|
Download README.md from GMLRVigil/BenchCheck-MetaBenchmark: direct link, hf CLI and curl.
- Browser
- Download file 4.33 kB
-
https://huggingface.co/datasets/GMLRVigil/BenchCheck-MetaBenchmark/resolve/main/README.md
- Command line
-
hf download hf://datasets/GMLRVigil/BenchCheck-MetaBenchmark/README.md
-
curl -L -o README.md https://huggingface.co/datasets/GMLRVigil/BenchCheck-MetaBenchmark/resolve/main/README.md
4.33 kB
| license: other | |
| license_name: mixed-source-licenses | |
| license_link: https://huggingface.co/datasets/GMLRVigil/BenchCheck-MetaBenchmark#licences | |
| task_categories: | |
| - video-classification | |
| - visual-question-answering | |
| language: | |
| - en | |
| pretty_name: BenchCheck meta-benchmark (v6) | |
| size_categories: | |
| - n<1K | |
| configs: | |
| - config_name: meta_benchmark | |
| data_files: items/meta_benchmark.jsonl | |
| default: true | |
| - config_name: coverage_census | |
| data_files: items/coverage_census.jsonl | |
| # BenchCheck meta-benchmark (v6) | |
| 840 multiple-choice video questions from 76 public video benchmarks, one item per video, | |
| four capability groups x 210 (Perception, Temporal, Spatial / physical, Reasoning / knowledge). The set is the | |
| difficulty-first census of the BenchCheck screened pool: items that none of the cheap attackers of the pool screening | |
| (text-only, single frame, 32-frame 2B model, options-only, shuffled frames) could solve, ranked by the worst-case attacker | |
| percentile, with the marked answers verified against the frames; no cap on any model's accuracy (v6, 2026-09-19: the earlier | |
| v5 composition capped the reference model's share of correct answers at 45 %, the cap is dropped so that every row of the leaderboard is a measurement). | |
| ## Construction (five steps; details in the paper appendix) | |
| 1. Candidates: the screened pool; the 3,000 items with the smallest worst-case attacker percentile (one per video). | |
| 2. Capability labels: Claude Opus 5 assigns one of 18 fine categories from the question, options and gold answer; the | |
| categories map to the four groups. | |
| 3. Verification of the marked answer: Claude Opus 5 reads the frames (32, then 512 for unconfirmed items), the question, | |
| the options and the marked answer and returns supported / contradicted / not verifiable / definition dependent. Kept: | |
| supported; reversed clips (AoTBench RTime, TempCompass direction); not verifiable on videos >= 60 s. Dropped: | |
| contradicted; definition dependent; not verifiable on videos < 60 s. Ordering questions whose options are the events | |
| themselves are excluded (not scorable as single choice). | |
| 3b. Selection: per group the verified items in worst-case attacker percentile order (hardest first) until 210, one item per | |
| video, no reference cap. The reference model (Claude Opus 5, VIDEO protocol) answers 56.8 % of the resulting items; this is reported, not constrained. | |
| 4. Protocols: VIDEO (1 fps, <= 512 frames, short side 224) and BLIND (question + options, 4 permutations); see code/prompt.md. | |
| Version v6: final composition. | |
| ## Files | |
| - `items/coverage_census.jsonl`: the coverage census used as the appendix comparison (832 items, at most 8 per benchmark from | |
| 106 benchmarks, one per video where possible; same fields without the verification). | |
| - `items/meta_benchmark.jsonl`: one line per item (fields: item_id, benchmark, capability_group, fine_category, question, | |
| options, answer, answer_idx, chance, video, video_id, duration_s, attack_pct, verification, source_version, license). | |
| - `items/summary.json`: counts per group, benchmark and verification reason. | |
| - `videos/<video_id>.mp4`: normalized videos (840 files; H.264, at most 1,024 stored frames per video). | |
| - `code/prompt.md`: the exact prompts, the frame sampling rule (1 fps over the timeline, at most 512 frames uniformly thinned, | |
| short side 224; the stored videos hold at most 1,024 frames) and the scoring rule. | |
| - `results/leaderboard.csv`: current leaderboard (Overall = VIDEO, Blind, Gain, per-group accuracy; n = items scored). | |
| ## Composition | |
| Groups: {'perception': 210, 'temporal': 210, 'spatial_physical': 210, 'reasoning_knowledge': 210}. Largest sources: {'S-EMBER': 109, 'AoTBench': 92, 'ALLVB': 45, 'TOC-Bench': 38, 'UrbanVideo-Bench': 34, 'Camera Movement Understanding (BU_Google)': 33, 'VideoAds': 31, 'Video-MME-v2': 28}. | |
| Verification reasons: {'reversed_clip': 88, 'supported': 533, 'not_verifiable_long': 219}. Median duration 42 s; 282 videos >= 300 s; | |
| mean chance 29.7%. | |
| ## Licences | |
| Every item keeps the licence of its source benchmark (field `license`; {'unknown': 185, 'CC-BY-4.0': 57, '': 42, 'custom': 72, 'CC-BY-NC-SA-4.0': 137, 'MIT': 99, 'CC-BY-NC-4.0': 156, 'Apache-2.0': 84, 'CC-BY-SA-4.0': 7, 'CC-BY-NC-ND-4.0': 1}). Items are redistributed for | |
| research evaluation only; check the source benchmark's terms before any other use. | |