This dataset is the public result layer for
GioiaZheng/msmarco-genqa. It collects
the exact ranked outputs behind three benchmark studies:
TREC Deep Learning 2019 and 2020 passage retrieval;
cross-domain retrieval on NFCorpus and SciFact;
the NFCorpus video-query representation ablation.
The 14 runs are available both as the original TREC-format TSV files and as one
normalized Parquet table for the Dataset Viewer. The three checksummed GitHub… See the full description on the dataset page:
https://huggingface.co/datasets/GioiaZheng/msmarco-genqa-benchmark-runs.
