README.md

metadata

language:
  - en
tags:
  - information-retrieval
license: cc-by-4.0
configs:
  - config_name: brief-corpus
    data_files:
      - split: test
        path: brief-corpus/*.parquet
  - config_name: brief-queries
    data_files:
      - split: test
        path: brief-queries/*.parquet
  - config_name: brief-qrels
    data_files:
      - split: test
        path: brief-qrels/*.parquet
  - config_name: full-corpus
    data_files:
      - split: test
        path: full-corpus/*.parquet
  - config_name: full-queries
    data_files:
      - split: test
        path: full-queries/*.parquet
  - config_name: full-qrels
    data_files:
      - split: test
        path: full-qrels/*.parquet

SRA-Bench

MTEB v2 text retrieval dataset layout for SRA-Bench. The brief subset uses the compact corpus, and the full subset uses the expanded corpus-full source text.