3 skills found
naivoder / MCTSrMonte Carlo Tree Search Self-Refine (MCTSr)
billlin0904 / ParallelMCTSResearchParallel MCTS research
JianChengXingYun / Mctsr Zero[ AAAI-26 ]: MCTSr-Zero is an MCTS framework designed to generate high-quality, principle-aligned conversational data for LLMs by optimizing for "domain alignment" instead of objective correctness in human-centric dialogues.