i implemented my personal local recommendation algorithm based on #ytss scripts and my youtube likes data.
i fetched likes from my youtube account using this script:
https://ipfs.chainsafe.io/ipfs/QmTdJcMEaqN6RFDmZqvvyxPgNv6RnuU2HfvgUSMo43tEmi#eaed6c0dcd9aa9a08641404c812cecac54bd0428e01871230ff01391ece06928/inline/quiet/markdown
then i queried against 12 M youtube videos:
python3 ytss/gui/semantic_query.py ytss/data_12M/ytcorpus12M_merge.txt ytss/data_12M/embeddings_int8_256_12M_merge.npy 256 "`cat ytlikes.txt`"
nostr:note1qqqq999ez59srenftes8dhxf7wyf5a7rahakk3un0vrvs8hh0ems0tdsg5