Chart Understanding on ECDBench
78.27Description ScoreGPT-5
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-5Model Size=-, # Training Data=-2026.05 | 78.27 | 63.15 | 70.71 | |
| Gemini-2.5-ProModel Size=-, # Training Data=-2026.05 | 76.88 | 44.36 | 60.62 | |
| Qwen3-VLModel Size=8B, # Training Data=-2026.05 | 71.65 | 41.83 | 56.74 | |
| Qwen3-VL + ChartCF (Ours)Model Size=8B, # Training Data=4K pairs (DPO)2026.05 | 70.67 | 42.08 | 56.38 | |
| Qwen3-VL + ECDModel Size=8B, # Training Data=300K (SFT)2026.05 | 70.51 | 43.22 | 56.87 | |
| GPT-4oModel Size=-, # Training Data=-2026.05 | 70.18 | 35.62 | 52.9 | |
| Claude-3.5-SonnetModel Size=-, # Training Data=-2026.05 | 68.14 | 41.99 | 55.07 | |
| Qwen2.5-VL + ECDModel Size=7B, # Training Data=300K (SFT)2026.05 | 66.34 | 35.38 | 50.86 | |
| Qwen2.5-VL + ChartCF (Ours)Model Size=7B, # Training Data=4K pairs (DPO)2026.05 | 66.09 | 37.42 | 51.76 | |
| InternVL3.5 + ECDModel Size=8B, # Training Data=300K (SFT)2026.05 | 65.77 | 39.79 | 52.78 | |
| InternVL3.5 + ChartCF (Ours)Model Size=8B, # Training Data=4K pairs (DPO)2026.05 | 65.28 | 38.89 | 52.08 | |
| Qwen2.5-VLModel Size=7B, # Training Data=-2026.05 | 57.35 | 19.04 | 38.19 | |
| GPT-4o miniModel Size=-, # Training Data=-2026.05 | 57.27 | 24.26 | 40.77 | |
| InternVL3.5Model Size=8B, # Training Data=-2026.05 | 49.67 | 32.27 | 40.97 |