Reasoning on MMMU-Pro (Accuracy)
55.9Accuracy (Reasoning on MMMU-Pro)HEED
Evaluation results
| Method | Links | |
|---|---|---|
| HEEDPipeline Stage=C4+, Post-training=SFT+DPO2026.05 | 55.9 | |
| TeacherPipeline Stage=C0, Post-training=SFT+DPO2026.05 | 55.2 | |
| RSAPipeline Stage=C3+, Post-training=SFT+DPO2026.05 | 54.5 | |
| HSAPipeline Stage=C2+, Post-training=SFT+DPO2026.05 | 53.3 | |
| KDPipeline Stage=C1+, Post-training=SFT+DPO2026.05 | 52.4 | |
| LoMoModel=Qwen3.5-9B, Evaluation Protocol=Standard Evaluation2026.05 | 46.18 | |
| LoMoModel=Qwen3.5-9B, Evaluation Protocol=Rendered Evaluation2026.05 | 45.29 | |
| Standard SFTModel=Qwen3.5-9B, Evaluation Protocol=Standard Evaluation2026.05 | 39.83 | |
| LoMoModel=LLaVA-OV1.5-8B, Evaluation Protocol=Standard Evaluation2026.05 | 36.36 | |
| Standard SFTModel=LLaVA-OV1.5-8B, Evaluation Protocol=Standard Evaluation2026.05 | 35.24 | |
| Standard SFTModel=Qwen3.5-9B, Evaluation Protocol=Rendered Evaluation2026.05 | 33.06 | |
| LoMoModel=LLaVA-OV1.5-8B, Evaluation Protocol=Rendered Evaluation2026.05 | 27.59 | |
| Standard SFTModel=LLaVA-OV1.5-8B, Evaluation Protocol=Rendered Evaluation2026.05 | 16.01 |