CARFT: Boosting LLM Reasoning via Contrastive Learning with Annotated Chain-of-Thought-based Reinforced Fine-Tuning Paper • 2508.15868 • Published Aug 21 • 3
