{"slug": "show-hn-theoremsmith-mass-producing-hard-proof-problems-to-train-llms", "title": "Show HN: TheoremSmith: Mass-Producing Hard Proof Problems to Train LLMs", "summary": "Abundant AI released TheoremSmith, an open-source pipeline on GitHub for generating synthetic formal verification environments that reliably produce multi-hour, frontier-difficulty proof tasks to train LLMs via reinforcement learning. The project's author said the pipeline can scale throughput effectively and asked for feedback on how models could still cheat the verifier and whether synthetic tasks will transfer to real proofs.", "body_md": "Making models better at formal verification will help push the boundaries of math research and significantly reduce the number of bugs in our code. Recently, training models on RL environments (tasks paired with verifiers that reward correct solutions) has driven big gains in model performance, and proofs are unusually easy to verify. Here we introduce TheoremSmith, a pipeline for generating synthetic formal verification environments. It reliably creates multi-hour, frontier-difficulty tasks and can scale throughput effectively.\n\nI'd appreciate any feedback, especially on ways models could still cheat the verifier, and on whether synthetic tasks like these are likely to help on real proofs.\n\nCode: [https://github.com/abundant-ai/theoremsmith](https://github.com/abundant-ai/theoremsmith)\n\nComments URL: [https://news.ycombinator.com/item?id=50444276](https://news.ycombinator.com/item?id=50444276)\n\nPoints: 3\n\n# Comments: 0", "url": "https://wpnews.pro/news/show-hn-theoremsmith-mass-producing-hard-proof-problems-to-train-llms", "canonical_source": "https://charleshuang.me/2026/09/27/theoremsmith/", "published_at": "2026-11-10 18:33:17+00:00", "updated_at": "2026-11-10 18:52:59.401368+00:00", "lang": "en", "topics": ["ai-research", "large-language-models", "ai-safety", "developer-tools", "machine-learning"], "entities": ["TheoremSmith", "Abundant AI", "GitHub", "Hacker News"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/show-hn-theoremsmith-mass-producing-hard-proof-problems-to-train-llms", "markdown": "https://wpnews.pro/news/show-hn-theoremsmith-mass-producing-hard-proof-problems-to-train-llms.md", "text": "https://wpnews.pro/news/show-hn-theoremsmith-mass-producing-hard-proof-problems-to-train-llms.txt", "jsonld": "https://wpnews.pro/news/show-hn-theoremsmith-mass-producing-hard-proof-problems-to-train-llms.jsonld"}}