Ir Datasets Versions Save

Provides a common interface to many IR ranking datasets.

v0.4.1

2 years ago
  • Adds version 2 of the MS MARCO document collection.
  • Using mirror.ir-datsets.com as a fallback for some small files
  • More examples in the documentation (the python API is now joined by the CLI and a PyTerrier example)
  • Improved bibtex, including a master bib file that can be imported papers (e.g., in overleaf).
  • Other minor improvements

v0.4.0

3 years ago

New datasets:

  • BEIR suite
  • Cranfield
  • CLIRMatrix
  • DPR-W100
  • NQ
  • TREC DL Hard
  • TREC News
  • TripClick

Other:

  • Download dashboard
  • Improved documentation for non-downloadable datasets
  • A beta "more pythonic API"
  • Speeding up library load time
  • Minor bug fixes, improvements, etc.

v0.3.3

3 years ago

dataset migration bugfix

v0.3.2

3 years ago

v0.3.1

3 years ago

v0.3.0

3 years ago

v0.2.0

3 years ago

Now includes language codes for queries and docs

v0.1.7

3 years ago