Categories

40 个页面

Reasoning

CARFT：Boosting LLM Reasoning via Contrastive Learning with Annotated Chain-of-Thought-based Reinforced Fine-Tuning

ThinkDial：An Open Recipe for Controlling Reasoning Effort in Large Language Models

Think in Blocks：Adaptive Reasoning from Direct Response to Deep Reasoning

Your Reward Function for RL is Your Best PRM for Search：Unifying RL and Search-Based TTS

BEYOND PASS@1：SELF-PLAY WITH VARIATIONAL PROBLEM SYNTHESIS SUSTAINS RLVR

Beyond the 80 20 Rule：High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning

CURE：Critical-Token-Guided Re-concatenation for Entropy-collapse Prevention

EDGE-GRPO：Entropy-Driven GRPO with Guided Error Correction for Advantage Diversity

ETTRL：BALANCING EXPLORATION AND EXPLOITATION IN LLM TEST-TIME REINFORCEMENT LEARNING VIA ENTROPY MECHANISM

Promoting Efficient Reasoning with Verifiable Stepwise Reward