Datasets:
Tasks:
Other
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
machine-learning
active-learning
representation-learning
sparse-fourier-transform
experimental-design
synthetic-data
| { | |
| "status": "passed", | |
| "scope": "readonly derivedtable audit, no oracle queries", | |
| "table_counts": { | |
| "strong_evaluation": 192, | |
| "probe_evaluation": 120, | |
| "warmup_usages": 41, | |
| "core_evaluation": 324, | |
| "contexts": 74, | |
| "unified_evaluation": 32 | |
| }, | |
| "source_files_byte_equal_original": 9, | |
| "all_source_pointers_models_risks_call_partitions_valid": true, | |
| "parquet_equals_jsonl": true, | |
| "bigint_masks_decimal_strings": true, | |
| "maximum_mask_bits": 255, | |
| "teacher_case_ids_consistent": 169, | |
| "warmup_context_charges_exact": true, | |
| "certified": 661, | |
| "refused": 7, | |
| "false_certificates": 0, | |
| "notes": [ | |
| "Repeated case IDs denote shared latent teachers, not independent oracle traces.", | |
| "Warmup usage records can be charged to multiple methods or suites; contexts prevent per-row double-counting.", | |
| "Search_and_screen_calls includes generic searches, probes, and fallbacks after subtracting readout and fresh audit calls." | |
| ] | |
| } | |