Image2Code Generation on Synthetic Eval Dataset Block
0.8676BLEULlama-VL-TUG
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Llama-VL-TUG2026.01 | 0.8676 | 86.7621 | 0.9432 | 94.6115 | 0.3255 | 0.8953 | |
| Gemma3-12B-Instruction-Tuned2026.01 | 0.0488 | 5.6815 | 0.2353 | 32.8237 | -0.5744 | 0.2474 | |
| Qwen2.5-VL-7B-Instruct2026.01 | 0.0148 | 3.0378 | 0.2211 | 24.828 | -0.6823 | 0.2275 | |
| Llama3.2-11B-Vision-Instruct2026.01 | 0.0059 | 1.9446 | 0.1823 | 27.9335 | -0.6176 | 0.2278 | |
| GPT-4o-mini2026.01 | 0.0021 | 2.4243 | 0.196 | 18.9216 | -0.7492 | 0.2077 | |
| MiniCPM-V-2-62026.01 | 0 | 0.4034 | 0.0669 | 2.9787 | -1.5357 | 0.0231 |