/sts53aJ5KPMnuqcpwwrvkq4/uSz6V2Qq+S76Wq14l6LGW+JI/LLMKrzGg9 nXup9zLVd0r1sIz1Eb0st4JLq56ce9HnctVHQvW5QvWXVH2uYD1EnysswutP mg3Ar9R6JswrwK+kHv8GfqV9BP2aNNcBX078y8CXM99/AV9O/LXAr2K8L+F7 lfZb2pwDfpUVP0qY9cCvZv72Ar+a9TkL/GoLuXuHzTr87lH9J82LwHu4vzPA exjfNcCvUb2mzAvAr1G9xc2fwK/ReDLmauArLNppb8o8D3wF6/UH8BWs/1XA V1KPzwFfyfhOA1/J/r0S+Cor/pgwzwJfxf3/DnwV8VcAv5b98Qzwa7Xfk+YU 8Gu5X+TPU839Sn6ryYf891YzXqlPteo/LfXLVrMfpL7VjBf199TwfdFHjeon KfrpdZ6Lvmr4PvSXraHeRZ811Bf066lVv46rvmu1/1Oi/95ai+0UpLQ/ai3s ui8h/ZOt5frSX7XMn8iljveiszrqF+27vU75MtrfddQj+j9bd4E/ePkc/uHx XuAv3v/7z3av9mdC/clLP4Z/Zb1WztMJf6vnfuG3nnrixW/rmU/47fZ69fOk +q3zHH6brbe47cqo3zYwf/BbTwPPK/HbBta3APwNej7GzX7wN5BvE/gbuL9p 4G/Ufhox74K/kfq8DfyN1NvF4G9kft8DfyP1eTv4G5nf6eBv4v374G9iPe/A 703M5wzwN9nUOuzIfAD+JvLfCX7n/Zngb6a/fgj+ZvrXXeBvZv/NAn8z++0A +JvZr13gd+4Lwd9C/zoI/hb6z93gb6GeZ4O/hXr4CPwtjO8e8LdQ33PA30r8 x+BvpT42g7+V/jMX/K30t0/A30r+e8HfynrNA0+b+nPSfAr+Nvb7feBvI998 8LdRT5+Bv43nxf3gb+P8UwT+duprAPzt9Nct4G/nesXgb+c8OAjeds5z3eBv 53nkBn8H74fA38F+fQD8HdTDAvB3sD6HwN/B/DwI/g7dT9wsBL/h/uLgN9TL Q+A3zMci8BvqOYH3DPP3MPgN8z1Lxmf2C/TotnwuerWMd6bAuR703mmpD/RD t+V60i9W55eE9FPMUo/ot31W65OQfhyy7C/pV8v8op9zlvlCv7t89Cf4gdvH fIlf+Lge/MT62F/wm04f9wM/6vbRn8SvfOxv+FnMx3lW5suJ9ePih0M+9o/4 pY/+CD/N+XgeyXzrZ//Aj91++o34tZ/zFfzc+tn/Mq/6OY/gPOj2c79yXvip P5wnMT/3K/Oqc4/zaMjP+OS88lNvOM9yzvriAwHmB+G4A9SLnJcB6gHHqQlw PZlXA1qPuJzH3QHqT87rAOcnnOexAPMr82pA/U//Tw0668u8EKA/Yp7IBVhf mVcD1BPmEXeQ+5V5JUh9Y54xQfaXzKtB+gnmoe4g45N5yXmOeSoW5Pkm86rz HPPqYJDzPubVbJD6wLyaCzJemVeD7A/Mq+4Q+wHzqidEv8W8akLMv8yrIfJj Xu0O8XzFvNobYv4xr8ace5lXQ8wn5tVBhx/zajbEfsO8mgvx/4XMqw4e86o7 zHxgXvWEqQfMqyZMvijiD/P/TD/iD1O/Y4jfuY8g/rDqMW6eRvxh+s1viD/M +oYRf5h+/hTiD9NfTiJ+J94Q4o+w359E/BE+/xXxR8gfxHeE/bkH8Ufox78g focvgPgjrNcTiCNCPf2M+CPsfz/id9bbjfgj7N8TiN9Z34f4o8xHH+KP8vlP iD/K/AtxVPVwyMQQf5T7zSH+KPMPg4hFOT88jvijnId+RPxRxteB+KM6z6fN Y4g/yn7+AfFHud//AEG/pLY= "]]}, {}}, {{ Hue[0.59, 0.7, 0.75], Line[CompressedData[" 1:eJxTTMoPSmViYGAwAmIQXViSMan+4Ct7+dbXgTvkzthvkFy2qyjvr73HwyqR de7H7Sf/4b4mtI/NASJ/3N7/zRG/N0f4HL5pxPQf+nrUnvXdvsePUoXh/Mep 0gv8M8Wh/CP2B++Wx+zJkXYAG/fwiP280CsSmgXyDrNmgsAR+31pM16d2qno ADa+9Yj92kXHPBp3Kzukp4HAEfs5974sM9+r6mDHdX1xge0R+24pZaCV6nD5 qrDA5CUHNB1EgaZXiRy1B/km6pC2w5ICW6COI/bh59bKCx7RdTj8FeSgI/Zu XHdqjx/VdwA6DihwxF4aZNxjAweQ6VzXD9sLOcrIznpi6PAG5JzXR+w54419 Sp8aOUDC44g9Y61XdcAzY4dqkHeqjtj/mJW4Svu5CdT8w/YfdlTcZHth6hAE 0h54xP75tX6ORy/M4Pz7IO+8NIe7Hxi4wBCwgJt/1uDq1OLXlg5g73IdsT/q BwpwK3j47clh/qL51hpu3pYuUIDYwN2/eoVh0IN3tlD3H7ZfDArO93bQ+Dti P+tJ/IZpH+wdGMDgiL0nUHZHhYMDOPjWHbZ3EXBvV/vp4FAICr6CI/YO8y/J Tal0hJtvoxe3jfGXIzz8LPa+9M2vcnKIAQscsQeF1p1fTnD9+rcZgCHmDI9f 7awekR2/neHuUf8pvka1xgUenkodi50n/3FxOHsGBA7by4mDTHCF2y8FTp6u cP2ipqAYdYPLCx65uNDznxvUfYfteYNjLbfXucPDg/PRiwsq/92h4XfYngWc ATzg8QcJF094/AAAqyt8EQ== "]]}, { AbsolutePointSize[2.735294117647059], Hue[0.59, 0.7, 0.65], Point[CompressedData[" 1:eJxTTMoPSmViYGAwAmIQXViSMan+4Ct7+dbXgTvkzthvkFy2qyjvr73HwyqR de7H7Sf/4b4mtI/NASJ/3N7/zRG/N0f4HL5pxPQf+nrUnvXdvsePUoXh/Mep 0gv8M8Wh/CP2B++Wx+zJkXYAG/fwiP280CsSmgXyDrNmgsAR+31pM16d2qno ADa+9Yj92kXHPBp3Kzukp4HAEfs5974sM9+r6mDHdX1xge0R+24pZaCV6nD5 qrDA5CUHNB1EgaZXiRy1B/km6pC2w5ICW6COI/bh59bKCx7RdTj8FeSgI/Zu XHdqjx/VdwA6DihwxF4aZNxjAweQ6VzXD9sLOcrIznpi6PAG5JzXR+w54419 Sp8aOUDC44g9Y61XdcAzY4dqkHeqjtj/mJW4Svu5CdT8w/YfdlTcZHth6hAE 0h54xP75tX6ORy/M4Pz7IO+8NIe7Hxi4wBCwgJt/1uDq1OLXlg5g73IdsT/q BwpwK3j47clh/qL51hpu3pYuUIDYwN2/eoVh0IN3tlD3H7ZfDArO93bQ+Dti P+tJ/IZpH+wdGMDgiL0nUHZHhYMDOPjWHbZ3EXBvV/vp4FAICr6CI/YO8y/J Tal0hJtvoxe3jfGXIzz8LPa+9M2vcnKIAQscsQeF1p1fTnD9+rcZgCHmDI9f 7awekR2/neHuUf8pvka1xgUenkodi50n/3FxOHsGBA7by4mDTHCF2y8FTp6u cP2ipqAYdYPLCx65uNDznxvUfYfteYNjLbfXucPDg/PRiwsq/92h4XfYngWc ATzg8QcJF094/AAAqyt8EQ== "]]}}}, Prolog -> {{ Text[10, Offset[{-2, -2}, {10., 0.7575000000000001}], {1., 1.}], Text[20, Offset[{-2, -2}, {20., 0.7575000000000001}], {1., 1.}], Text[30, Offset[{-2, -2}, {30., 0.7575000000000001}], {1., 1.}], Text[40, Offset[{-2, -2}, {40., 0.7575000000000001}], {1., 1.}], Text[50, Offset[{-2, -2}, {50., 0.7575000000000001}], {1., 1.}]}, { Text[ PercentForm[0.], Offset[{3, 0}, {0, 0.}], {-1, -1.}], Text[ PercentForm[0.2], Offset[{3, 0}, {0, 0.2}], {-1, -1.}], Text[ PercentForm[0.4], Offset[{3, 0}, {0, 0.4}], {-1, -1.}], Text[ PercentForm[0.6], Offset[{3, 0}, {0, 0.6}], {-1, -1.}]}}, PlotRange -> {{ 0, 50.00213675213675}, {-0.0075000000000000015`, 0.7575000000000001}}, ImageSize -> {302, 182.}, GridLines -> {{10., 20., 30., 40., 50.}, {{0., GrayLevel[0.8501]}, {0.05, GrayLevel[0.8501]}, {0.1, GrayLevel[0.8501]}, {0.15, GrayLevel[0.8501]}, {0.2, GrayLevel[0.8501]}, {0.2, GrayLevel[0.8501]}, {0.25, GrayLevel[0.8501]}, {0.3, GrayLevel[0.8501]}, {0.35, GrayLevel[0.8501]}, {0.4, GrayLevel[0.8501]}, {0.4, GrayLevel[0.8501]}, {0.45, GrayLevel[0.8501]}, {0.5, GrayLevel[0.8501]}, {0.55, GrayLevel[0.8501]}, {0.6, GrayLevel[0.8501]}, {0.6, GrayLevel[0.8501]}, {0.65, GrayLevel[0.8501]}, {0.7, GrayLevel[0.8501]}, {0.75, GrayLevel[0.8501]}, {0.8, GrayLevel[0.8501]}, {0., GrayLevel[0.3001]}, {0.2, GrayLevel[0.3001]}, {0.4, GrayLevel[0.3001]}, {0.6, GrayLevel[0.3001]}, {0.8, GrayLevel[0.3001]}}}, Frame -> True, Axes -> None, AspectRatio -> Full, BaseStyle -> { FontFamily -> "Verdana", FontSize -> 8, FontColor -> GrayLevel[0.5], ScriptSizeMultipliers -> 0.2, ScriptMinSize -> 6}, PlotRangePadding -> None, PlotRangeClipping -> True, Background -> GrayLevel[1], FrameStyle -> GrayLevel[0.5], ImagePadding -> 1, FrameTicks -> None], {None, "error rate", "rounds"}]}}, TagBox[GridBox[{ { DynamicBox[ToBoxes[ Grid[{{ Button[ MouseAppearance[ Style["\:2039", FontColor :> If[ CurrentValue["MouseOver"], RGBColor[0.27, 0.54, 0.79], GrayLevel[0.25]], FontFamily -> "MS Gothic", FontSize -> 14, FontWeight -> "Thin"], "LinkHand"], NeuralNetworks`Private`NetTrain`index$$ = Mod[NeuralNetworks`Private`NetTrain`index$$ - 1, 2, 1], Appearance -> None], Dynamic[ Part[{ Tooltip["loss", Column[{ Row[{"Measurement: ", "loss"}]}]], Tooltip["error rate", Column[{ Row[{"Measurement: ", "ErrorRate"}]}]]}, NeuralNetworks`Private`NetTrain`index$$], TrackedSymbols :> { NeuralNetworks`Private`NetTrain`index$$}], Button[ MouseAppearance[ Style["\:203a", FontColor :> If[ CurrentValue["MouseOver"], RGBColor[0.27, 0.54, 0.79], GrayLevel[0.25]], FontFamily -> "MS Gothic", FontSize -> 14, FontWeight -> "Thin"], "LinkHand"], NeuralNetworks`Private`NetTrain`index$$ = Mod[NeuralNetworks`Private`NetTrain`index$$ + 1, 2, 1], Appearance -> None]}}, ItemSize -> {{3, 20, 3}, {2}}], StandardForm], ImageSizeCache->{265., {7.9150390625, 14.0849609375}}, TrackedSymbols:>{NeuralNetworks`Private`NetTrain`index$$}]}, { DynamicBox[ToBoxes[ Part[ NeuralNetworks`Private`NetTrain`plotList$$, NeuralNetworks`Private`NetTrain`index$$], StandardForm], ImageSizeCache->{313., {95., 99.}}, TrackedSymbols:>{ NeuralNetworks`Private`NetTrain`index$$}]} }, DefaultBaseStyle->"Column", GridBoxAlignment->{"Columns" -> {{Center}}}, GridBoxItemSize->{ "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"], DynamicModuleValues:>{}], Alignment->Center, StripOnInput->False], "\[SpanFromLeft]"}, { ItemBox[ StyleBox[ RowBox[{ GraphicsBox[{}, Axes->None, Background->Hue[0.083, 1., 1.], ImagePadding->{{0, 0}, {5, 2}}, ImageSize->{13, 1}], " ", "\<\"training set\"\>", "\t", GraphicsBox[{}, Axes->None, Background->Hue[0.59, 0.7, 0.75], ImagePadding->{{0, 0}, {5, 2}}, ImageSize->{13, 1}], " ", "\<\"validation set\"\>"}], FontFamily->"Verdana", FontSize->8, FontColor->GrayLevel[0.4]], Alignment->Center, StripOnInput->False], "\[SpanFromLeft]"} }, AutoDelete->False, ColumnsEqual->False, GridBoxAlignment->{ "Columns" -> {Right, {Left}}, "Rows" -> {{Baseline}}}, GridBoxDividers->{"Columns" -> {False, { Opacity[0.15]}, False}}, GridBoxItemSize->{"Columns" -> {8, {26}}, "Rows" -> {{1.}}}, GridBoxSpacings->{"Columns" -> { Offset[0.27999999999999997`], Offset[1.1199999999999999`], { Offset[1.75]}, Offset[0.27999999999999997`]}, "Rows" -> { Offset[0.2], { Offset[0.8]}, Offset[0.2]}}], "Grid"], FrameMargins->{{10, 10}, {10, 5}}, FrameStyle->None, StripOnInput->False], Deploy, DefaultBaseStyle->"Deploy"], Alignment->Center, BaseStyle->{ FontWeight -> "Light", FontSize -> 11, FontFamily -> "DejaVu Sans", NumberMarks -> False, Deployed -> False}, StripOnInput->False]} }, DefaultBaseStyle->"Column", GridBoxAlignment->{"Columns" -> {{Left}}, "Rows" -> {{Baseline}}}, GridBoxDividers->{"Columns" -> {{False}}, "Rows" -> {{False}}}, GridBoxItemSize->{"Columns" -> {{Automatic}}, "Rows" -> {{1.}}}, GridBoxSpacings->{"Columns" -> { Offset[0.27999999999999997`], { Offset[0.5599999999999999]}, Offset[0.27999999999999997`]}, "Rows" -> { Offset[0.2], Offset[0.8], { Offset[0.4]}, Offset[0.2]}}], Background->GrayLevel[1], FrameMargins->{{0, 0}, {0, 0}}, FrameStyle->GrayLevel[0.85], RoundingRadius->5, StripOnInput->False]], "Output", CellChangeTimes->{{3.763468422838024*^9, 3.763468455380887*^9}, 3.7634863216612873`*^9, 3.763486377409851*^9, 3.763486481919368*^9, { 3.763486530827507*^9, 3.763486635707397*^9}, 3.763486817806835*^9, 3.763487456940959*^9, 3.763747836554844*^9, 3.763816564003991*^9, 3.763915676246413*^9, 3.764082207345428*^9, {3.764591547767344*^9, 3.764591556491619*^9}, {3.764591608543784*^9, 3.76459162361261*^9}, 3.7645918250792933`*^9, 3.764593130126831*^9, 3.764593390358643*^9, 3.7645935759368477`*^9}, CellLabel->"Out[8]="] }, Open ]], Cell["Check the classification error rate on the validation data:", "Text", CellChangeTimes->{{3.728707259494198*^9, 3.7287072707902927`*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"bertresults", "[", RowBox[{"\"\\"", ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{{3.764591876685998*^9, 3.764591885765334*^9}}, CellLabel->"In[10]:="], Cell[BoxData["0.15937500000000002`"], "Output", CellChangeTimes->{{3.764592187436552*^9, 3.764592202671833*^9}, 3.764593130216865*^9, {3.764593576029887*^9, 3.7645935975256233`*^9}}, CellLabel->"Out[10]="] }, Open ]], Cell["\<\ Let\[CloseCurlyQuote]s compare the results with the