Pyserini Release Notes (v0.25.0)

March 31, 2024 ยท View on GitHub

  • Release date: March 31, 2024
  • Anserini dependency: v0.25.0
  • Lucene dependency: v9.9.1

Summary of Changes

  • Added cohere-embed-english-v3.0 2CRs for MS MARCO v1 passage.
  • Added BGE-base-en-v1.5 2CRs for MS MARCO v1 passage and BEIR.
  • Added support for DL22 doc and DL23 doc and passages from MS MARCO v2 and added corresponding 2CRs.
  • Added initial support for CLIP dense encoder and multimodal retrieval.
  • Added option for users to specify different distance metrics when building Faiss indexes.
  • Refactored and recalibrated 2CR scores, increased tolerance as needed.
  • Refactored method to get topics.
  • Replaced deprecated pkg_resources with importlib.resources and other deprecation fixes.
  • Updated CIRAL 2CRs.
  • Updated SPLADE 2CRs for BEIR.

Contributors

This Release

Sorted by number of commits:

All Time

All contributors with five or more commits, sorted by number of commits, according to GitHub: