Lily Ge
Lily Ge
# Pull Request Checklist Adding to reproduction logs doesn't quite fit into any of the options currently provided by the checklist in PULL_REQUEST_TEMPLATE.md but is a common PR, so I...
# Pull Request Checklist UW Basilisk Results: Model | dl | Metric -- | -- | -- RankZephyr | dl20 | 0.8198 E2E First Mistral | dl20 | 0.7885 ##...
Addresses #2856 with the problems mentioned.
Pyserini's prebuilt index definitions contain more metadata than Anserini, specifically the following fields: size compressed (bytes), total terms, documents, unique terms, downloaded. Anserini's prebuilt indexes don't have these fields. Given...
After looking at #2772 , Lucene's implementation for RRF seems to be faster than the current Anserini implementation. Modifying Lucene's implementation to work with ScoredDocs and Anserini's current fusion design,...
Most of Anserini's prebuilt indexes have been uploaded to Hugging Face per this [issue](https://github.com/castorini/anserini/issues/2820). We should update Pyserini to point to the indexes on Hugging Face and upload any additional...
# Pull Request Checklist ## Reference Issue Please provide the reference to issue this PR is addressing (# followed by the issue number). If there is no associated issue, write...
Lucene inverted BM25 index for ACL Anthology. To make sure all fields are indexed without getting flagged as 'empty', empty abstracts and authors have 'n/a' as placeholders. Fields are title,...