Task Progression Failure Detection on Close Box In-Distribution
100TPRTemporal Non-Distr. Min.
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Temporal Non-Distr. Min.Failure Detector Category=Diff.2026.03 | 100 | 97 | 4.67 | |
| STAC For. KLFailure Detector Category=STAC2026.03 | 100 | 97 | 5.07 | |
| STAC Rev. KLFailure Detector Category=STAC2026.03 | 100 | 97 | 6.13 | |
| STAC MMD*Failure Detector Category=STAC2026.03 | 100 | 97 | 5.47 | |
| GPT-4o Image QAFailure Detector Category=VLM2026.03 | 100 | 0 | 23.2 | |
| GPT-4o Video QA*Failure Detector Category=VLM2026.03 | 100 | 89 | 21.2 | |
| Gemini 1.5 Pro Image QAFailure Detector Category=VLM2026.03 | 100 | 0 | 21.2 | |
| Gemini 1.5 Pro Video QAFailure Detector Category=VLM2026.03 | 100 | 57 | 17.2 | |
| Claude 3.5 Sonnet Image QAFailure Detector Category=VLM2026.03 | 100 | 6 | 23.2 | |
| SentinelComponents=STAC MMD* + GPT-4o Video QA*2026.03 | 100 | 86 | 5.47 | |
| Claude 3.5 Sonnet Video QAFailure Detector Category=VLM2026.03 | 83 | 31 | 23.2 | |
| Diffusion Output VarianceFailure Detector Category=Diff.2026.03 | 0 | 100 | — |