prasannad28's picture
Update README.md
f73c10a verified
metadata
license: mit

These are the (anchor,positive,negative) triplets which were used to finetune the embedding model. 20 negatives were mined per anchor-positive pairs