Skip to content

Latest commit

 

History

History
63 lines (55 loc) · 3.25 KB

release-notes-v0.16.1.md

File metadata and controls

63 lines (55 loc) · 3.25 KB
Raw

Pyserini Release Notes (v0.16.1)

Release date: May 12, 2022

  • Installed pre-built indexes:
    • BEIR (v1.0.0): "flat" baseline, "multfield" baseline, and SPLADE-distill CoCodenser-medium.
    • MS MARCO V1/V2 segmented doc condition (updated indexes).
    • MS MARCO V1 doc/passage uniCOIL noexp.
  • Created initial repro-matrix for MS MARCO V1/V2 doc/passage.
  • Improved indexing util for encoding a corpus.
  • Added util to dump out BM25 document vectors from a corpus.
  • Added bindings to new Anserini feature: multi-threaded method to fetch raw documents from index in batch.
  • Added option in trec_eval to compute metrics with unjudged docs removed and to computed judged@k.
  • Added IRST retrieval model, with integration tests.
  • Added SimpleGeoSearcher.
  • Refactored LTR pipeline, improved documentation.
  • Refactored IndexReader, moved from pyserini.index to pyserini.index.lucene.
  • Refactored Vector-PRF implementation and added negative PRF passages for Rocchio.
  • Refactored code for loading topics and qrels.
  • Updated documentation for dense encoding.

Contributors (This Release)

Sorted by number of commits:

All Contributors

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