Image2Description on Synthetic Eval Dataset Block
89.52BLEULlama-VL-TUG
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Llama-VL-TUG2026.01 | 89.52 | 89.5213 | 95.91 | 97.7836 | 71.9 | 95.61 | |
| Gemma3-12B-Instruction-Tuned2026.01 | 77.17 | 77.174 | 87.65 | 88.7206 | 45.29 | 80 | |
| Llama3.2-11B-Vision-Instruct2026.01 | 69.19 | 69.1903 | 82.96 | 82.1518 | 32.38 | 76.18 | |
| GPT-4o-mini2026.01 | 69.16 | 69.1613 | 81.12 | 83.1176 | 42.94 | 75.4 | |
| Qwen2.5-VL-7B-Instruct2026.01 | 66.44 | 66.4355 | 76.62 | 80.0331 | 30.56 | 74.24 | |
| MiniCPM-V-2-62026.01 | 33.66 | 33.6623 | 48.48 | 45.9604 | -81.65 | 55.42 |