ExpThink reduces average CoT response length by up to 77% while improving accuracy on math benchmarks via experience-guided reward shaping and difficulty-adaptive advantage in RL.
Guidelines: • The answer [N/A] means that the paper does not involve crowdsourcing nor research with human subjects
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
ExpThink: Experience-Guided Reinforcement Learning for Adaptive Chain-of-Thought Compression
ExpThink reduces average CoT response length by up to 77% while improving accuracy on math benchmarks via experience-guided reward shaping and difficulty-adaptive advantage in RL.