performance of a \ classifier trained on context-independent word embeddings. Precompute the \ GloVe vectors for the training and the validation dataset:\ \>", "Text", CellChangeTimes->{{3.728665924590131*^9, 3.728665936518705*^9}, { 3.728706974627112*^9, 3.728706991043274*^9}, {3.728710495308913*^9, 3.728710495757037*^9}, {3.731757039956691*^9, 3.7317570426927567`*^9}, { 3.731757084365018*^9, 3.731757096374072*^9}, {3.763742497516614*^9, 3.763742506564638*^9}, {3.7637425640148363`*^9, 3.7637426249933434`*^9}, 3.7637466824005527`*^9, 3.763916566780933*^9, 3.763979496101955*^9, 3.764089492728236*^9, 3.764092533835499*^9, {3.764593092225754*^9, 3.764593096526781*^9}, {3.7722832247550783`*^9, 3.772283225123702*^9}, { 3.773419762815422*^9, 3.773419763055723*^9}}], Cell[BoxData[{ RowBox[{ RowBox[{"trainembeddingsglove", "=", RowBox[{ RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], "[", RowBox[{ RowBox[{"train", "[", RowBox[{"[", RowBox[{"All", ",", "1"}], "]"}], "]"}], ",", RowBox[{"TargetDevice", "\[Rule]", "\"\\""}]}], "]"}], "\[Rule]", RowBox[{"train", "[", RowBox[{"[", RowBox[{"All", ",", "2"}], "]"}], "]"}]}]}], ";"}], "\[IndentingNewLine]", RowBox[{ RowBox[{"validembeddingsglove", "=", RowBox[{ RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], "[", RowBox[{ RowBox[{"valid", "[", RowBox[{"[", RowBox[{"All", ",", "1"}], "]"}], "]"}], ",", RowBox[{"TargetDevice", "\[Rule]", "\"\\""}]}], "]"}], "\[Rule]", RowBox[{"valid", "[", RowBox[{"[", RowBox[{"All", ",", "2"}], "]"}], "]"}]}]}], ";"}]}], "Input", CellChangeTimes->{{3.7634861277443953`*^9, 3.7634861327671947`*^9}, { 3.763486406746982*^9, 3.763486409709483*^9}, {3.763487998167246*^9, 3.7634880009647417`*^9}, {3.764592886385256*^9, 3.7645929163158092`*^9}, 3.764593299514226*^9}, CellLabel->"In[12]:="], Cell["Train the classifier on the precomputed GloVe vectors:", "Text", CellChangeTimes->{ 3.763321126516569*^9, {3.7633214392298937`*^9, 3.7633214468289948`*^9}, 3.763468410163715*^9, {3.763486158173032*^9, 3.7634861642197733`*^9}, { 3.763746685872386*^9, 3.763746686911667*^9}, {3.764089505288436*^9, 3.764089509032104*^9}, 3.764092508410285*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"gloveresults", "=", RowBox[{"NetTrain", "[", RowBox[{ "classifierhead", ",", "trainembeddingsglove", ",", "All", ",", "\[IndentingNewLine]", RowBox[{"ValidationSet", "\[Rule]", "validembeddingsglove"}], ",", "\[IndentingNewLine]", RowBox[{"TrainingStoppingCriterion", "\[Rule]", RowBox[{"<|", RowBox[{ RowBox[{"\"\\"", "\[Rule]", "\"\\""}], ",", RowBox[{"\"\\"", "\[Rule]", "50"}]}], "|>"}]}], ",", "\[IndentingNewLine]", RowBox[{"TargetDevice", "\[Rule]", "\"\\""}], ",", "\[IndentingNewLine]", RowBox[{"MaxTrainingRounds", "\[Rule]", "50"}]}], "]"}]}]], "Input", CellChangeTimes->{{3.7633209380446663`*^9, 3.763320949020236*^9}, 3.763320988762795*^9, {3.763321410608366*^9, 3.763321453212701*^9}, 3.76332170959013*^9, 3.76333481934881*^9, {3.763363037985634*^9, 3.763363038054048*^9}, {3.763468322792994*^9, 3.7634683760863247`*^9}, { 3.7634684534895773`*^9, 3.7634684547441673`*^9}, {3.7634864118313217`*^9, 3.7634864147782917`*^9}, 3.7645928785600357`*^9, 3.7645929721676598`*^9}, CellLabel->"In[14]:="], Cell[BoxData[ FrameBox[GridBox[{ { ItemBox[ FrameBox[ StyleBox["\<\"NetTrain Results\"\>", "SuggestionsBarText", StripOnInput->False, FontSize->12], FrameMargins->{{10, 5}, {-4, 2}}, FrameStyle->None, StripOnInput->False], Alignment->{Left, Bottom}, Background->RGBColor[0.96, 0.96, 0.96], Frame->{{False, False}, {True, False}}, FrameStyle->Opacity[0.1], ItemSize->{Automatic, 1}, StripOnInput->False]}, { ItemBox[ TagBox[ FrameBox[ TagBox[GridBox[{ { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"summary\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{"\[ThinSpace]", "\" \"", "\"batches:\"", "11700"}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", "\"rounds:\"", "50"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"time:\"", "\"1.2min\""}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"examples/s:\"", "5105"}, "RowWithSeparators"]}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"data\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{ "\[ThinSpace]", "\" \"", "\"training examples:\"", "7462"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"validation examples:\"", "3200"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"processed examples:\"", "374400"}, "RowWithSeparators"], TemplateBox[{ "\[ThinSpace]", "\" \"", "\"skipped examples:\"", "0"}, "RowWithSeparators"]}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"method\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{ "\[ThinSpace]", "\" \"", "\"ADAM\"", "\"optimizer\""}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", "\"batch size\"", "32"}, "RowWithSeparators"], "\"CPU\""}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"round\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"loss:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"loss\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "loss"}]}], "Tooltip"]& ], TagBox[ InterpretationBox[ RowBox[{"\"5.3\"", "\[Times]", SuperscriptBox["10", "\"-1\""]}], 0.5303649358054858, AutoDelete -> True], ScientificForm[#, 3]& ]}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"error:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"ErrorRate\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "ErrorRate"}]}], "Tooltip"]& ], "\"26.9%\""}, "RowWithSeparators"]}, "RowWithSeparators"]}, { TemplateBox[{ TemplateBox[{5}, "Spacer1"], StyleBox["\"validation\"", GrayLevel[0.4], StripOnInput -> False]}, "RowDefault"], TemplateBox[{",", "\",\"", TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"loss:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"loss\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "loss"}]}], "Tooltip"]& ], TagBox[ InterpretationBox[ RowBox[{"\"5.15\"", "\[Times]", SuperscriptBox["10", "\"-1\""]}], 0.515397813525924, AutoDelete -> True], ScientificForm[#, 3]& ]}, "RowWithSeparators"], TemplateBox[{"\[ThinSpace]", "\" \"", TagBox[ TooltipBox["\"error:\"", TagBox[ GridBox[{{ TemplateBox[{"\"Measurement: \"", "\"ErrorRate\""}, "RowDefault"]}}, GridBoxAlignment -> {"Columns" -> {{Left}}}, DefaultBaseStyle -> "Column", GridBoxItemSize -> { "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"]], Annotation[#, Column[{ Row[{"Measurement: ", "ErrorRate"}]}], "Tooltip"]& ], "\"26.1%\""}, "RowWithSeparators"]}, "RowWithSeparators"]}, { ItemBox[ DynamicModuleBox[{NeuralNetworks`Private`NetTrain`index$$ = 2, NeuralNetworks`Private`NetTrain`plotList$$ = { GeneralUtilities`TightLabeled[ Graphics[{ AbsoluteThickness[1.25], {{ Hue[0.083, 1., 1.], Line[CompressedData[" 1:eJw1l3lczdkfxtOGLDNSIiRtctv3Vfepe+teMrYUGUsUEanBWCa0kC2MoWI0 DIZElhgt1hPCIPU93zT5ScwwM5IlUWOtfnfO99z7T6/7us9993zP+ZznPHfo rKQJs3V1dHQcu+jo/Pc3KvK/1zX5sF2r9Lf7XSHS+1p55oZrCw9/vkJ02OsP efKvensCW6/yz/+WR28tXLXM8Ro5dvS/V5Nc2RHX8GHMNdKF6V/L8+OnONbV Xeefv5G3qtx8DcpvkEns+23yza+/CE58ep3oMv17uZFsaqbc8iY5zvQf5UNO JBrUBd4ik5m+XV4XXv88M+kW5+sg7LeCw1mTKrmfLjj7aM/7vpeqSCH7vi6u JqWF9ex9h/vXR2bUs9zQ9ZWEySMNoDjzsOBV0h3C5EcNYXun66CYuZWkk726 oqvjfP2g1moykem7o3/kzvj2+VXkCNMbIa/TzD5zYhXpYPoemDI68a715CoS wfS94Dwn9us52QIpYPreyNrvZWdyuYq0M/0XWFC61KZzssB5X8LpJNJWbNf+ vz74yqndY/hoyv30gblThNJ1EOV+jeE3PSnroGUVf56+cF5u3PVeosCf1wRP /BKtR04Q+HqYoGz+5arMMoGvlynyxrgWpDcIfD37IaL1Y8aBTdV8v/pheZeL htFGAt8vM+yy/7nw3UqB71d/FL4tXrY4WOD71R/m75JbiqIo368BmDsy6Hpe f4HoMb05TjQvGGxmSMkJpjfHyCdLYKHxE830A3Hu2J0HljYC0Wf6QSi41/TH /mECOcn0g5CQUTd587eUTGH6wdgU4h2Tl0OJAdNbYMaSgx4zb1JSxPQW+M13 +v4PK6rJ10w/BNeuL7UuLKLcjyVM2mzGvvkocL+WcH0e2/7YhPLnsYT73RPr aiZQ/rxDUeQ+pHu3Zu16DEXKotJSiwsCX6+hyD66bMbhau16WqF1l889vY+U r7cVJrpnr+x5UbsfVrDtlnDumwztfllj96e4FPNvRL6f1rip2qtr8K12v63R YSsPfptbzefBGnsc6u17X9HOiw3inIaOoROq+TzZYFpjl/JbHiKfTxvYhbrH FDkJfD5tEf8o8qRXDOXzaYvirR8OkWOUz6ctfpCfOHuxn0gmML0derd7bHkY K5LDTG+H5fM/3jlgSclnprfDJ68D8m97UjKe6YdB/5cHi8qXCySf6YchYlCo scsTSj4x/TDMrihqKJ4nkHFMb4/79rKlNTWUHGJ6e9wb89e6+74i+cj09lDM GtJy25eSsUw/HMuVGysrKqrJQaYfjpbWTeouzwTygemHo7wkQPnMUyBjmF6G JqN5Bp75lPzC9DIMdhnzYNhjkbxnehn2RjUUvg2g5Cumd8DY9tyWaY6UHGB6 B+QbTx9pWSqQd0zvgD2RNsk/a+Z/NNM7wvC8oUGJByX7md4RK/WM7dJaBfIv 0zvC0rqmMTNZJOFM74TKz50pX1ZRso/pnfC8x0/6t24LpI3pnXBpsbnafDwl o5jeGSELk3ptDqXkZ6Z3hrxZ3Nxzm0hamd4Z/omdFQc0eTGS6V0gm/HgXfhv AtnL9C64a227NjdeJG+Z3gUel+MXrwyk0vrJXNH4oLv+z3UCX19X0BE6Ad3/ oNL6p7qiX3q3lY0+2v1xhezep6gTd6m0f7Wu2Jo0M/XVDZHvryuK7020szrP 91/mhhIYfG+UrZ0PN3SOKgnYXSxI85Pqho3rkzoMsgQ+X25wOB++tq2FSvNX 64bJ11323B5Qw+fTDatWqD+OesnzVeaO2HN78heMo3y+3WG/d2naN158/lPd 8afZzbLSDJGfD837KN/jZJYgnZ9adzg83Fm7YZ7Iz5c7glu3FTUe5edP5oGP T3sONPHTnk8PNMyfa520XpTOb5oH4mT7PGbG1/Dz7QHrAYXrzebxvK31gOe+ N05/x2jzwROGVDdk4F4q5YfME+UuxwuGT6Q8XzwRe/fuu/CBNVL+pHki6Zsf Khala/PWE9v9x7wuvcjzttYTvwdYjr84SuT55oXHR1fdSXgqSHkr80Lhv0cV l8xqeN56ocufo8PJFirlbZoXfpw9PbNppsjz1gvK7AAxfibP21ovfLn9x0Fr Bmvz1hvOM4sKDP143sq8Yb8ievZXlwWet95Qfh7ntFpzHxgyvjcqlum3OT6l 5BTje6PI69moDTMomcr43oidnJ+c5SCSrozvg/GjOvRd71BymvF9cG9r7/sz D1EyjfF9EPCq7vtSQkk3xveBbsZsq4Z1lPzK+D64dPD5vtUuIpnO+D6IGdp1 4OMllHRnfF8kb49QnRkpkDOM7wv9zvLE7/dQMoPxfbGtZseToGeUGDG+L1QL TG5s1dwvxYzviy3xbtkvNecphvF98d0m+6WpfwmkB+P7wchDSH87hJISxvdD 0g1zVXSLQGYyvh/Od5/sd9JXID0Z3w8RJ9N7e/UUSSnj+8E5wizZ31sksxjf D+HNjfOzNc/Ti/H98cN026izIykpY3x/HLlV3+PIVJHEMr4/cGxQ+sF8gfRm fH9k/TK2nmjm7Szj+2OZbbz3LMMaEsf4/ghxdy+qf0jJF4wfgOOh1q8/aPLm HOMHIPpNZOvpvSKZzfgBWKa6OsHhtEi+ZPwA5FcNqVukyd/zjB+Aow/f3Onc Rckcxg9At0c3WxPeCKQP4wei2n5n8aVkSi4wfiDkPoUpOxQCiWf8QGx9qGgY LhOIMeMHIv9/Vt67PojkIuMHQh1669KNLJHMZfxAqA6V1KieCKQv44+AffA/ gyNuU3KJ8UdgRd9bH/Q053ke449Ahk/svJ8sRGLC+COgE3I321szz9KBH4Hs nPO6McspSWD8EbiVefK+eohITBk/CJtu/LP5vg4l5YwfhJInbx2OL6ZkPuMH 4fLoBW1d1lDSj/GDUDkru+L6bkouM34Q1ng/S1Np5mkB4wfh0aS2u90GCMSM 8eVYOKtji+tGSq4wvhz6exZ6ercJJJHx5bjtuDq/2Fok/RlfjnqDHBuTzZRc ZXw5UsNWVjcvEchCxpejR4E8LTKSSvOmA4Qv9zU/oekXbB5NAe8jXSI8bgp8 XoHAhGPWq8aL0jwDsL9bS6M283mPBD4kWRiv8Rek85AA2OX4bnnqpj0vwCD6 +O8BRJDOUw7wz9lPi23aROm8HQUeFOulJvcQpPNYDuy+OPdK5TaBn1fAtL7D 7sJAfp6bgL2n6+sUXvy86wRDqLifmnFZlPLANBjRwukRaZr1kPIiGKOHXKKv S6iUJwhGQcqxDGt9UcqbyGBU33rhevWkIOVRQjByO/NyD/UTeF4FI7zneb1n i0Qpz3KCcUzZ/Ysnn7X9MhibdNsH52rmh+VheTDyvR9vfKzpR1JeBsOzq+kk lxzeX5uCcbXR5JRVAM9bnRCkTRSmzKnh/ddU837S+hdL83g/loVgcf+JUbIU nucIgf3qmykpY7T9OgQRK6Pj0tJ4X00IgfOKnMhJlbyvpoUgZpHtlf42onSf 5ISgYoyl/0NdkffVEExp8U0/10uU7qPyEKQ7f2uXRXlfrQ3B4ObrL97VUek+ awpB0PAHKyqjKe+rCsSti/77n0lUug9NFegMT+nYN1Hk96UChy1M56eY8PtU rkDMyiNlLR8o76sKGL25lxXznkr3cYICHr0mDH5ay39PpSowatwZa6vP/D7P UaB0izrx9nKR91UF8leUFy+9z/sqUSBh3XKHw+1VvC8osP7s+obmz9VSn2hS oPrtvCU1B0XeVxWY2T45IeSUIPURUyUGyBPKHPK0fUWJm2f2b+z5He+rciVS C1plF6byvhOphFnCb9M7K3gfSlAiZ4W9zeyr2r6kRKYsLu/zoBqpT+UoYXqw YdFgTV5KfVWJqDtLX1sYiFJfJUrkWbY3K49Tqa/WKrHm+ue1bWW8rzYpQeWX leaavJT6qhIpSffq8qaIUl81DYU4qnuQYTzvq7JQKF81vt/aRqW+Kg/Fxidd hzqs1fbVUFT/ZNOmyuR9NUGjL12qG1vO+2pqKNYmTzj+2J/31ZxQlFXmHdeL 0fbVUBwrKgk/orkvWV8loVi84f3psve8r9aG4kfjNx9PHuF9tSkUG6bUOmUc F3hfDcVhRe6N0f14XzUNQ3N64cAd83lflYVhTdqp17sDeV+Vh2F7yZyWgZ2U qJn/MFzZ6eiv1vwe3cP8h0FpPW7h0emUvGH+w6DI7xB7W4lExfyHoV1ofT5W cx/9xPyHQV1PtqXOo6SF+dfwt+fUf+pPSRjzH4Z3vxtu+NSXkjzmPwzxnyJf /tAkkNfMfxgW2RxaYxxQQ0KZfxVSRjc073wvkN3MvwpxfS7v89X8Pmlm/lWw 9/cUlqlriJL5V2FgcWTx+whKfmT+VVjtfyGVpgjkFfOvwuH6aRnxC0WiYP5V 8Ho+9i8LzXruYv5VyLVd75ayTCQvmX8VjAt3uMZqfk+EMP8q6MmCMto067uT +VehxOzXQ9umUfKC+VfhUcvv3lONRBLM/Ksx4vkxvTRNvucy/2ocdpjkVOoq kufMvxrb9k7C18NEAuZfjXR/y4wJCZTkMP9qnNhprtuhydcm5l+N0fet1pWt riZy5l+Nr7qvjKMTBZLN/KvRTe/FqaaXInnG/KvhcubE3nWaPhfE/KtB0Sdx zkOB7GD+1bCYNq9huqYPNjL/asz993mj2XCR/B+C8k/H "]]}, {}}, {{ Hue[0.59, 0.7, 0.75], Line[CompressedData[" 1:eJwdk3lQzGEYx1diKCUtrW4hlag90GHVk9pNRIdhBx3C5qhp2dKxFYVIa4TF mB03k1wzjSPraB+ZLSOj2d87ypWr2MKWYzC1lby/3pl33j+eeZ/v8/1839dr nSJRbsXhcIR0s+e2nE1HdtZ/DVeWqW+cUz3BGueqe8qswfCe162Vl92bUTNg 2+qoHw2z8rrn9UuMGGc2LDMb7IH/08Va8deIo3r0He1yLhSYx1ntjGewQ+56 Nm4zD25auLM5agbr3+YlPch0haopxtUTnzJ4esXzyX5bPUGvtdu7/w2D+vQT X5vuekGa1a8XTZ8ZvH6+cVHp/WkQ5K25JDIzePLd76qgOm8Q3X+YZ+plUO0y jUr6QHXWobW5fxlUrUxYf/GhH3iNO/Fu91iCrJvVj/yhqbzzlrqfQVnzdc8J htmwocOhJENIUGrTVvy4IRDkg9yZwlCCrmy7Dj5E9u7ZeHopQccIN3ftJwFY jH8qDqYSHJsqit3+WQgat4wl+9MIjiheXBhvEoG5te1kegrBXm3aFf/OOVBm AfFQPMEfuvxXo7vmwpT6WJDQ+52tlWPau+bBpQDlEdlmgu9ZO1+CoLpmudO9 JIIULiUQDC+FjrWFWwk+47ccy/4WAl3vX7xNkBNsWMYCD4VmX5nKL5Hgg8yR v/2650P1UY2WJyN4q4IFIgbZHKdZ+9IJXq0WJH7oWQDW3PG25bTfBRbn9zCQ Jk4O2J5FUPspteb4j3Aggm5vfQnBGFrV5QMINPbbCnIIRjlE75vRB5DNdW50 yyUIZ4jH0YIIyOavstFsIigOSKkdYYkA2xLDYg+qF1z3ZalCtRBapjruGkN5 sLTaLAvhblh8qprWA99wKLFIsPYxnUqh+v5bDkzU9UdCrbNAV7eFoE8f75p3 URSYPlZkFK0hOLX8QqRmIAoyGgKPeWYS9OCxHSTQclu30on6cRl+nhKQOPQk S5UEJ81lE5WCSRRbak/zmWBgzsX8k8JM8Y38ZDqv3fLkkDs7okFZ+Nq3UUHz bO8yTh+KhhS+zeFmFUHr4Q+wCAbsknbwighyhlcMlImHKiOo/n8v4o2U "]]}, { AbsolutePointSize[2.735294117647059], Hue[0.59, 0.7, 0.65], Point[CompressedData[" 1:eJwdk3lQzGEYx1diKCUtrW4hlag90GHVk9pNRIdhBx3C5qhp2dKxFYVIa4TF mB03k1wzjSPraB+ZLSOj2d87ypWr2MKWYzC1lby/3pl33j+eeZ/v8/1839dr nSJRbsXhcIR0s+e2nE1HdtZ/DVeWqW+cUz3BGueqe8qswfCe162Vl92bUTNg 2+qoHw2z8rrn9UuMGGc2LDMb7IH/08Va8deIo3r0He1yLhSYx1ntjGewQ+56 Nm4zD25auLM5agbr3+YlPch0haopxtUTnzJ4esXzyX5bPUGvtdu7/w2D+vQT X5vuekGa1a8XTZ8ZvH6+cVHp/WkQ5K25JDIzePLd76qgOm8Q3X+YZ+plUO0y jUr6QHXWobW5fxlUrUxYf/GhH3iNO/Fu91iCrJvVj/yhqbzzlrqfQVnzdc8J htmwocOhJENIUGrTVvy4IRDkg9yZwlCCrmy7Dj5E9u7ZeHopQccIN3ftJwFY jH8qDqYSHJsqit3+WQgat4wl+9MIjiheXBhvEoG5te1kegrBXm3aFf/OOVBm AfFQPMEfuvxXo7vmwpT6WJDQ+52tlWPau+bBpQDlEdlmgu9ZO1+CoLpmudO9 JIIULiUQDC+FjrWFWwk+47ccy/4WAl3vX7xNkBNsWMYCD4VmX5nKL5Hgg8yR v/2650P1UY2WJyN4q4IFIgbZHKdZ+9IJXq0WJH7oWQDW3PG25bTfBRbn9zCQ Jk4O2J5FUPspteb4j3Aggm5vfQnBGFrV5QMINPbbCnIIRjlE75vRB5DNdW50 yyUIZ4jH0YIIyOavstFsIigOSKkdYYkA2xLDYg+qF1z3ZalCtRBapjruGkN5 sLTaLAvhblh8qprWA99wKLFIsPYxnUqh+v5bDkzU9UdCrbNAV7eFoE8f75p3 URSYPlZkFK0hOLX8QqRmIAoyGgKPeWYS9OCxHSTQclu30on6cRl+nhKQOPQk S5UEJ81lE5WCSRRbak/zmWBgzsX8k8JM8Y38ZDqv3fLkkDs7okFZ+Nq3UUHz bO8yTh+KhhS+zeFmFUHr4Q+wCAbsknbwighyhlcMlImHKiOo/n8v4o2U "]]}}}, Prolog -> {{ Text[10, Offset[{-2, -2}, {10., 0.21100000000000002`}], {1., 1.}], Text[20, Offset[{-2, -2}, {20., 0.21100000000000002`}], {1., 1.}], Text[30, Offset[{-2, -2}, {30., 0.21100000000000002`}], {1., 1.}], Text[40, Offset[{-2, -2}, {40., 0.21100000000000002`}], {1., 1.}], Text[50, Offset[{-2, -2}, {50., 0.21100000000000002`}], {1., 1.}]}, { Text["1", Offset[{3, 0}, {0, 0.}], {-1, -1.}]}}, PlotRange -> {{0, 50.00213675213675}, {-0.911, 0.21100000000000002`}}, ImageSize -> {302, 182.}, GridLines -> {{10., 20., 30., 40., 50.}, {{-1., GrayLevel[0.3001]}, {0., GrayLevel[0.3001]}, {-0.6989700043360187, GrayLevel[0.8501]}, {-0.5228787452803375, GrayLevel[0.8501]}, {-0.39794000867203755`, GrayLevel[0.8501]}, {-0.30102999566398114`, GrayLevel[0.8501]}, {-0.22184874961635626`, GrayLevel[0.8501]}, {-0.1549019599857432, GrayLevel[0.8501]}, {-0.09691001300805638, GrayLevel[0.8501]}, {-0.04575749056067506, GrayLevel[0.8501]}, {0.30102999566398114`, GrayLevel[0.8501]}, {0.47712125471966244`, GrayLevel[0.8501]}, {0.6020599913279623, GrayLevel[0.8501]}, {0.6989700043360187, GrayLevel[0.8501]}, {0.7781512503836435, GrayLevel[0.8501]}, {0.8450980400142567, GrayLevel[0.8501]}, {0.9030899869919434, GrayLevel[0.8501]}, {0.9542425094393249, GrayLevel[0.8501]}}}, Frame -> True, Axes -> None, AspectRatio -> Full, BaseStyle -> { FontFamily -> "Verdana", FontSize -> 8, FontColor -> GrayLevel[0.5], ScriptSizeMultipliers -> 0.2, ScriptMinSize -> 6}, PlotRangePadding -> None, PlotRangeClipping -> True, Background -> GrayLevel[1], FrameStyle -> GrayLevel[0.5], ImagePadding -> 1, FrameTicks -> None], {None, "loss", "rounds"}], GeneralUtilities`TightLabeled[ Graphics[{ AbsoluteThickness[1.25], {{ Hue[0.083, 1., 1.], Line[CompressedData[" 1:eJxdV1lwVFUQHZBFQPYMYERMCBAYQsi+J31nH7egaNzALYpKXPjjCytU+YVf VGnALa644YobuJGo8EWZvDfJIBQl+IOlERVBFFDUzOmek5KpVzX15p13bt/u 0+f2FLavW7lmdCAQKBkVCGS/b2jLfvbK3tZje1qPHRa9z8jmcWuHr8MSwOc7 Obpq9/D1rT0/Ko92H9jUfeCQvLk9+xkS/Nx2SEYBf1wad8wcvg7a8xNSlXpk +DogNwJ3yvAHZTTwpwXwxm/kLeDPSnvx+uFrv9wE3DnB61X7jT/g9H7Q4hnl 8Nr2tLyB79EuG92m7kGLf4w7cmLZ8JWxdce6XRv27dywb0D0vXHuZPbxkUH5 F5/xTuMfkOuBn0D868BPdKuz6Tg6IP8AP8np/gblOuAnu/XYQFpeA36KW3dm y/A1IOeAn2p8vvFNc7r/3HrTHeDrfItnukO6N/kW7wzy6X5mEq/7zXNaH9/y kec0n2nLV9Dhtt23fM5yK7Ll35O2es2y/KWtXrNdumBy9rJ6zWG+tV5zLH7f 6nWxK0XCfbkA+HyHcqV8eRv4fMZ3M/CXuPxt4eErLWOAn2v5TMs7wM/l/m4B /lLGPxb4ebY/T94Ffp7VMy2rgL/MIfqCXDwFjF/jLeD6up8ClwcB+LbfQody r8rlo9B9j37wLV+FzI/mc77px7d8z2c+tB7zWQ+tV5HD8vme1bPI4XZbrt5F TvszbXooYv5ULwtsfzl9LnBo382e6XMB96f6XOh03Zw+F5qefNPnQsufJyuB X8T9vAr8Iur3b+AX8f1rgS9m/V4Bvthl0Ts3+PIX8MWs5zXAL+b9y8Av5npn gV/sHgaBJyuAX+KwvXEDsg34JYznDPBLuL9W4EPkfwn4kEM4+/rlNPAh6vNq 4JdSTy8Cv5T9/CfwS1m/q4Av4fsvAF/C9f8AvsShPXb0y5XALzP/6JfngV/G +E8Bn3vuyxXAlzq0Z2tangO+1PTmye/Al7L/Lgd+Od9/Fvjl1NdJ4JdTD8hf qIz+ovkts/U8zX9nGfWt9SmzfvC0fpky06dn9S1jvVH/UDnjU32U8zn001nO eFVf5W4rBGX6y5Sbf/ebPsuZL+g3VMF+UX1XML/Qf2cF86X9UcF8on8yFRZP v/VXhfmhp/0XqrT1fevPSvKhzBsrLd4+6+9K+in6P1N5nt9WUY/wj1AV/Ur9 pYr7hf9srGJ+1J+qzJ899a9MFftV/a2a+YHfhqpNj575bTXx8NuN1ay3+m01 /Q1+m6lmvdRva5hv+G2ohuup39bY+ZyWceCvcZo3X3aAv4Z+sRr8NU7z5sl4 8NdSj++Bv9b63ZdbwVNLPV8I/lrq5X3w1HK/t4G/lvwT8Hud0/kmLR+Av45+ fzv461wh5O7JRPDXsb8/BM/I8zvAX8d8TAJ/PfP1EfjrmY87wV/P8+Ai8NdT TzvBX8/9toO/nnqcDP4Grr8L/A30q7vA30D/mQL+Bvrnx+Bv4PlwN/gbmJ+p eK+R58sn4G+kf60BfyPzPw38jVzvU/A0sh73gL+R+Z0OXBPvPwN/k+0nLfeC v4l8M8Cfu/fkc/A3mX765T7wN1EvM4Frph/tBn8zz++14G9mvvPA38z69oC/ mfXrAH8z8xsEfwvr3Qv+Fp5X94O/hXqfBf4Wzh9fgL/F9OfJA+BvoV/MBr+w Pl+CX8j/IHiF9ZoDfqFevsLvQn09BH7h+tBbwFFP0GPQUZ+qV2d+1a96ds7y 0696b3PsH/RDh6NfaL84npfop64cv6f9tt2xvujHXsd5QfvV0R/Qz0OOekG/ B8LUM/wgGGa+1S/C3D/8xIWN31O/aQvzPIMfdYRNzzm/ClM/8LOuMM8vnS/D //fD3jDnCfXLMP0QfjoUZj11vo3Qz+HHwQj3p34dMX/uUz93kfPm6wjXx3nQ EWG+9LyI8HzAedIVoX51Xo2wHjiPeiPUj55XEeYXOhsaWU/rFqU/IJxglH6k 52WU5wmOU4ly/zqvRnn+4DzuiPL/jJ7XUcaL87wrSr/ReTVK/WMe6Ily/tV5 IUo9YZ4Yijr9//q1zatR6gPzSDBG/9B5JcZ8YJ6RGPOv82qM9cI81BGj3+m8 FKNfYp7qilF/Oq/G2H+YV3tirAfm1UyM+cK8OhRjf+m8OsKHeTUYJx7zaihO PsyrEud+dV6Nc33Mqx1xzieYVzvj9CfMq11x1kvn1Tj1i3m1J878YF7NxOmn mFeH4uxPnVfjzC/m1WCC/ox5NZRg/jCvSoJ6SIEnwfW6EX+C+juB+BPMTxLx Jzi/PYP4E5avtPyG+BPUdwLxj9w/jfgTrO9xxJ+w+blP4og/yX55CvEn+f/5 V8SfdLCTPX0SQ/xJ9s+TiD9Jf/4F8Sc5v0YRf9LW8+QJxJ/k+fcz4k9yHo8g /iTzvxXxjzw/hviTrH8Y8afo91sQf4r4nxB/ivrSQqbo712IP0W/HEL8Kf6/ gEF0pWy/A/I44k9xvv0R8aeohxbEn6JeHkP8Kcb7A+JPsd7/Ab3OooM= "]]}, {}}, {{ Hue[0.59, 0.7, 0.75], Line[CompressedData[" 1:eJxTTMoPSmViYGAwAmIQXViSMan+4Cv7h1Ui69wfXrXfILlsV1HeX/tCW67r iwsu20/+w31NaB+bQzVIuuqSvf+bI35vjvA59B/6qhHTf9Ge9d2+x49ShR1m zQQCyYv2j1OlF/hniju0vg7cIcd60f7g3fKYPTnSDpoxQB0rLtjPC70ioVkg 7yAPUjDvgv2+tBmvTu1UhOifecF+7aJjHo27leHyc+59WWa+V9UB6Bigiy7Y d0spA61Ud0gDgWcX7KvCApOXHNB0gLj/gj3IN1GHtB3OngECn4v24efWygse 0XUI2iHX+vriBXs3rju1x4/qO9iBvLf3gr00yLjHBg6HQd45dMFeyFFGdtYT Q7j9nPHGPqVPjaD+v2DPWOtVHfDM2AHkHY07F+x/zEpcpf3cxIEBDC7af9hR cZPthSncvufX+jkevTCD8++DvPPSHG4/MHCBIWABt++swdWpxa8tHUDStl0X 7I/6gQLcCq5+Tw7zF8231nD3bOkCBYgNJDzSLtivXmEY9OCdLcQ9DRfsF4OC 870dNDwu2M96Er9h2gd7iHsCL9h7AmV3VDhA4qv1gr2LgHu72k8HuHsc5l+S m1LpCA9fG724bYy/HCHmnblgb7H3pW9+lRM8fkChdeeXEzw+9W8zAEPMGe5+ 7awekR2/nR08QAYev2Cv/lN8jWqNiwPYOTsu2Ct1LHae/McFHh9y4iATXB2+ gbilF+ylwMnTFZ6eRE1BMeoGDy/BIxcXev5zcwCFjsi6C/a8wbGW2+vc4eo5 H724oPIfwWcBZwAPhyUg5y++YA+JR0+4fgBZn4Tu "]]}, { AbsolutePointSize[2.735294117647059], Hue[0.59, 0.7, 0.65], Point[CompressedData[" 1:eJxTTMoPSmViYGAwAmIQXViSMan+4Cv7h1Ui69wfXrXfILlsV1HeX/tCW67r iwsu20/+w31NaB+bQzVIuuqSvf+bI35vjvA59B/6qhHTf9Ge9d2+x49ShR1m zQQCyYv2j1OlF/hniju0vg7cIcd60f7g3fKYPTnSDpoxQB0rLtjPC70ioVkg 7yAPUjDvgv2+tBmvTu1UhOifecF+7aJjHo27leHyc+59WWa+V9UB6Bigiy7Y d0spA61Ud0gDgWcX7KvCApOXHNB0gLj/gj3IN1GHtB3OngECn4v24efWygse 0XUI2iHX+vriBXs3rju1x4/qO9iBvLf3gr00yLjHBg6HQd45dMFeyFFGdtYT Q7j9nPHGPqVPjaD+v2DPWOtVHfDM2AHkHY07F+x/zEpcpf3cxIEBDC7af9hR cZPthSncvufX+jkevTCD8++DvPPSHG4/MHCBIWABt++swdWpxa8tHUDStl0X 7I/6gQLcCq5+Tw7zF8231nD3bOkCBYgNJDzSLtivXmEY9OCdLcQ9DRfsF4OC 870dNDwu2M96Er9h2gd7iHsCL9h7AmV3VDhA4qv1gr2LgHu72k8HuHsc5l+S m1LpCA9fG724bYy/HCHmnblgb7H3pW9+lRM8fkChdeeXEzw+9W8zAEPMGe5+ 7awekR2/nR08QAYev2Cv/lN8jWqNiwPYOTsu2Ct1LHae/McFHh9y4iATXB2+ gbilF+ylwMnTFZ6eRE1BMeoGDy/BIxcXev5zcwCFjsi6C/a8wbGW2+vc4eo5 H724oPIfwWcBZwAPhyUg5y++YA+JR0+4fgBZn4Tu "]]}}}, Prolog -> {{ Text[10, Offset[{-2, -2}, {10., 0.7423500000000001}], {1., 1.