sure, i have 20M lib of congress amazon.titan-embed-text-v2:0 embeddings, 5.5M have DDC, so I did:
1. Extracted centroids for each ddc ### class (labels from wikipedia)
2. Plotted PCA / UMAP
Web viewer: thisismattmiller.github.io/dcc-embedding/
Code: github.com/thisismattmi...