List your company

Visual Reasoning on Sudoku (Scale-Specific Accuracy)

100Accuracy (Scale 40)EndoCoT
-4235077Mar 12, 2026

Evaluation results

MethodLinks
EndoCoTTraining Setting=Task-Specific Training2026.031009564
DiffThinkerTraining Setting=Task-Specific Training2026.03945513
EndoCoTTraining Setting=Unified Training2026.03888059
DiffThinkerTraining Setting=Unified Training2026.03452637
Qwen3-VL-8BTraining Setting=Task-Specific Training, Optimization Method=SFT2026.031720
ThinkGenTraining Setting=Zero-Shot2026.03000
ChronoEditTraining Setting=Zero-Shot2026.03000
Qwen3-VL-8BTraining Setting=Zero-Shot2026.03000
Qwen-Image-Edit-2511Training Setting=Zero-Shot2026.03000
Qwen3-VL-8BTraining Setting=Task-Specific Training, Optimization Method=GRPO2026.03000