}], Text[20, Offset[{-2, -2}, {20., 0.7423500000000001}], {1., 1.}], Text[30, Offset[{-2, -2}, {30., 0.7423500000000001}], {1., 1.}], Text[40, Offset[{-2, -2}, {40., 0.7423500000000001}], {1., 1.}], Text[50, Offset[{-2, -2}, {50., 0.7423500000000001}], {1., 1.}]}, { Text[ PercentForm[0.], Offset[{3, 0}, {0, 0.}], {-1, -1.}], Text[ PercentForm[0.2], Offset[{3, 0}, {0, 0.2}], {-1, -1.}], Text[ PercentForm[0.4], Offset[{3, 0}, {0, 0.4}], {-1, -1.}], Text[ PercentForm[0.6], Offset[{3, 0}, {0, 0.6}], {-1, -1.}]}}, PlotRange -> {{ 0, 50.00213675213675}, {-0.0073500000000000015`, 0.7423500000000001}}, ImageSize -> {302, 182.}, GridLines -> {{10., 20., 30., 40., 50.}, {{0., GrayLevel[0.8501]}, {0.05, GrayLevel[0.8501]}, {0.1, GrayLevel[0.8501]}, {0.15, GrayLevel[0.8501]}, {0.2, GrayLevel[0.8501]}, {0.2, GrayLevel[0.8501]}, {0.25, GrayLevel[0.8501]}, {0.3, GrayLevel[0.8501]}, {0.35, GrayLevel[0.8501]}, {0.4, GrayLevel[0.8501]}, {0.4, GrayLevel[0.8501]}, {0.45, GrayLevel[0.8501]}, {0.5, GrayLevel[0.8501]}, {0.55, GrayLevel[0.8501]}, {0.6, GrayLevel[0.8501]}, {0.6, GrayLevel[0.8501]}, {0.65, GrayLevel[0.8501]}, {0.7, GrayLevel[0.8501]}, {0.75, GrayLevel[0.8501]}, {0.8, GrayLevel[0.8501]}, {0., GrayLevel[0.3001]}, {0.2, GrayLevel[0.3001]}, {0.4, GrayLevel[0.3001]}, {0.6, GrayLevel[0.3001]}, {0.8, GrayLevel[0.3001]}}}, Frame -> True, Axes -> None, AspectRatio -> Full, BaseStyle -> { FontFamily -> "Verdana", FontSize -> 8, FontColor -> GrayLevel[0.5], ScriptSizeMultipliers -> 0.2, ScriptMinSize -> 6}, PlotRangePadding -> None, PlotRangeClipping -> True, Background -> GrayLevel[1], FrameStyle -> GrayLevel[0.5], ImagePadding -> 1, FrameTicks -> None], {None, "error rate", "rounds"}]}}, TagBox[GridBox[{ { DynamicBox[ToBoxes[ Grid[{{ Button[ MouseAppearance[ Style["\:2039", FontColor :> If[ CurrentValue["MouseOver"], RGBColor[0.27, 0.54, 0.79], GrayLevel[0.25]], FontFamily -> "MS Gothic", FontSize -> 14, FontWeight -> "Thin"], "LinkHand"], NeuralNetworks`Private`NetTrain`index$$ = Mod[NeuralNetworks`Private`NetTrain`index$$ - 1, 2, 1], Appearance -> None], Dynamic[ Part[{ Tooltip["loss", Column[{ Row[{"Measurement: ", "loss"}]}]], Tooltip["error rate", Column[{ Row[{"Measurement: ", "ErrorRate"}]}]]}, NeuralNetworks`Private`NetTrain`index$$], TrackedSymbols :> { NeuralNetworks`Private`NetTrain`index$$}], Button[ MouseAppearance[ Style["\:203a", FontColor :> If[ CurrentValue["MouseOver"], RGBColor[0.27, 0.54, 0.79], GrayLevel[0.25]], FontFamily -> "MS Gothic", FontSize -> 14, FontWeight -> "Thin"], "LinkHand"], NeuralNetworks`Private`NetTrain`index$$ = Mod[NeuralNetworks`Private`NetTrain`index$$ + 1, 2, 1], Appearance -> None]}}, ItemSize -> {{3, 20, 3}, {2}}], StandardForm], ImageSizeCache->{265., {7.9150390625, 14.0849609375}}, TrackedSymbols:>{NeuralNetworks`Private`NetTrain`index$$}]}, { DynamicBox[ToBoxes[ Part[ NeuralNetworks`Private`NetTrain`plotList$$, NeuralNetworks`Private`NetTrain`index$$], StandardForm], ImageSizeCache->{313., {95., 99.}}, TrackedSymbols:>{ NeuralNetworks`Private`NetTrain`index$$}]} }, DefaultBaseStyle->"Column", GridBoxAlignment->{"Columns" -> {{Center}}}, GridBoxItemSize->{ "Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}], "Column"], DynamicModuleValues:>{}], Alignment->Center, StripOnInput->False], "\[SpanFromLeft]"}, { ItemBox[ StyleBox[ RowBox[{ GraphicsBox[{}, Axes->None, Background->Hue[0.083, 1., 1.], ImagePadding->{{0, 0}, {5, 2}}, ImageSize->{13, 1}], " ", "\<\"training set\"\>", "\t", GraphicsBox[{}, Axes->None, Background->Hue[0.59, 0.7, 0.75], ImagePadding->{{0, 0}, {5, 2}}, ImageSize->{13, 1}], " ", "\<\"validation set\"\>"}], FontFamily->"Verdana", FontSize->8, FontColor->GrayLevel[0.4]], Alignment->Center, StripOnInput->False], "\[SpanFromLeft]"} }, AutoDelete->False, ColumnsEqual->False, GridBoxAlignment->{ "Columns" -> {Right, {Left}}, "Rows" -> {{Baseline}}}, GridBoxDividers->{"Columns" -> {False, { Opacity[0.15]}, False}}, GridBoxItemSize->{"Columns" -> {8, {26}}, "Rows" -> {{1.}}}, GridBoxSpacings->{"Columns" -> { Offset[0.27999999999999997`], Offset[1.1199999999999999`], { Offset[1.75]}, Offset[0.27999999999999997`]}, "Rows" -> { Offset[0.2], { Offset[0.8]}, Offset[0.2]}}], "Grid"], FrameMargins->{{10, 10}, {10, 5}}, FrameStyle->None, StripOnInput->False], Deploy, DefaultBaseStyle->"Deploy"], Alignment->Center, BaseStyle->{ FontWeight -> "Light", FontSize -> 11, FontFamily -> "DejaVu Sans", NumberMarks -> False, Deployed -> False}, StripOnInput->False]} }, DefaultBaseStyle->"Column", GridBoxAlignment->{"Columns" -> {{Left}}, "Rows" -> {{Baseline}}}, GridBoxDividers->{"Columns" -> {{False}}, "Rows" -> {{False}}}, GridBoxItemSize->{"Columns" -> {{Automatic}}, "Rows" -> {{1.}}}, GridBoxSpacings->{"Columns" -> { Offset[0.27999999999999997`], { Offset[0.5599999999999999]}, Offset[0.27999999999999997`]}, "Rows" -> { Offset[0.2], Offset[0.8], { Offset[0.4]}, Offset[0.2]}}], Background->GrayLevel[1], FrameMargins->{{0, 0}, {0, 0}}, FrameStyle->GrayLevel[0.85], RoundingRadius->5, StripOnInput->False]], "Output", CellChangeTimes->{{3.763468422838024*^9, 3.763468455380887*^9}, { 3.7634863373348703`*^9, 3.763486360245412*^9}, {3.7634865986326637`*^9, 3.7634866457716217`*^9}, 3.763486821639071*^9, 3.7634875458884287`*^9, 3.763488008849189*^9, 3.7637478923111258`*^9, 3.7638166247548733`*^9, 3.763915730005238*^9, 3.7640822719588547`*^9, 3.7645932076953783`*^9, 3.7645936853753023`*^9}, CellLabel->"Out[14]="] }, Open ]], Cell["Compare the results obtained with GPT and with GloVe:", "Text", CellChangeTimes->{{3.728707259494198*^9, 3.7287072707902927`*^9}, { 3.764586920919835*^9, 3.76458693199161*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"Dataset", "[", RowBox[{"<|", RowBox[{ RowBox[{"\"\\"", "\[Rule]", RowBox[{"bertresults", "[", "\"\\"", "]"}]}], ",", RowBox[{"\"\\"", "\[Rule]", RowBox[{"gloveresults", "[", "\"\\"", "]"}]}]}], "|>"}], "]"}]], "Input", CellChangeTimes->{{3.764587142435939*^9, 3.7645871797950993`*^9}, { 3.764587214969513*^9, 3.7645872419460917`*^9}, {3.764592993679694*^9, 3.764593001496254*^9}}, CellLabel->"In[16]:="], Cell[BoxData[ TemplateBox[{ TagBox[ DynamicModuleBox[{ TypeSystem`NestedGrid`PackagePrivate`$state$$ = Data`UnorderedAssociation[ "DisplayedRowCount" -> 20, "RowCount" -> 2, "Length" -> 2, "InitialType" -> TypeSystem`Assoc[ TypeSystem`Atom[String], TypeSystem`Struct[{"Loss", "ErrorRate"}, { TypeSystem`Atom[Real], TypeSystem`Atom[Real]}], 2], "DisplayedColumnCount" -> 20, "DataRowCount" -> 0, "Shape" -> TypeSystem`PackageScope`HeaderShape[ Association[{All, TypeSystem`PackageScope`KeyDummy[0]} -> 1, {All, "Loss"} -> 1, {All, "ErrorRate"} -> 1], TypeSystem`PackageScope`Limited[ TypeSystem`PackageScope`ColumnShape[ TypeSystem`PackageScope`KeyValueShape[ TypeSystem`PackageScope`AtomShape[48.00000000000001], TypeSystem`PackageScope`RowShape[ Association[ "Loss" -> TypeSystem`PackageScope`AtomShape[60], "ErrorRate" -> TypeSystem`PackageScope`AtomShape[60]]]]], 2, {}]], "DataColumnCount" -> 0, "InitialData" -> TypeSystem`PackageScope`CacheHandle[502547336], "Type" -> TypeSystem`Assoc[ TypeSystem`Atom[String], TypeSystem`Struct[{"Loss", "ErrorRate"}, { TypeSystem`Atom[Real], TypeSystem`Atom[Real]}], 2], "HandleQ" -> True, "Meta" -> {}, "RowTarget" -> 20, "Data" -> TypeSystem`PackageScope`CacheHandle[502547336]], TypeSystem`NestedGrid`PackagePrivate`$path$$ = {}, TypeSystem`NestedGrid`PackagePrivate`$pos$$ = 1, TypeSystem`NestedGrid`PackagePrivate`$grid$$ = DynamicModule[{ TypeSystem`NestedGrid`PackagePrivate`renderedGrid = Deploy[ Style[ Grid[{{ Item[ Annotation[ EventHandler[ MouseAppearance[ Mouseover[ Pane[ Style[ RawBoxes[ StyleBox["\"\"", FontColor -> GrayLevel[0.4]]], Italic, GrayLevel[0.7]], {{20, Full}, 15}], Style[ Pane[ Style[ RawBoxes[ StyleBox[ "\"\"", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], Italic, GrayLevel[0.7]], {{20, Full}, 15}], FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966][{All, Keys}]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[All, Keys]], "Mouse"], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Annotation[ EventHandler[ MouseAppearance[ Mouseover[ Pane[ RawBoxes[ StyleBox["\"Loss\"", FontColor -> GrayLevel[0.4]]], {{ 20, Full}, 15}], Style[ Pane[ RawBoxes[ StyleBox[ "\"Loss\"", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], {{20, Full}, 15}], FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966][{All, "Loss"}]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[All, "Loss"]], "Mouse"], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Item[ Annotation[ EventHandler[ MouseAppearance[ Mouseover[ Pane[ RawBoxes[ StyleBox[ "\"ErrorRate\"", FontColor -> GrayLevel[0.4]]], {{ 20, Full}, 15}], Style[ Pane[ RawBoxes[ StyleBox[ "\"ErrorRate\"", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], {{20, Full}, 15}], FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966][{All, "ErrorRate"}]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[All, "ErrorRate"]], "Mouse"], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}]}, { Item[ Pane[ Annotation[ EventHandler[ MouseAppearance[ Mouseover["BERT", Style[ "BERT", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966][{ Key["BERT"]}]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ 1188092966][ GeneralUtilities`Slice[ Key["BERT"]]], "Mouse"], ImageSize -> {{48.00000000000001, Full}, Automatic}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Pane[ Annotation[ RawBoxes["0.370675"], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[ Key["BERT"], Key["Loss"]]], "Mouse"], ImageSize -> {{60, Full}, Automatic}, Alignment -> {Left, Baseline}], Pane[ Annotation[ RawBoxes["0.159375"], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[ Key["BERT"], Key["ErrorRate"]]], "Mouse"], ImageSize -> {{60, Full}, Automatic}, Alignment -> {Left, Baseline}]}, { Item[ Pane[ Annotation[ EventHandler[ MouseAppearance[ Mouseover["GloVe", Style[ "GloVe", FontColor -> RGBColor[ 0.27450980392156865`, 0.5372549019607843, 0.792156862745098]]], "LinkHand"], {"MouseClicked", 1} :> TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966][{ Key["GloVe"]}]], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[ 1188092966][ GeneralUtilities`Slice[ Key["GloVe"]]], "Mouse"], ImageSize -> {{48.00000000000001, Full}, Automatic}], Background -> GrayLevel[0.95], Alignment -> {Left, Baseline}], Pane[ Annotation[ RawBoxes["0.515398"], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[ Key["GloVe"], Key["Loss"]]], "Mouse"], ImageSize -> {{60, Full}, Automatic}, Alignment -> {Left, Baseline}], Pane[ Annotation[ RawBoxes["0.260625"], TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ GeneralUtilities`Slice[ Key["GloVe"], Key["ErrorRate"]]], "Mouse"], ImageSize -> {{60, Full}, Automatic}, Alignment -> {Left, Baseline}]}}, BaseStyle -> {ContextMenu -> { MenuItem["Copy position to clipboard", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentPosition[ TypeSystem`NestedGrid`PackagePrivate`copyClip]], MenuEvaluator -> Automatic], MenuItem["Copy data to clipboard", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentData[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`copyClip]], MenuEvaluator -> Automatic], Delimiter, MenuItem["Paste position in new cell", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentPosition[ TypeSystem`NestedGrid`PackagePrivate`cellPaste]], MenuEvaluator -> Automatic], MenuItem["Paste data in new cell", KernelExecute[ TypeSystem`NestedGrid`PackagePrivate`toCurrentData[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`cellPaste]], MenuEvaluator -> Automatic]}, FontFamily -> "Verdana", FontSize -> 12}, Alignment -> Left, RowMinHeight -> 1.5, Dividers -> All, FrameStyle -> GrayLevel[0.85], BaseStyle -> {FontFamily -> "Verdana", FontSize -> 12}], LineBreakWithin -> False, ContextMenu -> {}, NumberMarks -> False, ShowAutoStyles -> False]], TypeSystem`NestedGrid`PackagePrivate`initialQ = True, TypeSystem`NestedGrid`PackagePrivate`self = $Failed}, Dynamic[ TypeSystem`NestedGrid`PackagePrivate`setupViewPath[ TypeSystem`NestedGrid`PackagePrivate`$path$$, If[ Not[TypeSystem`NestedGrid`PackagePrivate`initialQ], Module[{TypeSystem`NestedGrid`PackagePrivate`tmpGrid$ = $Failed, TypeSystem`NestedGrid`PackagePrivate`tmpData$ = TypeSystem`NestedGrid`PackagePrivate`LookupCacheValue[ TypeSystem`NestedGrid`PackagePrivate`$state$$["Data"]]}, TypeSystem`NestedGrid`PackagePrivate`tmpGrid$ = If[TypeSystem`NestedGrid`PackagePrivate`tmpData$ === TypeSystem`NestedGrid`PackagePrivate`$NotCached, TypeSystem`NestedGrid`PackagePrivate`renderedGrid, TypeSystem`NestedGrid`PackagePrivate`renderGrid[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966][ TypeSystem`NestedGrid`PackagePrivate`tmpData$]]; If[ Not[ FailureQ[TypeSystem`NestedGrid`PackagePrivate`tmpGrid$]], TypeSystem`NestedGrid`PackagePrivate`renderedGrid = TypeSystem`NestedGrid`PackagePrivate`tmpGrid$]; Null]]; TypeSystem`NestedGrid`PackagePrivate`initialQ = False; TypeSystem`NestedGrid`PackagePrivate`$pos$$; If[ FailureQ[TypeSystem`NestedGrid`PackagePrivate`renderedGrid], TypeSystem`SparseGrid[ TypeSystem`H["(data no longer present)"]], If[GeneralUtilities`$DebugMode, Row[{TypeSystem`NestedGrid`PackagePrivate`renderedGrid, " ", TypeSystem`NestedGrid`PackagePrivate`formatState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$]}], TypeSystem`NestedGrid`PackagePrivate`renderedGrid]]], TrackedSymbols :> {TypeSystem`NestedGrid`PackagePrivate`$pos$$}], DynamicModuleValues :> {}], TypeSystem`NestedGrid`PackagePrivate`$posCell$$, TypeSystem`NestedGrid`PackagePrivate`$topBar$$ = Dynamic[ TypeSystem`NestedGrid`PackagePrivate`alignBar[ TypeSystem`NestedGrid`PackagePrivate`$state$$][ TypeSystem`NestedGrid`PackagePrivate`makeFramedBar[ TypeSystem`PackageScope`Pathbar[ TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`updateState[ TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$path$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$, 1188092966]]]], TrackedSymbols :> {TypeSystem`NestedGrid`PackagePrivate`$path$$}], TypeSystem`NestedGrid`PackagePrivate`$bottomBar$$ = Framed[ Dynamic[ Replace[ MouseAnnotation[Null], { TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][Null] -> "", TypeSystem`NestedGrid`PackagePrivate`$SliceMarker[1188092966][ Pattern[TypeSystem`NestedGrid`PackagePrivate`path, Blank[]]] :> ( TypeSystem`NestedGrid`PackagePrivate`$lastPath = TypeSystem`NestedGrid`PackagePrivate`path; TypeSystem`NestedGrid`PackagePrivate`makePathTrail[ TypeSystem`NestedGrid`PackagePrivate`path, TypeSystem`NestedGrid`PackagePrivate`makePathElements]), Null :> Spacer[10], Blank[] :> Spacer[10]}], TrackedSymbols :> {}], FrameStyle -> None, ImageMargins -> 0, FrameMargins -> 0, Alignment -> Top, ImageSize -> {Automatic, 14}]}, DynamicBox[ ToBoxes[ Column[ If[TypeSystem`NestedGrid`PackagePrivate`$path$$ == {}, { TypeSystem`NestedGrid`PackagePrivate`displayScrollBar[ TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$], TypeSystem`NestedGrid`PackagePrivate`onDesktopShow[ TypeSystem`NestedGrid`PackagePrivate`alignBar[ TypeSystem`NestedGrid`PackagePrivate`$state$$][ TypeSystem`NestedGrid`PackagePrivate`$bottomBar$$]]}, { TypeSystem`NestedGrid`PackagePrivate`$topBar$$, TypeSystem`NestedGrid`PackagePrivate`displayScrollBar[ TypeSystem`NestedGrid`PackagePrivate`$grid$$, TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$pos$$], TypeSystem`NestedGrid`PackagePrivate`onDesktopShow[ TypeSystem`NestedGrid`PackagePrivate`alignBar[ TypeSystem`NestedGrid`PackagePrivate`$state$$][ TypeSystem`NestedGrid`PackagePrivate`$bottomBar$$]]}], Spacings -> If[TypeSystem`NestedGrid`PackagePrivate`$path$$ == {}, 0, {{}, {0, 0.05, 0}}]], StandardForm], ImageSizeCache -> {198., {42.634033203125, 48.365966796875}}, TrackedSymbols :> { TypeSystem`NestedGrid`PackagePrivate`$state$$, TypeSystem`NestedGrid`PackagePrivate`$grid$$}], BaseStyle -> {LineBreakWithin -> False}, Deinitialization :> TypeSystem`NestedGrid`PackagePrivate`deleteState[ TypeSystem`NestedGrid`PackagePrivate`$state$$], DynamicModuleValues :> {}, Initialization :> Block[{$ContextPath = $ContextPath}, Needs["TypeSystem`"]]], Deploy, DefaultBaseStyle -> "Deploy"]}, "CopyTag", DisplayFunction->(#& ), InterpretationFunction->("Dataset[<>]"& )]], "Output", CellChangeTimes->{3.76459313029712*^9, 3.764593691617923*^9}, CellLabel->"Out[16]="] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell["Net information", "Subsection", CellChangeTimes->{{3.7259790342745*^9, 3.7259790377150373`*^9}}], Cell["Inspect the number of parameters of all arrays in the net:", "Text", CellChangeTimes->{{3.725979178627933*^9, 3.7259791789799137`*^9}, { 3.7259792303321257`*^9, 3.725979245908435*^9}, {3.730020487832671*^9, 3.730020494904923*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetInformation", "[", RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{3.735313527695961*^9, 3.763323159740761*^9, 3.7634042165150414`*^9, 3.7634170397885*^9, 3.7978486436663446`*^9}], Cell[BoxData[ InterpretationBox[ TagBox[ FrameBox[GridBox[{ { ItemBox[ TagBox[ RowBox[{"\[LeftAssociation]", RowBox[{ RowBox[{ RowBox[{"{", RowBox[{"\<\"embedding\"\>", ",", "\<\"embeddingpos\"\>", ",", "\<\"Weights\"\>"}], "}"}], "\[Rule]", "393216"}], ",", RowBox[{ RowBox[{"{", RowBox[{"\<\"embedding\"\>", ",", "\<\"embeddingsegments\"\>", ",", "\<\"Weights\"\>"}], "}"}], "\[Rule]", "1536"}], ",", RowBox[{ RowBox[{"{", RowBox[{"\<\"embedding\"\>", ",", "\<\"embeddingtokens\"\>", ",", "\<\"Weights\"\>"}], "}"}], "\[Rule]", "23440896"}], ",", RowBox[{ RowBox[{"{", RowBox[{"\<\"embedding\"\>", ",", "\<\"normalize\"\>", ",", "\<\"Biases\"\>"}], "}"}], "\[Rule]", "768"}], ",", TemplateBox[{"982"}, "OutputSizeLimit`Skeleton"], ",", RowBox[{ RowBox[{"{", RowBox[{"\<\"encoder\"\>", ",", "9", ",", "1", ",", "\<\"attention\"\>", ",", "9", ",", "\<\"query\"\>", ",", "\<\"Net\"\>", ",", "\<\"Weights\"\>"}], "}"}], "\[Rule]", "49152"}], ",", RowBox[{ RowBox[{"{", RowBox[{"\<\"encoder\"\>", ",", "9", ",", "1", ",", "\<\"attention\"\>", ",", "9", ",", "\<\"value\"\>", ",", "\<\"Net\"\>", ",", "\<\"Biases\"\>"}], "}"}], "\[Rule]", "64"}], ",", RowBox[{ RowBox[{"{", RowBox[{"\<\"encoder\"\>", ",", "9", ",", "1", ",", "\<\"attention\"\>", ",", "9", ",", "\<\"value\"\>", ",", "\<\"Net\"\>", ",", "\<\"Weights\"\>"}], "}"}], "\[Rule]", "49152"}]}], "\[RightAssociation]"}], Short[#, 5]& ], BaseStyle->{Deployed -> False}, StripOnInput->False]}, {GridBox[{ { PaneBox[ TagBox[ TooltipBox[ StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource[ "FEStrings", "sizeBriefExplanation"], StandardForm], ImageSizeCache->{59., {2., 8.}}], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLText", StripOnInput->False], StyleBox[ DynamicBox[ ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeExplanation"], StandardForm]], DynamicUpdating -> True, LineIndent -> 0, LinebreakAdjustments -> {1., 100, 0, 0, 0}, LineSpacing -> {1, 2}, StripOnInput -> False]], Annotation[#, Style[ Dynamic[ FEPrivate`FrontEndResource["FEStrings", "sizeExplanation"]], DynamicUpdating -> True, LineIndent -> 0, LinebreakAdjustments -> {1., 100, 0, 0, 0}, LineSpacing -> {1, 2}], "Tooltip"]& ], Alignment->Center, BaselinePosition->Baseline, ImageSize->{Automatic, {25, Full}}], ButtonBox[ PaneSelectorBox[{False-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeShowLess"], StandardForm], ImageSizeCache->{51., {0., 8.