README.md

metadata

license: cc-by-nc-sa-4.0
language: en
tags:
  - splade
  - query-expansion
  - document-expansion
  - bag-of-words
  - passage-retrieval
  - knowledge-distillation
  - document encoder
datasets:
  - ms_marco

Efficient SPLADE

Efficient SPLADE model for passage retrieval. This architecture uses two distinct models for query and document inference. This is the query one, please also download the doc one (https://huggingface.co/naver/efficient-splade-V-large-doc). For additional details, please visit:

• paper:
• code: https://github.com/naver/splade

  	MRR@10 (MS MARCO dev)	R@1000 (MS MARCO dev)	Latency (PISA) ms	Latency (Inference) ms
  naver/efficient-splade-V-large	38.8	98.0	29.0	45.3
  naver/efficient-splade-VI-BT-large	38.0	97.8	31.1	0.7

Citation

If you use our checkpoint, please cite our work (need to update):

@misc{https://doi.org/10.48550/arxiv.2205.04733,
  doi = {10.48550/ARXIV.2205.04733},
  url = {https://arxiv.org/abs/2205.04733},
  author = {Formal, Thibault and Lassance, Carlos and Piwowarski, Benjamin and Clinchant, Stéphane},
  keywords = {Information Retrieval (cs.IR), Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective},
  publisher = {arXiv},
  year = {2022},
  copyright = {Creative Commons Attribution Non Commercial Share Alike 4.0 International}
}