Image2Code on Real-world images corpus D3 (Block)
0.76CErLlama3.2-11B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Llama3.2-11Bstatus=Vanilla2026.01 | 0.76 | 0.33 | 0.2 | 0.05 | |
| MiniCPM-V-2-62026.01 | 0.76 | 0.18 | 0.01 | — | |
| Qwen2.5-VL-7B2026.01 | 0.68 | 0.36 | 0.18 | — | |
| Gemma3-12B2026.01 | 0.55 | 0.51 | 0.21 | — | |
| GPT-4o-mini2026.01 | 0.37 | 0.64 | 0.41 | — | |
| Qwen2.5-VL-7B2026.01 | 0.27 | 0.81 | 0.66 | 0.41 | |
| MiniCPM-V-2-62026.01 | 0.16 | 0.81 | 0.5 | 0.19 | |
| Gemma3-12B2026.01 | 0.08 | 0.92 | 0.77 | 0.38 | |
| GPT-4o-mini2026.01 | 0.07 | 0.92 | 0.6 | 0.43 | |
| Llama3.2-11Btstatus=Vanilla2026.01 | 0.06 | 0.07 | 0.2 | — | |
| Llama-VL-TUGstatus=Fine-tuned, backbone=Llama3.2-11B-Vision-Instruct2026.01 | 0.03 | 0.8 | 0.58 | 0.35 | |
| Llama-VL-TUGstatus=Fine-tuned, backbone=Llama3.2-11B-Vision-Instruct2026.01 | 0.03 | 0.71 | 0.35 | — |