{"slug": "training-a-chess-reasoning-agent", "title": "Training a Chess Reasoning Agent", "summary": "Luke Salamone trained LLM reasoning agents at 0.8B, 2B and 9B parameters to play chess by calling tools that track the board state, enabling strong performance with relatively few parameters. The demo code is available on GitHub at lukesalamone/chess_reasoner_demo. Salamone cites the DeepSeek R1 paper's demonstration that thinking tokens let an LLM solve verifiable tasks such as math problems, and its finding that supervised examples are needed to bootstrap the thinking process and prevent nonsense chains of thought.", "body_md": "**TLDR: I trained an LLM reasoning agent to play chess, with model sizes 0.8B, 2B and 9B. The agent calls tools which help it keep track of the board state, which enables good performance with a relatively small number of parameters.**\n\n**The demo code is [available on Github](https://github.com/lukesalamone/chess_reasoner_demo).**\n\nI became excited about the power of RLVR after the [Deepseek R1 paper](./posts/notes-on-deepseek-r1/) came out. They demonstrated that through the use of “thinking” tokens, an LLM could solve verifiable tasks like math problems. Additionally, they described the necessity of bootstrapping the thinking process with supervised examples, which prevents the model from generating a nonsense chain of thought.", "url": "https://wpnews.pro/news/training-a-chess-reasoning-agent", "canonical_source": "https://blog.lukesalamone.com/posts/chess-reasoning-llm/", "published_at": "2026-10-12 01:06:06+00:00", "updated_at": "2026-10-15 05:53:21.009839+00:00", "lang": "en", "topics": ["large-language-models", "ai-agents", "ai-research", "machine-learning"], "entities": ["Luke Salamone", "DeepSeek R1", "GitHub"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/training-a-chess-reasoning-agent", "markdown": "https://wpnews.pro/news/training-a-chess-reasoning-agent.md", "text": "https://wpnews.pro/news/training-a-chess-reasoning-agent.txt", "jsonld": "https://wpnews.pro/news/training-a-chess-reasoning-agent.jsonld"}}