}}], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControl", StripOnInput->False], True-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeShowLess"], StandardForm]], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControlActive", StripOnInput->False]}, Dynamic[ CurrentValue["MouseOver"]], Alignment->Center, FrameMargins->0, ImageSize->{Automatic, {25, Full}}], Appearance->None, BaselinePosition->Baseline, ButtonFunction:>OutputSizeLimit`ButtonFunction[ OutputSizeLimit`Defer, 128, 31510516668702153272, 5/2], Enabled->True, Evaluator->Automatic, Method->"Queued"], ButtonBox[ PaneSelectorBox[{False-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeShowMore"], StandardForm], ImageSizeCache->{56., {0., 8.}}], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControl", StripOnInput->False], True-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeShowMore"], StandardForm]], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControlActive", StripOnInput->False]}, Dynamic[ CurrentValue["MouseOver"]], Alignment->Center, FrameMargins->0, ImageSize->{Automatic, {25, Full}}], Appearance->None, BaselinePosition->Baseline, ButtonFunction:>OutputSizeLimit`ButtonFunction[ OutputSizeLimit`Defer, 128, 31510516668702153272, 5 2], Enabled->True, Evaluator->Automatic, Method->"Queued"], ButtonBox[ PaneSelectorBox[{False-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeShowAll"], StandardForm], ImageSizeCache->{42., {0., 8.}}], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControl", StripOnInput->False], True-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeShowAll"], StandardForm]], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControlActive", StripOnInput->False]}, Dynamic[ CurrentValue["MouseOver"]], Alignment->Center, FrameMargins->0, ImageSize->{Automatic, {25, Full}}], Appearance->None, BaselinePosition->Baseline, ButtonFunction:>OutputSizeLimit`ButtonFunction[ OutputSizeLimit`Defer, 128, 31510516668702153272, Infinity], Enabled->True, Evaluator->Automatic, Method->"Queued"], ButtonBox[ PaneSelectorBox[{False-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeChangeLimit"], StandardForm], ImageSizeCache->{77., {0., 8.}}], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControl", StripOnInput->False], True-> StyleBox[ StyleBox[ DynamicBox[ToBoxes[ FEPrivate`FrontEndResource["FEStrings", "sizeChangeLimit"], StandardForm]], StripOnInput->False, DynamicUpdating->True, LineSpacing->{1, 2}, LineIndent->0, LinebreakAdjustments->{1., 100, 0, 0, 0}], "OSLControlActive", StripOnInput->False]}, Dynamic[ CurrentValue["MouseOver"]], Alignment->Center, FrameMargins->0, ImageSize->{Automatic, {25, Full}}], Appearance->None, BaselinePosition->Baseline, ButtonFunction:>FrontEndExecute[{ FrontEnd`SetOptions[ FrontEnd`$FrontEnd, FrontEnd`PreferencesSettings -> {"Page" -> "Advanced"}], FrontEnd`FrontEndToken["PreferencesDialog"]}], Evaluator->None, Method->"Preemptive"]} }, AutoDelete->False, FrameStyle->GrayLevel[0.85], GridBoxDividers->{"Columns" -> {False, {True}}}, GridBoxItemSize->{"Columns" -> {{Automatic}}, "Rows" -> {{Automatic}}}, GridBoxSpacings->{"Columns" -> {{2}}}]} }, DefaultBaseStyle->"Column", GridBoxAlignment->{"Columns" -> {{Left}}, "Rows" -> {{Baseline}}}, GridBoxDividers->{"Columns" -> {{False}}, "Rows" -> {{False}}}, GridBoxItemSize->{"Columns" -> {{Automatic}}, "Rows" -> {{1.}}}, GridBoxSpacings->{"Columns" -> { Offset[0.27999999999999997`], { Offset[0.5599999999999999]}, Offset[0.27999999999999997`]}, "Rows" -> { Offset[0.2], Offset[1.2], { Offset[0.4]}, Offset[0.2]}}], BaseStyle->"OutputSizeLimit", FrameMargins->{{12, 12}, {0, 15}}, FrameStyle->GrayLevel[0.85], RoundingRadius->5, StripOnInput->False], Deploy, DefaultBaseStyle->"Deploy"], If[31510516668702153272 === $SessionID, Out[128], Message[ MessageName[Syntax, "noinfoker"]]; Missing["NotAvailable"]; Null]]], "Output", CellChangeTimes->{3.733052474940061*^9, 3.735313599538179*^9, 3.754054778193301*^9, 3.761571611987797*^9, 3.763323161881771*^9, 3.763364821727583*^9, 3.763417673287936*^9, 3.763748113553153*^9, 3.7638166282048187`*^9, 3.7639157340495367`*^9, 3.76408227746482*^9}, CellLabel->"Out[128]="] }, Open ]], Cell["Obtain the total number of parameters:", "Text", CellChangeTimes->{{3.725979366532411*^9, 3.725979381541073*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetInformation", "[", RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{3.7353135255040083`*^9, 3.763323166972258*^9, 3.763404216518153*^9, 3.763417039797091*^9, 3.797848643674078*^9}], Cell[BoxData["108891648"], "Output", CellChangeTimes->{3.733052474988936*^9, 3.73531359982904*^9, 3.754054779571116*^9, 3.761571612030385*^9, 3.7633231678831253`*^9, 3.763364822761517*^9, 3.763417674947563*^9, 3.763748113669395*^9, 3.76381662855267*^9, 3.7639157343502607`*^9, 3.764082277755018*^9}, CellLabel->"Out[129]="] }, Open ]], Cell["Obtain the layer type counts:", "Text", CellChangeTimes->{{3.725979260148272*^9, 3.725979271188596*^9}, 3.725979320962792*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetInformation", "[", RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{3.735313523584717*^9, 3.763323173868146*^9, 3.76340421652094*^9, 3.763417039798827*^9, 3.7978486436755323`*^9}], Cell[BoxData[ RowBox[{"\[LeftAssociation]", RowBox[{ RowBox[{"PartLayer", "\[Rule]", "2"}], ",", RowBox[{"EmbeddingLayer", "\[Rule]", "3"}], ",", RowBox[{"NeuralNetworks`SequenceIndicesLayer", "\[Rule]", "1"}], ",", RowBox[{"ElementwiseLayer", "\[Rule]", "157"}], ",", RowBox[{"ThreadingLayer", "\[Rule]", "25"}], ",", RowBox[{"NormalizationLayer", "\[Rule]", "25"}], ",", RowBox[{"DropoutLayer", "\[Rule]", "25"}], ",", RowBox[{"LinearLayer", "\[Rule]", "468"}], ",", RowBox[{"DotLayer", "\[Rule]", "144"}], ",", RowBox[{"CatenateLayer", "\[Rule]", "12"}]}], "\[RightAssociation]"}]], "Output", CellChangeTimes->{3.733052475040373*^9, 3.735313599873169*^9, 3.754054780653843*^9, 3.7615716120398293`*^9, 3.763323174853578*^9, 3.763364824120543*^9, 3.763417676849184*^9, 3.763748113720254*^9, 3.763816628651339*^9, 3.7639157345057573`*^9, 3.764082277867124*^9}, CellLabel->"Out[130]="] }, Open ]], Cell["Display the summary graphic:", "Text", CellChangeTimes->{{3.726915136428933*^9, 3.726915145334137*^9}}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"NetInformation", "[", RowBox[{ RowBox[{ "NetModel", "[", "\"\\"", "]"}], ",", "\"\\""}], "]"}]], "Input", CellChangeTimes->{3.735313522304438*^9, 3.763323186619059*^9, 3.763404216523693*^9, 3.76341703980059*^9, 3.797848643677421*^9}], Cell[BoxData[ GraphicsBox[{ {GrayLevel[0.7], LineBox[{{-3., 1.8369701987210297`*^-16}, {-2., 1.2246467991473532`*^-16}}], LineBox[{{-2., 1.2246467991473532`*^-16}, {-1., 6.123233995736766*^-17}}], LineBox[{{-1., 6.123233995736766*^-17}, {0., 0.}}]}, {InsetBox[ GraphicsBox[ {GrayLevel[0.7], PolygonBox[{{-1., 0.3333333333333333}, {0., 0.}, {-1., -0.3333333333333333}, {-0.6666666666666666, 0.}, {-1., 0.3333333333333333}}]}], \ {-2.4000000000000004, 1.4695761589768238*^-16}, {-0.25, 0}, Offset[8], { 1., -6.123233995736765*^-17}], InsetBox[ GraphicsBox[ {GrayLevel[0.7], PolygonBox[{{-1., 0.3333333333333333}, {0., 0.}, {-1., -0.3333333333333333}, {-0.6666666666666666, 0.}, {-1., 0.3333333333333333}}]}], {-1.4, 8.572527594031473*^-17}, {-0.25, 0}, Offset[8], {1., -6.123233995736766*^-17}], InsetBox[ GraphicsBox[ {GrayLevel[0.7], PolygonBox[{{-1., 0.3333333333333333}, {0., 0.}, {-1., -0.3333333333333333}, {-0.6666666666666666, 0.}, {-1., 0.3333333333333333}}]}], {-0.4, 2.4492935982947065*^-17}, {-0.25, 0}, Offset[8], {1., -6.123233995736766*^-17}]}, {EdgeForm[AbsoluteThickness[1]], StyleBox[{ {GrayLevel[0.85], EdgeForm[GrayLevel[0.5]], DiskBox[{-3., 1.8369701987210297*^-16}, Offset[{2., 2.}, {0., 0.}]]}, { {EdgeForm[GrayLevel[0.85]], FaceForm[RGBColor[0.9, 0.9, 0.9]], RectangleBox[Offset[{-12, -12}, {-2., 1.2246467991473532*^-16}], Offset[{12, 12}, {-2., 1.2246467991473532*^-16}]]}, InsetBox[ GraphicsBox[ {Thickness[0.0149], {GrayLevel[0.5], Thickness[0.03], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{1, 4, 3}}}, {{{39.192120463394616`, 30.51535353195944}, {39.192120463394616`, 30.51535353195944}, { 38.69442248591872, 35.17935629818005}, {28.246036107633145`, 35.21413676830481}}}, CurveClosed->{0}], JoinedCurveBox[{{{1, 4, 3}}}, {{{39.335627149246974`, 22.022653388846493`}, {39.335627149246974`, 22.022653388846493`}, {39.028664759644585`, 18.313538367818456`}, {28.898905902765012`, 18.118321787764348`}}}, CurveClosed->{0}]}, GeometricTransformationBox[GraphicsGroupBox[ {Thickness[0.0149], GraphicsGroupBox[{ {FaceForm[RGBColor[0.631, 0.8160000000000001, 0.498]], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{15.5, 40.5}, {22.5, 40.5}, {22.5, 33.5}, {15.5, 33.5}}}]}, {FaceForm[RGBColor[0.40800000000000003`, 0.627, 0.196]], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{23., 41.}, {15., 41.}, {15., 33.}, {23., 33.}}, {{22., 34.}, {16., 34.}, {16., 40.}, {22., 40.}}}]}}]}], {{{{1.6209767772132715`, 0.}, {0., 1.302327038770047}}, {-8.296516058106576, \ -14.790162389666165`}}}], GeometricTransformationBox[GraphicsGroupBox[ {Thickness[0.0149], GraphicsGroupBox[{ {FaceForm[RGBColor[0.38, 0.753, 0.89]], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{15.5, 19.5}, {22.5, 19.5}, {22.5, 12.5}, {15.5, 12.5}}}]}, {FaceForm[RGBColor[0.24300000000000002`, 0.608, 0.71]], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{23., 20.}, {15., 20.}, {15., 12.}, {23., 12.}}, {{22., 13.}, {16., 13.}, {16., 19.}, {22., 19.}}}]}}]}], {{{{1.6446663930506598`, 0.}, {0., 1.2828873028183188`}}, {-8.702521921403617, \ -2.012105908138385}}}], GeometricTransformationBox[GraphicsGroupBox[ {Thickness[0.0149], GraphicsGroupBox[{ {FaceForm[RGBColor[0.584, 0.51, 0.8390000000000001]], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{30.5, 30.5}, {37.5, 30.5}, {37.5, 23.5}, {30.5, 23.5}}}]}, {FaceForm[RGBColor[0.4, 0.396, 0.659]], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{38., 31.}, {30., 31.}, {30., 23.}, {38., 23.}}, {{37., 24.}, {31., 24.}, {31., 30.}, {37., 30.}}}]}}]}], {{{{1.5182866155506958`, 0.