Document Visual Question Answering on InfoVQA (val)
86.53AccuracyQwen-8B-DeltaThinker
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen-8B-DeltaThinkerBackbone=Qwen-8B2026.05 | 86.53 | |
| GLM-9B-DeltaThinkerBackbone=GLM-9B2026.05 | 85.78 | |
| Qwen3-VL-8B-ThinkingBackbone=Qwen3-VL-8B2026.05 | 85.33 | |
| GLM-4.1V-9B-ThinkingBackbone=GLM-4.1V-9B2026.05 | 85.04 | |
| ARES-RL-7BBackbone=ARES-RL-7B2026.05 | 81.29 | |
| REVisual-R1Model Family=REVisual2026.05 | 80.01 | |
| Vision-R1-7BBackbone=Vision-R1-7B2026.05 | 78.79 | |
| Bee-8B-RLBackbone=Bee-8B2026.05 | 69.26 | |
| Uni-DPOModel=Qwen2-VL-2B2025.06 | 45.28 | |
| SimPOModel=Qwen2-VL-2B2025.06 | 41.51 | |
| DPOModel=Qwen2-VL-2B2025.06 | 39.73 | |
| BaselineModel=Qwen2-VL-2B2025.06 | 37.77 |