Papers for

automated theorem proving teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Generative reward models improve verification of formal reasoning translations

Beyond Solver Verdicts: Generative Reward Models for Autoformalization

Abstract: Neurosymbolic systems rely on mathematical solvers to guarantee reasoning correctness, yet solvers are fundamentally blind to whether a formal translation maintains strict reference-equivalence to a designated formalization. We formalize this vulnerability as Verdict-Preserving-Unfaithfulness (VPU): a failure mode where an incorrect encoding executes successfully and matches the expected verdict. We theoretically prove that structural, verdict-only verification heuristics are mathematically bounded to chance-level detection on these deceptively valid traces. To resolve this, we introduce Generative Verification (GenV), which distills an offline Z3-equivalence oracle into a reference-free, continuous reference-equivalence score by repurposing the language model's native vocabulary space. Mechanistic analysis via decision-projected logit lenses and sparse autoencoders shows this generative readout natively extracts precise spatial error coordinates without explicit localization training. Empirically, our oracle-mined verifier (GenV+HN) achieves 0.961 AUROC in reference-equivalence verification, generalizes zero-shot across unseen translators and divergent formal styles, and yields an 11.3-point downstream accuracy gain in agentic test-time compute allocation.

Thu 10 SeptMachine LearningComputation and Language
The gist
Sometimes computer programs translate math problems into precise language, but those translations can be wrong even if the computer solver says the answer is correct. The authors studied this hidden problem and showed that usual checks only spot errors by guessing. To fix this, they created a new method that uses language models to score how faithful translations are, without needing references. Their method can also find exactly where errors happen and works well even on new kinds of translations.
Open 2609.11085v1

Model training and inference enable advanced math problem solving

An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics

Abstract: We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems.

Wed 9 SeptArtificial Intelligence
The gist
Solving very hard math competition problems is tough for computers because they need to provide clear step-by-step explanations in natural language. The authors improved an existing language model, Nemotron 3 Ultra, by training specialized versions and designing a way to generate, check, and refine math proofs all in plain English without relying on other software tools. Their system scored highly on the International Math Olympiad 2026, enough for a gold medal. They also shared their trained models, data, code, and a new test set of challenging math problems.
Open 2609.10712v1