STEM Multimodal Reasoning on MathVista mini (Score)
56.7ScoreCPO
Evaluation results
| Method | Links | |
|---|---|---|
| CPOFine-tuning setting=CPO, Backbone=Qwen3-VL-2B2026.07 | 56.7 | |
| GRPOFine-tuning setting=GRPO, Backbone=Qwen3-VL-2B2026.07 | 52.4 | |
| GSPOFine-tuning setting=GSPO, Backbone=Qwen3-VL-2B2026.07 | 51.6 | |
| LoRAFine-tuning setting=LoRA, Backbone=Qwen3-VL-2B2026.07 | 51.3 | |
| BaseFine-tuning setting=Base, Backbone=Qwen3-VL-2B2026.07 | 48 | |
| FFTFine-tuning setting=FFT, Backbone=Qwen3-VL-2B2026.07 | 41.6 |