The MCTS canon, curated and (mostly) coded

What it does This repository is a curated awesome-list that catalogs Monte Carlo tree search papers from major machine learning and AI conferences—NeurIPS, ICML, AAAI, ICLR, CVPR, and others—organizing them by year. Each entry typically links to both the paper and its corresponding implementation code. It functions as a living bibliography for researchers and practitioners tracking how MCTS techniques are evolving across reinforcement learning, theorem proving, code generation, and combinatorial optimization.

The interesting bit The list captures the recent expansion of MCTS well beyond its game-playing roots, surfacing appearances in LLM reasoning, molecule generation, SQL synthesis, and software agents. It enforces a “paper plus code” standard for a classical planning algorithm that predates the deep-learning era yet keeps colonizing new research niches.

Key highlights

• Covers papers from 2022 through 2025 (and likely earlier in the full list), spanning venues from NeurIPS to CVPR to KDD.
• Entries include emerging applications: LLM self-training, neural theorem proving, text-to-SQL, embodied multi-agent collaboration, and protein sequence optimization.
• Organized chronologically with direct links to both paper PDFs and source repositories.
• Maintained as part of a family of similar awesome-lists by the same curator.

Caveats

• Many paper and code links in the visible README are empty placeholders, suggesting the list is partially incomplete or awaiting community contributions.
• As a curated markdown list, it offers no search, tagging, or automated indexing beyond scrolling.

Verdict Worth bookmarking if you are doing a literature review on MCTS or hunting for reference implementations of specific variants. Skip it if you are looking for a software framework or a gentle tutorial introduction to the algorithm itself.