distilbert-dot-tas_b-b256-msmarco EU AI Act Compliance Profile

DistilBert for Dense Passage Retrieval trained with Balanced Topic Aware Sampling (TAS-B)

We provide a retrieval trained DistilBert-based model (we call the dual-encoder then dot-product scoring architecture BERT_Dot) trained with Balanced Topic Aware Sampling on MSMARCO-Passage.

This instance was trained with a batch size of 256 and can be used to re-rank a candidate set or directly for a vector index based dense retrieval. The architecture is a 6-layer DistilBERT, without architecture additions or modifications (we only change the weights during training) - to receive a query/passage representation we pool the CLS vector. We use the same BERT layers for both query and passage encoding (yields better results, and lowers memory requirements).

If you want to know more about our efficient (can be done on a single consumer GPU in 48 hours) batch composition procedure and dual supervision for dense retrieval training, check out our paper: https://arxiv.org/abs/2104.06967 🎉

For more information and a minimal usage example please visit: https://github.com/sebastian-hofstaetter/tas-balanced-dense-retrieval

Effectiveness on MSMARCO Passage & TREC-DL'19

We trained our model on the MSMARCO standard ("small"-400K query) training triples re-sampled with our TAS-B method. As teacher models we used the BERT_CAT pairwise scores as well as the ColBERT model for in-batch-negative signals published here: https://github.com/sebastian-hofstaetter/neural-ranking-kd

MSMARCO-DEV (7K)

| | MRR@10 | NDCG@10 | Recall@1K | |----------------------------------|--------|---------|-----------------------------| | BM25 | .194 | .241 | .857 | | TAS-B BERT_Dot (Retrieval) | .347 | .410 | .978 |

TREC-DL'19

For MRR and Recall we use the recommended binarization point of the graded relevance of 2. This might skew the results when

Your obligations as a distilbert-dot-tas_b-b256-msmarco deployer

Track your progress

Sign up to track obligations, receive deadline reminders, and generate audit packages.

Tags

Languages

Library