Unleashing the Reasoning Potential of Pre-trained LLMs by Critique Fine-Tuning on One Problem Paper β’ 2506.03295 β’ Published Jun 3, 2025 β’ 17 β’ 2