Task Progression Failure Detection on Close Box Combined
100TPRGPT-4o Image QA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4o Image QAFailure Detector Category=VLM2026.03 | 100 | 0 | 60 | |
| Gemini 1.5 Pro Video QAFailure Detector Category=VLM2026.03 | 98 | 57 | 82 | |
| Gemini 1.5 Pro Image QAFailure Detector Category=VLM2026.03 | 96 | 0 | 57 | |
| SentinelComponents=STAC MMD* + GPT-4o Video QA*2026.03 | 96 | 86 | 92 | |
| GPT-4o Video QA*Failure Detector Category=VLM2026.03 | 88 | 89 | 89 | |
| Claude 3.5 Sonnet Image QAFailure Detector Category=VLM2026.03 | 81 | 6 | 51 | |
| Claude 3.5 Sonnet Video QAFailure Detector Category=VLM2026.03 | 81 | 31 | 61 | |
| Temporal Non-Distr. Min.Failure Detector Category=Diff.2026.03 | 71 | 97 | 82 | |
| STAC For. KLFailure Detector Category=STAC2026.03 | 65 | 97 | 78 | |
| STAC Rev. KLFailure Detector Category=STAC2026.03 | 65 | 97 | 78 | |
| STAC MMD*Failure Detector Category=STAC2026.03 | 65 | 97 | 78 | |
| Diffusion Output VarianceFailure Detector Category=Diff.2026.03 | 25 | 100 | 55 |