Datasets:
Tasks:
Text Retrieval
Formats:
parquet
Sub-tasks:
document-retrieval
Languages:
English
Size:
1M - 10M
License:
Commit History
-
Jev false-negative judgments (judgments config)
Verified signature{ "judgments": { "rows_in_run": 530674, "rows_dropped": 0, "queries_removed": 0, "queries_trimmed": 0, "rows": 530674, "queries": 39999, "positives": 39999, "candidates_per_query": 12.3, "top_up_rows": 300720, "judge": "typesafe/jev-1.13.0" } }16c14ef
-
Card: documents cut to 1,024 reranker tokens when scored
Verified signaturec9f6762 -
s2orc: hard negatives and teacher scores (jina35-u2)
Verified signaturea05fb24
-
s2orc: re-mined candidates (100 to depth 1000); teacher scores to come
Verified signature96ffc5a -
S2ORC title–abstract: domain-v1.1: near copies of evaluation documents that no evaluation query judges relevant restored (middle-ground decontamination)
Verified signature680c9de -
S2ORC title–abstract: the unified schema (domain-v1, miner scores)
Verified signaturea8b871b -
initial commit
Verified signaturea50335c