Multimodal Large Language Model Evaluation on MME-RealWorld
43.7ReasoningThyme
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ThymeBackbone=QwenVL2.5-7B2025.10 | 43.7 | 60.1 | 58.1 | |
| UG-SearchBackbone=QwenVL2.5-7B2025.10 | 35.1 | 58.5 | 55.7 | |
| ViCropBackbone=QwenVL2.5-7B2025.10 | 34.2 | 53.4 | 51.1 | |
| TextCoTBackbone=QwenVL2.5-7B2025.10 | 33.7 | 53.3 | 50.9 | |
| QwenVL2.5-7B2025.10 | 33.4 | 51.5 | 49.3 |