{"name":"ReMCTS","repo":"MUSE-AI-Lab/ReMCTS","url":"https://midorreal.com/project/muse-ai-lab-remcts","repository":"https://github.com/MUSE-AI-Lab/ReMCTS","description":"ReMCTS","facts":{"license":{"value":"MIT","source":"https://github.com/MUSE-AI-Lab/ReMCTS","checked":"21 Nov 2026"},"language":{"value":"Python","source":"https://github.com/MUSE-AI-Lab/ReMCTS","checked":"21 Nov 2026"},"releases":{"value":"0","source":"https://github.com/MUSE-AI-Lab/ReMCTS/releases","checked":"21 Nov 2026"},"contributors":{"value":"2","source":"https://github.com/MUSE-AI-Lab/ReMCTS/graphs/contributors","checked":"21 Nov 2026"},"archived":{"value":"no","source":"https://github.com/MUSE-AI-Lab/ReMCTS","checked":"21 Nov 2026"}},"stars":251.0,"writeup":{"what_it_is":"ReMCTS is a code generation strategy that uses Monte Carlo Tree Search over a tree of candidate programs. Each candidate is executed in a sandbox, failures are stored as structured reflections, and an evidence-based reward guides the search, with or without visible tests.","audience":"Researchers and developers working on LLM-based code generation","claims":[{"kind":"specific","claim":"ReMCTS with visible tests beats direct generation in 8 of 10 model-dataset pairs on HumanEval and MBPP-Sanitized","excerpt":"ReMCTS with visible tests beats direct generation in 8 of 10 model–dataset pairs","status":"not_checked"},{"kind":"specific","claim":"Candidate programs are executed in a Python or C++17 subprocess","excerpt":"Run each child in a Python or C++17 subprocess","status":"not_checked"},{"kind":"specific","claim":"Failures are stored as structured reflections in path and global memory","excerpt":"write a structured reflection (error type, root cause, fix hint, severity) into the path memory and the global memory","status":"not_checked"},{"kind":"specific","claim":"The search also works when no tests are given","excerpt":"so the search also works when no tests are given","status":"not_checked"},{"kind":"specific","claim":"Models are called through any OpenAI-compatible Chat Completions API and no GPU is needed","excerpt":"models are called through any OpenAI-compatible Chat Completions API, so no GPU is needed","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-11-16"},"why_now":null,"owner_supplied":null,"score":{"verdict":"real","hype":42,"reality":66,"gap":-23,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T08:00:00+00:00","verdict_source":"rule"},"early_read":null,"changes":[{"date":"2026-10-27T02:57:21+00:00","text":"Repository created","source":"https://github.com/MUSE-AI-Lab/ReMCTS"}],"cite":"ReMCTS is a GitHub project at https://github.com/MUSE-AI-Lab/ReMCTS written mainly in Python, described there as \"ReMCTS\".","attribution":"Data from Mid or Real, https://midorreal.com/project/muse-ai-lab-remcts, checked 22 November 2026. Attribution with a link is required.","still_real":null}