}, {0., 1.2065857650683098`}}, {-11.402865979580374`, \ -6.527851836053724}}}]}, AspectRatio->1, ImageMargins->0., ImagePadding->0, ImageSize->15, PlotRange->All, PlotRangePadding->0], {-2., 1.2246467991473532*^-16}]}, { {EdgeForm[GrayLevel[0.85]], FaceForm[RGBColor[0.9, 0.9, 0.9]], RectangleBox[Offset[{-12, -12}, {-1., 6.123233995736766*^-17}], Offset[{12, 12}, {-1., 6.123233995736766*^-17}]]}, InsetBox[ GraphicsBox[ {Thickness[0.014925373134328358`], {GrayLevel[0.5], Thickness[0.018686567164179105`], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{0, 2, 0}}}, {{{23., 91.5}, {30., 91.5}}}, CurveClosed->{0}]}, {GrayLevel[0.5], Thickness[0.018686567164179105`], Opacity[1.], JoinForm[{"Miter", 10.}], JoinedCurveBox[{{{0, 2, 0}}}, {{{37., 91.5}, {44., 91.5}}}, CurveClosed->{0}]}, {FaceForm[{RGBColor[0.631, 0.8160000000000001, 0.498], Opacity[ 1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{15.5, 103.5}, {23.5, 103.5}, {23.5, 79.5}, {15.5, 79.5}}}]}, {FaceForm[{RGBColor[0.40800000000000003`, 0.627, 0.196], Opacity[ 1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{24., 104.}, {15., 104.}, {15., 79.}, {24., 79.}}, {{23., 80.}, {16., 80.}, {16., 103.}, {23., 103.}}}]}, {FaceForm[{RGBColor[0.38, 0.753, 0.89], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{29.5, 103.5}, {37.5, 103.5}, {37.5, 79.5}, {29.5, 79.5}}}]}, {FaceForm[{RGBColor[0.24300000000000002`, 0.608, 0.71], Opacity[ 1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{38., 104.}, {29., 104.}, {29., 79.}, {38., 79.}}, {{37., 80.}, {30., 80.}, {30., 103.}, {37., 103.}}}]}, {FaceForm[{RGBColor[0.584, 0.51, 0.8390000000000001], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{43.5, 103.5}, {51.5, 103.5}, {51.5, 79.5}, {43.5, 79.5}}}]}, {FaceForm[{RGBColor[0.4, 0.396, 0.659], Opacity[1.]}], FilledCurveBox[{{{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}, {{0, 2, 0}, {0, 1, 0}, {0, 1, 0}}}, {{{52., 104.}, {43., 104.}, {43., 79.}, {52., 79.}}, {{51., 80.}, {44., 80.}, {44., 103.}, {51., 103.}}}]}}, AspectRatio->0.7, ImageMargins->0., ImagePadding->0, ImageSize->16, PlotRange->All, PlotRangePadding->1], {-1., 6.123233995736766*^-17}]}, {GrayLevel[0.85], EdgeForm[GrayLevel[0.5]], DiskBox[{0., 0.}, Offset[{2., 2.}, {0., 0.}]]}}, FontSize->10]}, { StyleBox[{ InsetBox["\<\"Input\"\>", Offset[{0, -7}, {-3., 1.8369701987210297*^-16}], {0, 1.}, Automatic, { 1, 0}], InsetBox["\<\"embedding\"\>", Offset[{0, -15}, {-2., 1.2246467991473532*^-16}], {0, 1.}, Automatic, { 1, 0}], InsetBox["\<\"encoder\"\>", Offset[{0, -15}, {-1., 6.123233995736766*^-17}], {0, 1.}, Automatic, { 1, 0}], InsetBox["\<\"Output\"\>", Offset[{0, -7}, {0., 0.}], {0, 1.}, Automatic, {1, 0}]}, StripOnInput->False, FontColor->GrayLevel[0.5]], StyleBox[{InsetBox[ TemplateBox[{"\[Cross]", "\"\[Cross]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"2\""}, "RowWithSeparators"], Offset[{3.3677786976552205*^-16, 5.5}, \ {-2.4000000000000004, 1.4695761589768238*^-16}], {0., -0.17}, Automatic, { 1., -6.123233995736765*^-17}], InsetBox[ TemplateBox[{"\[Cross]", "\"\[Cross]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"768\""}, "RowWithSeparators"], Offset[{3.3677786976552215*^-16, 5.5}, {-1.4, 8.572527594031473*^-17}], {0., -0.17}, Automatic, { 1., -6.123233995736766*^-17}], InsetBox[ TemplateBox[{"\[Cross]", "\"\[Cross]\"", StyleBox["\"n\"", Italic, StripOnInput -> False], "\"768\""}, "RowWithSeparators"], Offset[{3.3677786976552215*^-16, 5.5}, \ {-0.4, 2.4492935982947065*^-17}], {0., -0.17}, Automatic, { 1., -6.123233995736766*^-17}]}, StripOnInput->False, FontSize->8, FontWeight->"Thin", FontColor->GrayLevel[0.7]]}}, AspectRatio->Full, BaseStyle->{TooltipBoxOptions -> {TooltipDelay -> 0.08}}, ImageMargins->{{0, 0}, {0, 0}}, ImagePadding->{{28., 28.}, {28.999999999999996`, 18.}}, ImageSize->{221., 47.00000000000001}, PlotRangePadding->0]], "Output", CellChangeTimes->{3.733052475080224*^9, 3.7353136000977163`*^9, 3.754054782257962*^9, 3.761571612207946*^9, 3.763323187542582*^9, 3.763364825969767*^9, 3.763417679540444*^9, 3.7637481138762627`*^9, 3.763816628963191*^9, 3.763915734750691*^9, 3.764082278063692*^9}, CellLabel->"Out[131]="] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell["Export to MXNet", "Subsection"], Cell[TextData[{ Cell[BoxData[ ButtonBox["Export", BaseStyle->"Link", ButtonData->"ref/Export"]], "InlineFormula"], " the net into a format that can be opened in MXNet:" }], "Text", CellChangeTimes->{3.724526659609247*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"jsonPath", "=", RowBox[{"Export", "[", RowBox[{ RowBox[{"FileNameJoin", "[", RowBox[{"{", RowBox[{"$TemporaryDirectory", ",", "\"\\""}], "}"}], "]"}], ",", RowBox[{ "NetModel", "[", "\"\\"", "]"}], ",", "\"\\""}], "]"}]}]], "Input", CellChangeTimes->{ 3.713712302144353*^9, 3.7137124390312243`*^9, 3.713712513980206*^9, { 3.713712563904804*^9, 3.7137125908504972`*^9}, 3.7325563581999598`*^9, 3.7633232047780943`*^9, 3.763404216526597*^9, 3.763417039808766*^9, 3.797848643682925*^9}], Cell[BoxData["\<\"/private/var/folders/pz/94mxs33x2l512z6wtjbthvy0000_ck/T/\ net.json\"\>"], "Output", CellChangeTimes->{3.713707095322916*^9, 3.714413964226924*^9, 3.7144149188491793`*^9, 3.732556367450382*^9, 3.754054785205312*^9, 3.761571617208313*^9, 3.763323207475675*^9, 3.763364828736257*^9, 3.763417685075694*^9, 3.763748127406538*^9, 3.763816630851956*^9, 3.763915736957664*^9, 3.764082280807953*^9}, CellLabel->"Out[132]="] }, Open ]], Cell[TextData[{ Cell[BoxData[ ButtonBox["Export", BaseStyle->"Link", ButtonData->"ref/Export"]], "InlineFormula"], " also creates a ", StyleBox["net.params ", FontSlant->"Italic"], "file containing parameters:" }], "Text", CellChangeTimes->{3.72452666228046*^9}], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"paramPath", "=", RowBox[{"FileNameJoin", "[", RowBox[{"{", RowBox[{ RowBox[{"DirectoryName", "[", "jsonPath", "]"}], ",", "\"\\""}], "}"}], "]"}]}]], "Input", CellLabel->"In[133]:="], Cell[BoxData["\<\"/private/var/folders/pz/94mxs33x2l512z6wtjbthvy0000_ck/T/\ net.params\"\>"], "Output", CellChangeTimes->{3.713707095539467*^9, 3.714413964336213*^9, 3.7144149189635878`*^9, 3.732556367531705*^9, 3.754054785632416*^9, 3.7615716172521143`*^9, 3.763323210537854*^9, 3.763364829310114*^9, 3.763417685512705*^9, 3.763748127457136*^9, 3.763816631108942*^9, 3.763915737204124*^9, 3.764082281014338*^9}, CellLabel->"Out[133]="] }, Open ]], Cell["Get the size of the parameter file:", "Text"], Cell[CellGroupData[{ Cell[BoxData[ RowBox[{"FileByteCount", "[", "paramPath", "]"}]], "Input", CellLabel->"In[134]:="], Cell[BoxData["435648425"], "Output", CellChangeTimes->{3.713707095600053*^9, 3.714413964402884*^9, 3.714414919033608*^9, 3.7325563675626717`*^9, 3.7540547865155067`*^9, 3.761571617285421*^9, 3.7633232125830803`*^9, 3.763364830263426*^9, 3.763417687243053*^9, 3.763748127495017*^9, 3.763816631162385*^9, 3.763915737359476*^9, 3.764082281166368*^9}, CellLabel->"Out[134]="] }, Open ]] }, Open ]] }, Open ]] }, WindowSize->Automatic, WindowMargins->Automatic, WindowTitle->"BERT Trained on BookCorpus and Wikipedia Data-examples", Visible->True, OutputSizeLimit->10000, AutoStyleOptions->{"UndefinedSymbolStyle"->{FontColor -> GrayLevel[0]}}, FrontEndVersion->"12.2 for Mac OS X x86 (64-bit) (December 7, 2020)", StyleDefinitions->"Default.nb" ] (* End of Notebook Content *) (* Internal cache information *) (*CellTagsOutline CellTagsIndex->{} *) (*CellTagsIndex CellTagsIndex->{} *) (*NotebookFileOutline Notebook[{ Cell[CellGroupData[{ Cell[648, 23, 619, 18, 70, "ResourceExampleTitle",ExpressionUUID->"381985df-0ffb-452e-a045-76a319f690eb", CellGroupingRules->{"TitleGrouping", 0}, PageBreakBelow->False], Cell[1270, 43, 617, 17, 70, "ResourceExampleSubtitle",ExpressionUUID->"0548992b-0f02-4a03-938b-b5e97d2d2e44", CellGroupingRules->{"TitleGrouping", 10}, PageBreakBelow->False, CounterIncrements->"Subtitle"], Cell[CellGroupData[{ Cell[1912, 64, 106, 1, 70, "Subsection",ExpressionUUID->"2a592303-980a-4f38-82e7-aad6e899702b"], Cell[2021, 67, 207, 3, 70, "Text",ExpressionUUID->"cdb99d33-7702-4f87-a65e-c92e3520cd3c"], Cell[CellGroupData[{ Cell[2253, 74, 569, 9, 70, "Input",ExpressionUUID->"23384d66-f2b5-4d14-9b75-3718d777d34f"], Cell[2825, 85, 37896, 628, 70, "Output",ExpressionUUID->"2074aa56-c207-47af-8e35-f1422caab8db"] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell[40770, 719, 259, 3, 70, "Subsection",ExpressionUUID->"2656e2c0-6f41-461c-bd04-53504dc912c7"], Cell[41032, 724, 504, 8, 70, "Text",ExpressionUUID->"c342b042-c5e3-46a8-bfe2-88b4ed61b470"], Cell[CellGroupData[{ Cell[41561, 736, 305, 7, 70, "Input",ExpressionUUID->"be866de0-ce77-48b9-8564-3ab60d25f580"], Cell[41869, 745, 37203, 685, 70, "Output",ExpressionUUID->"ddec08a8-208f-498a-b685-884096468130"] }, Open ]], Cell[79087, 1433, 231, 3, 70, "Text",ExpressionUUID->"80a52bae-5988-4c19-85ff-ab3be735e901"], Cell[CellGroupData[{ Cell[79343, 1440, 727, 14, 70, "Input",ExpressionUUID->"60248f4b-ad5a-450e-bde7-4ac3cfce4c1c"], Cell[80073, 1456, 38717, 641, 70, "Output",ExpressionUUID->"2e14ed78-0935-4857-8d07-9ba5a3497e25"] }, Open ]], Cell[118805, 2100, 371, 5, 70, "Text",ExpressionUUID->"75916610-7547-463d-b2e7-b15b6b7a4794"], Cell[CellGroupData[{ Cell[119201, 2109, 1034, 18, 70, "Input",ExpressionUUID->"396f4cb9-fb88-403a-82b5-13d04b9d5884"], Cell[120238, 2129, 42011, 695, 70, "Output",ExpressionUUID->"93948365-6e92-4274-ae29-d94812d3b07c"] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell[162298, 2830, 148, 2, 70, "Subsection",ExpressionUUID->"7440f510-89a2-48c7-ae8b-acc1fffa5134"], Cell[162449, 2834, 760, 11, 70, "Text",ExpressionUUID->"6cb7d4a8-2d28-415d-bf39-d4a4f0cdc4ea"], Cell[163212, 2847, 873, 18, 70, "Input",ExpressionUUID->"2b12fbe1-e8ef-49e4-865d-1ec10bea525f"], Cell[164088, 2867, 118, 1, 70, "Text",ExpressionUUID->"a5d3dc2a-1ac4-451c-9af3-79d7a7518dfa"], Cell[CellGroupData[{ Cell[164231, 2872, 157, 3, 70, "Input",ExpressionUUID->"8419cee6-f30d-4f0e-881a-2c1c69e5c919"], Cell[164391, 2877, 