Datasets:
Download validation_report.json from RegalFire/Agent-Failure-Recovery-Benchmark: direct link, hf CLI and curl.
- Browser
- Download file 705 Bytes
-
https://huggingface.co/datasets/RegalFire/Agent-Failure-Recovery-Benchmark/resolve/main/validation_report.json
- Command line
-
hf download hf://datasets/RegalFire/Agent-Failure-Recovery-Benchmark/validation_report.json
-
curl -L -o validation_report.json https://huggingface.co/datasets/RegalFire/Agent-Failure-Recovery-Benchmark/resolve/main/validation_report.json
705 Bytes
| { | |
| "status": "PASS", | |
| "records": 22573, | |
| "independently_checked": 22573, | |
| "unique_ids": 22573, | |
| "semantic_duplicates": 0, | |
| "group_count": 23, | |
| "splits": { | |
| "holdout": 6058, | |
| "train": 15150, | |
| "validation": 1200, | |
| "test": 165 | |
| }, | |
| "errors": [], | |
| "validation_scope": "Schema, strict UTF-8/JSON, identities, exact semantic duplicates, source hashes/relationships, attribution, labels, independent computational oracles, recovery replay and grouped split leakage", | |
| "near_duplicate_policy": "Scientific normalized text and question 5-shingle grouping; synthetic whole-template groups plus source symmetry-grid grouping. No claim of exhaustive semantic deduplication." | |
| } | |