Pyserini Release Notes (v0.12.0)

May 5, 2021 ยท View on GitHub

Release date: May 5, 2021

  • Added dense retrieval support for DistilBERT KD, SBERT, and ANCE.
  • Added support for KILT, including integration tests.
  • Added support for pre-tokenized collections and tokenization using sentencepiece and other models.
  • Added support and guide to reproduce Elasticsearch multi-field experiments
  • Added integration tests to reproduce text classification pseudo-relevance feedback experiments.
  • Added option to make cache directory configurable.
  • Improved pyserini.fusion.
  • Improved support on Windows.
  • Improved dense retrieval integration test cases.
  • Organized integration tests into sparse, dense, and clprf.
  • Refactored dense query encoders.
  • Fixed bugs related to iteration order of topics.
  • Cleaned up LTR related code.
  • Changed terminology in documentation "replicate" to "reproduce" per latest ACM policy

Contributors (This Release)

Sorted by number of commits:

All Contributors

Sorted by number of commits, according to GitHub: