Chart VQA on chatmimic
83.2AccuracyOpenAI GPT-4o
Evaluation results
| Method | Links | |
|---|---|---|
| OpenAI GPT-4oZero-shot=true2026.02 | 83.2 | |
| Qwen2.5-VL-32B + AT-RL (Ours)Zero-shot=true2026.02 | 79.4 | |
| Qwen2.5-VL-32B + VPPOZero-shot=true2026.02 | 79.2 | |
| Qwen2.5-VL-32B InstructZero-shot=true2026.02 | 78.8 | |
| Claude 3.5 SonnetZero-shot=true2026.02 | 65.4 |