Multimodal Reasoning on Counting
68.33AccuracyCCoT
Evaluation Results
| Method | Links | |
|---|---|---|
| CCoTBackbone=VLAA Thinking-7B2026.05 | 68.33 | |
| Unsilencing Latent ReasoningBackbone=VLAA Thinking-7B2026.05 | 68.33 | |
| VanillaBackbone=VLAA Thinking-7B2026.05 | 65 | |
| DMLRBackbone=VLAA Thinking-7B2026.05 | 64.17 | |
| MCoTBackbone=VLAA Thinking-7B2026.05 | 62.5 | |
| Unsilencing Latent ReasoningBackbone=R1 OneVision-7B2026.05 | 61.67 | |
| ICoTBackbone=VLAA Thinking-7B2026.05 | 60.83 | |
| CCoTBackbone=R1 OneVision-7B2026.05 | 60 | |
| VanillaBackbone=R1 OneVision-7B2026.05 | 59.17 | |
| ICoTBackbone=R1 OneVision-7B2026.05 | 59.17 | |
| MCoTBackbone=R1 OneVision-7B2026.05 | 58.33 | |
| DMLRBackbone=R1 OneVision-7B2026.05 | 54.17 |