Scientific Table & Figure Analysis on ANABENCH 1.0 (test)
57.17S_ACCANAGENT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ANAGENTSize=8B, Backbone=Qwen3-VL, Mode=SFT2026.02 | 57.17 | 54.83 | 83.83 | 74.67 | 73.83 | 68.87 | |
| ANAGENTSize=4B, Backbone=Qwen3-VL, Mode=SFT + RL2026.02 | 56.5 | 57.17 | 79.33 | 73.33 | 68.5 | 66.97 | |
| ANAGENTSize=8B, Backbone=InternVL-3.5, Mode=SFT2026.02 | 56.33 | 52.33 | 79 | 75.17 | 72.17 | 67 | |
| ANAGENTSize=8B, Backbone=InternVL-3.5, Mode=Training-Free2026.02 | 55.67 | 35.17 | 71.83 | 71.67 | 65.5 | 59.97 | |
| ANAGENTSize=7B, Backbone=Qwen2.5-VL, Mode=SFT2026.02 | 54.33 | 53.67 | 80.83 | 73 | 74.5 | 67.27 | |
| ANAGENTSize=8B, Backbone=Qwen3-VL, Mode=Training-Free2026.02 | 54.17 | 39.5 | 77.5 | 72.83 | 64.5 | 61.7 | |
| ANAGENTSize=3B, Backbone=Qwen2.5-VL, Mode=SFT + RL2026.02 | 54 | 55.17 | 78.83 | 72.17 | 67.33 | 65.5 | |
| ANAGENTSize=4B, Backbone=Qwen3-VL, Mode=SFT2026.02 | 53.5 | 50.17 | 76.17 | 72.17 | 65.33 | 63.47 | |
| Qwen3-VLSize=8B, Mode=Baseline2026.02 | 52.83 | 31.17 | 74.17 | 69.5 | 61.17 | 57.77 | |
| InternVL-3.5Size=8B, Mode=Baseline2026.02 | 52.67 | 30.5 | 71.67 | 68.5 | 60.83 | 56.83 | |
| ANAGENTSize=7B, Backbone=Qwen2.5-VL, Mode=Training-Free2026.02 | 52.17 | 37.5 | 74.17 | 69.83 | 63 | 59.33 | |
| ANAGENTSize=4B, Backbone=Qwen3-VL, Mode=Training-Free2026.02 | 51.67 | 31.83 | 68.5 | 67.5 | 60.83 | 56.07 | |
| Qwen2.5-VLSize=7B, Mode=Baseline2026.02 | 50.83 | 32.17 | 73.5 | 67.5 | 60.33 | 56.87 | |
| ANAGENTSize=4B, Backbone=InternVL-3.5, Mode=SFT2026.02 | 50.83 | 47.5 | 77.83 | 69.33 | 64.67 | 62.03 | |
| ANAGENTSize=3B, Backbone=Qwen2.5-VL, Mode=SFT2026.02 | 50.33 | 48.33 | 76.5 | 71.17 | 65.67 | 62.4 | |
| ANAGENTSize=3B, Backbone=Qwen2.5-VL, Mode=Training-Free2026.02 | 47.67 | 25.13 | 67.83 | 66.17 | 56.5 | 52.66 | |
| Qwen3-VLSize=4B, Mode=Baseline2026.02 | 46.5 | 28.5 | 68.33 | 67.17 | 57 | 53.5 | |
| ANAGENTSize=4B, Backbone=InternVL-3.5, Mode=Training-Free2026.02 | 44.17 | 30.83 | 70.33 | 66 | 59.5 | 54.17 | |
| Qwen2.5-VLSize=3B, Mode=Baseline2026.02 | 43.17 | 24.88 | 64.5 | 64.83 | 56 | 50.68 | |
| InternVL-3.5Size=4B, Mode=Baseline2026.02 | 42 | 27.17 | 69.67 | 64.33 | 55.33 | 51.7 |