SCoRe uses multi-turn online RL with regularization on self-generated traces to improve LLM self-correction, achieving 15.6% and 9.1% gains on MATH and HumanEval for Gemini models.
arXiv:2307.16039
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
RLAIF matches RLHF on summarization and dialogue tasks, with a direct-RLAIF variant achieving superior results by using LLM rewards directly during training.
AdaMCoT uses dynamic routing of chain-of-thought reasoning in intermediary languages with a reward-based selector to improve cross-lingual factual consistency in LLMs.
citing papers explorer
-
Training Language Models to Self-Correct via Reinforcement Learning
SCoRe uses multi-turn online RL with regularization on self-generated traces to improve LLM self-correction, achieving 15.6% and 9.1% gains on MATH and HumanEval for Gemini models.
-
RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback
RLAIF matches RLHF on summarization and dialogue tasks, with a direct-RLAIF variant achieving superior results by using LLM rewards directly during training.
-
AdaMCoT: Rethinking Cross-Lingual Factual Reasoning through Adaptive Multilingual Chain-of-Thought
AdaMCoT uses dynamic routing of chain-of-thought reasoning in intermediary languages with a reward-based selector to improve cross-lingual factual consistency in LLMs.