512, 9, 70, "Output",ExpressionUUID->"67ff8f9d-f1a6-4e4d-b80e-f05a05493f1a"] }, Open ]], Cell[164918, 2889, 329, 4, 70, "Text",ExpressionUUID->"6e94df1e-f42f-4246-871d-8bab9bb57a66"], Cell[CellGroupData[{ Cell[165272, 2897, 257, 5, 70, "Input",ExpressionUUID->"50b2bf46-11fe-490a-b0af-0938c9fe8196"], Cell[165532, 2904, 29682, 491, 70, "Output",ExpressionUUID->"ed5840a7-65af-4433-8abe-26b87c0649f2"] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell[195263, 3401, 289, 4, 70, "Subsection",ExpressionUUID->"0e625efb-0756-477d-a62f-e59ebac10e4d"], Cell[195555, 3407, 833, 13, 70, "Text",ExpressionUUID->"bd5125a0-9231-4f6e-b946-07b89dd90b65"], Cell[CellGroupData[{ Cell[196413, 3424, 631, 14, 70, "Input",ExpressionUUID->"43559428-9491-4323-8d5c-1894f145a824"], Cell[197047, 3440, 8491, 181, 70, "Output",ExpressionUUID->"5711da47-45ad-41d4-b802-44c3e8dec0ca"] }, Open ]], Cell[205553, 3624, 798, 12, 70, "Text",ExpressionUUID->"6e72f9fc-fc89-4a00-913a-5b5c25197982"], Cell[CellGroupData[{ Cell[206376, 3640, 338, 8, 70, "Input",ExpressionUUID->"445fa5b4-c55f-4d4a-96a7-106c0db6f352"], Cell[206717, 3650, 1061, 27, 70, "Output",ExpressionUUID->"bacebefe-b7f7-44c0-b1d7-2b60ba0c5656"] }, Open ]], Cell[207793, 3680, 604, 10, 70, "Text",ExpressionUUID->"b88ec0cd-5d60-4f04-9524-8f44c0a1b844"], Cell[CellGroupData[{ Cell[208422, 3694, 455, 12, 70, "Input",ExpressionUUID->"2a5a8367-e764-49a5-804d-a280339020a6"], Cell[208880, 3708, 617, 11, 70, "Output",ExpressionUUID->"53a92fac-c167-428c-88dd-d1a057883a1e"] }, Open ]], Cell[209512, 3722, 283, 5, 70, "Text",ExpressionUUID->"befdaec9-7a49-4425-9753-36ec7b456db9"], Cell[CellGroupData[{ Cell[209820, 3731, 1509, 35, 70, "Input",ExpressionUUID->"29a5cae6-f9db-45d5-8abe-bbd5fa468bb4"], Cell[211332, 3768, 120549, 1981, 70, "Output",ExpressionUUID->"5214c41e-550c-4676-a78d-b2249d531808"] }, Open ]], Cell[331896, 5752, 366, 10, 70, "Text",ExpressionUUID->"a0dc4426-05f5-4c1c-ae47-b7efe4928ded"], Cell[CellGroupData[{ Cell[332287, 5766, 193, 4, 70, "Input",ExpressionUUID->"1afa90c8-1d4b-4357-88dd-0348627b5193"], Cell[332483, 5772, 50406, 832, 70, "Output",ExpressionUUID->"fbf7fa55-219a-43a3-b4a3-a67232cd7dee"] }, Open ]], Cell[382904, 6607, 293, 5, 70, "Text",ExpressionUUID->"c8f11202-4294-47f8-9e50-a577b2494ea6"], Cell[CellGroupData[{ Cell[383222, 6616, 243, 5, 70, "Input",ExpressionUUID->"8ee1cccb-10e8-438e-973a-b7622ec0f509"], Cell[383468, 6623, 27887, 463, 70, "Output",ExpressionUUID->"c2c1c3a5-4f85-4579-a1da-636ab1c2eaf3"] }, Open ]], Cell[411370, 7089, 307, 6, 70, "Text",ExpressionUUID->"e60ee030-3278-4d14-851e-139f069bd21e"], Cell[CellGroupData[{ Cell[411702, 7099, 354, 9, 70, "Input",ExpressionUUID->"38c16b09-ab79-4063-a31c-4e4145397f0b"], Cell[412059, 7110, 81333, 1339, 70, "Output",ExpressionUUID->"e19e2b57-8404-47f6-ac49-7d950a426d71"] }, Open ]], Cell[493407, 8452, 430, 11, 70, "Text",ExpressionUUID->"5d3020a7-d5c5-4fbe-b1fe-f5e5cbc60839"], Cell[CellGroupData[{ Cell[493862, 8467, 351, 9, 70, "Input",ExpressionUUID->"8a87226c-0de7-4214-aba0-98ef4d6b8da8"], Cell[494216, 8478, 26993, 448, 70, "Output",ExpressionUUID->"249452b6-1a89-4640-a022-8c4c21a123ea"] }, Open ]], Cell[521224, 8929, 523, 9, 70, "Text",ExpressionUUID->"388625ef-fd72-4f07-9d44-c4ee03475f0e"], Cell[521750, 8940, 246464, 4047, 70, "PastedInput",ExpressionUUID->"2cf3665a-b8e1-4ae1-a36d-1a04556acd72"] }, Open ]], Cell[CellGroupData[{ Cell[768251, 12992, 182, 2, 70, "Subsection",ExpressionUUID->"b82efd10-dfc9-4287-a373-a43d6bbd6a5f"], Cell[768436, 12996, 311, 6, 70, "Text",ExpressionUUID->"6a68ab83-3d78-4924-a514-b6ecaec705b2"], Cell[CellGroupData[{ Cell[768772, 13006, 577, 13, 70, "Input",ExpressionUUID->"cd155b1a-c892-43f0-a1da-63bfc57c8951"], Cell[769352, 13021, 363434, 5834, 70, "Output",ExpressionUUID->"5a474b0e-6559-4f39-8d11-80e2a0e42037"] }, Open ]], Cell[1132801, 18858, 232, 4, 70, "Text",ExpressionUUID->"f4184f33-fe43-4385-8124-56e3f384b319"], Cell[1133036, 18864, 999, 19, 70, "Input",ExpressionUUID->"b139a2f1-4085-42c4-8594-5d45f40655d2"], Cell[1134038, 18885, 452, 6, 70, "Text",ExpressionUUID->"f959030d-495a-425b-ba5b-ee1c8729f998"], Cell[1134493, 18893, 525, 11, 70, "Input",ExpressionUUID->"203d3ebb-f98d-4e30-af1a-6ce13f76449e"], Cell[1135021, 18906, 424, 8, 70, "Text",ExpressionUUID->"0b49adf4-0b38-40b0-8fcb-40ad992aa2a1"], Cell[CellGroupData[{ Cell[1135470, 18918, 239, 5, 70, "Input",ExpressionUUID->"96877020-6c8e-4315-8889-67c55ada490c"], Cell[1135712, 18925, 31815, 528, 70, "Output",ExpressionUUID->"c4977bb2-ccc2-4219-82ea-04d209fca235"] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell[1167576, 19459, 293, 3, 70, "Subsection",ExpressionUUID->"576d60bd-cb7b-40e4-893a-e2e14c1ec015"], Cell[1167872, 19464, 262, 3, 70, "Text",ExpressionUUID->"e5a95929-5a78-4afb-8bc3-276514eb8b75"], Cell[1168137, 19469, 1392, 27, 70, "Input",ExpressionUUID->"d8606ab5-15ab-428e-ba1d-30c9f4ac7ae4"], Cell[1169532, 19498, 120, 1, 70, "Text",ExpressionUUID->"b144c051-0382-4b66-bd37-f1d26c5a13f0"], Cell[CellGroupData[{ Cell[1169677, 19503, 205, 5, 70, "Input",ExpressionUUID->"ff597362-8191-4b28-bea7-06d6b5da606e", CellID->1153613343], Cell[1169885, 19510, 712, 12, 70, "Output",ExpressionUUID->"76864bc3-b0e3-47a8-b400-7a9421201fe5"] }, Open ]], Cell[1170612, 19525, 794, 12, 70, "Text",ExpressionUUID->"7f6659db-e392-47fc-b03b-c8a8218e3c3f"], Cell[1171409, 19539, 1837, 44, 70, "Input",ExpressionUUID->"b599195c-dfb2-4389-8e71-99e44f133627"], Cell[1173249, 19585, 358, 7, 70, "Text",ExpressionUUID->"904a3515-e73f-4423-9ee9-1577a841e329"], Cell[CellGroupData[{ Cell[1173632, 19596, 1008, 23, 70, "Input",ExpressionUUID->"d2653fd0-bceb-4201-9462-8905344a94b4"], Cell[1174643, 19621, 37493, 621, 70, "Output",ExpressionUUID->"f1def8ea-03a8-43b2-b33d-2be5dd0ec302"] }, Open ]], Cell[1212151, 20245, 335, 5, 70, "Text",ExpressionUUID->"721b23c1-cc53-459c-96df-446da7200322"], Cell[CellGroupData[{ Cell[1212511, 20254, 1030, 19, 70, "Input",ExpressionUUID->"96eeb01c-c319-469e-907b-c95fadf24f79"], Cell[1213544, 20275, 30206, 622, 70, "Output",ExpressionUUID->"a7de9bbd-8dfc-4346-950b-de3f452154bb"] }, Open ]], Cell[1243765, 20900, 143, 1, 70, "Text",ExpressionUUID->"ae3018cb-69fc-4886-a6f3-e67984c05989"], Cell[CellGroupData[{ Cell[1243933, 20905, 225, 5, 70, "Input",ExpressionUUID->"2f18f524-8ff3-494a-89f2-f5356f9e94ae"], Cell[1244161, 20912, 211, 3, 70, "Output",ExpressionUUID->"93a23c4e-9a40-41e3-b898-5b95b07190cc"] }, Open ]], Cell[1244387, 20918, 861, 13, 70, "Text",ExpressionUUID->"161b19f4-3dbd-474f-b1b7-fe9e2ad00f56"], Cell[1245251, 20933, 1376, 40, 70, "Input",ExpressionUUID->"782ee35f-165c-429f-bec7-7b2d177192a2"], Cell[1246630, 20975, 361, 5, 70, "Text",ExpressionUUID->"8958ad96-5a49-448c-8c05-d4c88de46a69"], Cell[CellGroupData[{ Cell[1247016, 20984, 1171, 23, 70, "Input",ExpressionUUID->"cef4ddab-013e-46b5-ac67-4350581a0ebe"], Cell[1248190, 21009, 30447, 626, 70, "Output",ExpressionUUID->"ef9618f0-5423-4456-a24a-81928e1d65e3"] }, Open ]], Cell[1278652, 21638, 185, 2, 70, "Text",ExpressionUUID->"6f25a21d-3bd5-4e10-b91c-ad739739c5ad"], Cell[CellGroupData[{ Cell[1278862, 21644, 536, 13, 70, "Input",ExpressionUUID->"d51dabeb-3c5e-4463-96e7-44eb7bfefd1b"], Cell[1279401, 21659, 17885, 346, 70, "Output",ExpressionUUID->"e3deae3c-f5fb-4152-9f96-a7599a8a668e"] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell[1297335, 22011, 103, 1, 70, "Subsection",ExpressionUUID->"e7c7bece-5d65-43c5-8c52-c64c222a38e5"], Cell[1297441, 22014, 242, 3, 70, "Text",ExpressionUUID->"04c5071b-7cd5-47f7-a601-d09a5c3bbecf"], Cell[CellGroupData[{ Cell[1297708, 22021, 348, 7, 70, "Input",ExpressionUUID->"ebd795d7-013f-4513-b3ea-81727d4352c3"], Cell[1298059, 22030, 10694, 272, 70, "Output",ExpressionUUID->"6523e66e-df9f-4553-9e4a-208d8d1913a5"] }, Open ]], Cell[1308768, 22305, 120, 1, 70, "Text",ExpressionUUID->"e1de4363-b1b6-4316-90fb-4fe4921ff087"], Cell[CellGroupData[{ Cell[1308913, 22310, 352, 7, 70, "Input",ExpressionUUID->"4744c0b0-2c33-4ba4-83c2-7668ab0dcf10"], Cell[1309268, 22319, 334, 5, 70, "Output",ExpressionUUID->"c54c643a-c5ae-4736-937b-8296cd345d78"] }, Open ]], Cell[1309617, 22327, 137, 2, 70, "Text",ExpressionUUID->"bf6a4414-9d48-407d-85f3-b66684fab445"], Cell[CellGroupData[{ Cell[1309779, 22333, 343, 7, 70, "Input",ExpressionUUID->"ab200015-3691-4299-ae54-9de3636ed856"], Cell[1310125, 22342, 941, 18, 70, "Output",ExpressionUUID->"b2703865-683e-4aa6-81cb-6904fbc05f1b"] }, Open ]], Cell[1311081, 22363, 110, 1, 70, "Text",ExpressionUUID->"09c8cd84-9987-40e6-a36c-1877299869ac"], Cell[CellGroupData[{ Cell[1311216, 22368, 340, 7, 70, "Input",ExpressionUUID->"70fdd03b-92c6-4e13-a08e-67b591ff39d4"], Cell[1311559, 22377, 9603, 175, 70, "Output",ExpressionUUID->"b9b8ec13-86a0-4982-873b-38ca24068ef6"] }, Open ]] }, Open ]], Cell[CellGroupData[{ Cell[1321211, 22558, 37, 0, 70, "Subsection",ExpressionUUID->"eb30117e-8f35-4a37-86e7-28e1c855ecc6"], Cell[1321251, 22560, 232, 7, 70, "Text",ExpressionUUID->"4c91d268-c2cb-4dcc-9fd3-a9de7953c89d"], Cell[CellGroupData[{ Cell[1321508, 22571, 632, 15, 70, "Input",ExpressionUUID->"b8fe342c-5902-485f-a521-24464da5e161"], Cell[1322143, 22588, 447, 7, 70, "Output",ExpressionUUID->"013c11a7-5b91-450b-ac3a-b1a579c48baf"] }, Open ]], Cell[1322605, 22598, 277, 10, 70, "Text",ExpressionUUID->"cbb1eb2e-2153-49e1-802e-b2955dc9de41"], Cell[CellGroupData[{ Cell[1322907, 22612, 243, 7, 70, "Input",ExpressionUUID->"bda44c2a-8d27-4a47-a6f4-b056a72c6dde"], Cell[1323153, 22621, 451, 7, 70, "Output",ExpressionUUID->"ab18c737-51e4-46fe-9103-77ec9afd3382"] }, Open ]], Cell[1323619, 22631, 51, 0, 70, "Text",ExpressionUUID->"48f3e418-8da4-4840-a2ab-95ce65fa850e"], Cell[CellGroupData[{ Cell[1323695, 22635, 99, 2, 70, "Input",ExpressionUUID->"487ae75a-d2ab-4c80-96e7-bfd5b6cd448b"], Cell[1323797, 22639, 385, 6, 70, "Output",ExpressionUUID->"428a2d33-41fe-40d2-8d5c-56a0c503f797"] }, Open ]] }, Open ]] }, Open ]] } ] *) (* End of internal cache information *)