Image Captioning on ImageNet 200 images (test)
1,014.62Token CountLingoLoop
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LingoLoopMLLM=Qwen2.5-VL-3B2025.06 | 1,014.62 | 7,108.5 | 32.5 | |
| LingoLoopMLLM=InstructBLIP2025.06 | 984.65 | 2,814.71 | 54.75 | |
| LingoLoopMLLM=Qwen2.5-VL-7B2025.06 | 825.23 | 4,105.09 | 15.87 | |
| LingoLoopMLLM=InternVL3-8B2025.06 | 613.35 | 3,183.87 | 11.08 | |
| Verbose imagesMLLM=Qwen2.5-VL-3B2025.06 | 525.7 | 3,650.52 | 17.12 | |
| Verbose imagesMLLM=InstructBLIP2025.06 | 451.85 | 1,612.14 | 23.7 | |
| Verbose imagesMLLM=Qwen2.5-VL-7B2025.06 | 384.62 | 1,916.1 | 7.74 | |
| Verbose imagesMLLM=InternVL3-8B2025.06 | 329.02 | 1,634.89 | 5.8 | |
| NoneMLLM=Qwen2.5-VL-7B2025.06 | 82.35 | 405.87 | 1.7 | |
| NoiseMLLM=Qwen2.5-VL-7B2025.06 | 79.29 | 403.71 | 1.65 | |
| NoiseMLLM=InstructBLIP2025.06 | 74.19 | 381.49 | 4.32 | |
| NoneMLLM=InstructBLIP2025.06 | 73.03 | 356.94 | 3.96 | |
| NoiseMLLM=InternVL3-8B2025.06 | 67.1 | 321.29 | 1.22 | |
| NoiseMLLM=Qwen2.5-VL-3B2025.06 | 65.21 | 433.87 | 2.18 | |
| NoneMLLM=InternVL3-8B2025.06 | 65.04 | 318.14 | 1.19 | |
| NoneMLLM=Qwen2.5-VL-3B2025.06 | 64.09 | 427.3 | 2.12 |