Multimodal Reasoning on SIMMC 2.0
79.86ScoreQwen2.5-VL-3B-Instruct
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-3B-InstructBackbone=Qwen2.5-3B, Variant=Qwen2.5-VL-3B-Instruct2026.05 | 79.86 | |
| SKG-VLABackbone=Qwen2.5-3B, Variant=mm-our2026.05 | 78.56 | |
| Qwen2.5 (pre)Backbone=Qwen2.5-3B, Variant=pre2026.05 | 77.95 | |
| Qwen2.5 (llm)Backbone=Qwen2.5-3B, Variant=llm2026.05 | 77.82 | |
| Qwen2.5 (sft)Backbone=Qwen2.5-3B, Variant=sft2026.05 | 77.77 | |
| SKG-VLABackbone=Qwen2.5-1.5B, Variant=mm-our2026.05 | 76.86 | |
| Qwen2.5 (pre)Backbone=Qwen2.5-1.5B, Variant=pre2026.05 | 76.28 | |
| Qwen2.5 (llm)Backbone=Qwen2.5-1.5B, Variant=llm2026.05 | 76.11 | |
| Qwen2.5 (sft)Backbone=Qwen2.5-1.5B, Variant=sft2026.05 | 76.07 | |
| SKG-VLABackbone=Qwen2.5-0.5B, Variant=mm-our2026.05 | 71.63 | |
| Qwen2.5 (pre)Backbone=Qwen2.5-0.5B, Variant=pre2026.05 | 71.02 | |
| Qwen2.5 (sft)Backbone=Qwen2.5-0.5B, Variant=sft2026.05 | 70.96 | |
| Qwen2.5 (llm)Backbone=Qwen2.5-0.5B, Variant=llm2026.05 | 70.84 |