Visual Reasoning on HR-Bench FSP 8K
86.5ACCThyme
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Thyme2026.02 | 86.5 | — | |
| SCBackbone=DeepEyes-7B2026.02 | 86 | — | |
| ASCBackbone=DeepEyes-7B2026.02 | 86 | 58.1 | |
| ESCBackbone=DeepEyes-7B2026.02 | 86 | 47.2 | |
| DeepconfBackbone=DeepEyes-7B2026.02 | 86 | 53.7 | |
| RTWIBackbone=DeepEyes-7B2026.02 | 86 | 63.5 | |
| CISCBackbone=DeepEyes-7B2026.02 | 85.8 | 50.6 | |
| Self-Uncer.Backbone=DeepEyes-7B2026.02 | 85.3 | 57.5 | |
| Base*Backbone=DeepEyes-7B2026.02 | 81.3 | — | |
| GPT-4o2026.02 | 60.8 | — |