Schrier - san,
Thanks for your informative post.
Tham also visualizes clusters of chunk embedding in his post. I visualized this in Wolfram Language as it is also useful for understanding of the summarization.
In[]:=
clusterskeys=FindClusters[​​embeddings->Range[Length[embeddings]],​​DistanceFunction->CosineDistance]
Out[]=
{{1,2},{3,42,43,44,46,50,51,52,53,54,55},{4,6,40,41,45},{5,7,8,9,10,11,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32},{12,37,38,39},{33,34},{35,36},{47,48,49}}
In[]:=
col=ColorData[97]/@Range[Length[clusterskeys]];​​assoc=KeySort@Flatten@MapIndexed[Thread[#->col[[First@#2]]]&,clusterskeys];​​ArrayPlot{Values@assoc},

Out[]=