Image2Code Generation on Synthetic Eval Dataset Flowchart
0.9082BLEULlama-VL-TUG
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Llama-VL-TUG2026.01 | 0.9082 | 90.8177 | 0.9724 | 97.59 | 0.234 | 0.9864 | |
| Qwen2.5-VL-7B-Instruct2026.01 | 0.0094 | 2.3684 | 0.1568 | 23.2609 | -0.5725 | 0.2718 | |
| Llama3.2-11B-Vision-Instruct2026.01 | 0.0004 | 0.7182 | 0.1278 | 22.1123 | -0.6942 | 0.2966 | |
| GPT-4o-mini2026.01 | 0.0003 | 1.0921 | 0.0704 | 13.3751 | -1.0583 | 0.0437 | |
| Gemma3-12B-Instruction-Tuned2026.01 | 0 | 1.5966 | 0.1806 | 22.3908 | -0.6147 | 0.29 | |
| MiniCPM-V-2-62026.01 | 0 | 0.5603 | 0.0482 | 6.8664 | -1.1244 | 0.0517 |