Visual Color Reasoning on ColorBench cleaned
57.87ScoreJigsaw + CARE
Evaluation Results
| Method | Links | |
|---|---|---|
| Jigsaw + CARETraining Environment=Jigsaw puzzles, CARE=true, Backbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 57.87 | |
| Mix + CL + CARETraining Environment=Mixed (Jigsaw, PatchFit, Rotation), Curriculum Learning=true, CARE=true, Backbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 57.46 | |
| Visual JigsawBackbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 55.21 | |
| Jigsaw + CL + CARETraining Environment=Jigsaw puzzles, Curriculum Learning=true, CARE=true, Backbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 54.6 | |
| VisualSphinxBackbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 53.99 | |
| ViCritBackbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 52.35 | |
| Qwen2.5 VLBackbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 51.12 | |
| JigsawTraining Environment=Jigsaw puzzles, Backbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 50.31 | |
| GRPO-CAREBackbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 48.06 | |
| Vision-ZeroBackbone=Qwen 2.5 VL 7B, Evaluation Mode=thinking mode (think-answer)2025.12 | 46.01 |