Tool-Integrated Reasoning on TIR-Bench
20.8ScoreDeepEyesV2-RL
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepEyesV2-RLLearning Stage=RL, Inference Mode=Zero-shot2025.11 | 20.8 | |
| DeepEyesV2-SFTLearning Stage=SFT, Inference Mode=Zero-shot2025.11 | 18.7 | |
| DeepEyesInference Mode=Zero-shot2025.11 | 17.3 | |
| Qwen2.5-VL 7BLearning Stage=Base, Inference Mode=Zero-shot2025.11 | 16 |