-
Fixed api embeddings for Book-Crossing
-
Delete titles_only/Book-Crossing/Book-Crossing.text-embedding-3-large.npy
-
Delete titles_only/Book-Crossing/Book-Crossing.gemini-embedding-001.npy
-
Add book crossing embs
-
Delete prod_search
-
Merge branch 'main' of https://huggingface.co/datasets/McAuley-Lab/BLaIR-Bench-API
-
Add product search embs
-
delete reddit_movie in titles_only folder
-
Merge branch 'main' of https://huggingface.co/datasets/McAuley-Lab/BLaIR-Bench-API
-
Created a ps folder
-
Upload esci_titles_only_na2period.jsonl
-
Upload C4_titles_only_na2period.jsonl
-
Delete prod_search_dls/esci/placeholder.txt
-
Upload esci item metadata containing only item titles
-
Merge branch 'main' of https://huggingface.co/datasets/McAuley-Lab/BLaIR-Bench-API
-
Added esci folder with placefolder file
-
Add a few new embs
-
Delete place holder
-
Upload Amazon-C4 item dataset with metadata attribute containing item titles only.
-
Created new c4 item dataset with placeholder
-
Merge branch 'main' of https://huggingface.co/datasets/McAuley-Lab/BLaIR-Bench-API
-
Baby_Products encoded by gemini with d256
-
Merge branch 'main' of https://huggingface.co/datasets/McAuley-Lab/BLaIR-Bench-API
-
add more embeddings
-
seq_rec ds encoded by gemini with MRL d256
-
seq_rec ds encoded by openai with MRL d256
-
Use correct query and item reddit_movie encoded by openai
-
Processed with correct id2name
-
Used imdb_processed instead of id2name to preprocess. Wrong approach
-
Added ps downloads folder and bc gemini file
-
Merge branch 'main' of https://huggingface.co/datasets/McAuley-Lab/BLaIR-Bench-API
-
Book-Crossing encoded with gemini
-
Delete placeholder file in this folder
-
Upload reddit_movie's id2title and id2query
-
create reddit_movie folder
-
All seq_rec datasets encoded by gemini-emb-001
-
Add embs for seq rec
-
text-embedding-3-large encoded
-
Encode seq_rec tasks with gemini-emb-001
-
Add Qwen3 embeddings for seq_rec
-
Yelp seq_rec text-emb-3-large
-
Baby Products seq_rec encoded by text-emb-3-large
-
Organize them under seq_rec task
-
ML1M seq_rec encoded with text-emb-3-large
-
Video_Games encoded with text-emb-3-large
-
text-emb-3-large All_Beauty emb
-
Create titles_only/.gitkeep
Store API embeddings of titles as features. -
Delete title_only
-
Create title_only
Store API embeddings of titles as features. -
initial commit