# FaithformBench: Benchmarking Faithfulness of Mathematical Chain-of-Thought Autoformalisation

arXiv: https://arxiv.org/abs/2608.10916 · HTML: https://mlsanity.com/paper/2608.10916/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Mathematical reasoning, one of 17 directions in LLM reasoning.
- Builds on 19 of its 40 references that are in the map.
- First posted on arXiv 11 Aug 2026.

## Placed in

- LLM reasoning / [Mathematical reasoning](https://mlsanity.com/lr/mathematical-reasoning/)

## Builds on (most cited first, 19 of 19)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [LeanDojo: Theorem Proving with Retrieval-Augmented Language Models](https://mlsanity.com/paper/2306.15626/) (2023)
- [ProcessBench: Identifying Process Errors in Mathematical Reasoning](https://mlsanity.com/paper/2412.06559/) (2024)
- [Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs](https://mlsanity.com/paper/2210.12283/) (2022)
- [DeepSeek-Prover-V2: Advancing Formal Mathematical Reasoning via Reinforcement Learning for Subgoal Decomposition](https://mlsanity.com/paper/2504.21801/) (2025)
- [SelfCheck: Using LLMs to Zero-Shot Check Their Own Step-by-Step Reasoning](https://mlsanity.com/paper/2308.00436/) (2023)
- [Goedel-Prover-V2: Scaling Formal Theorem Proving with Scaffolded Data Synthesis and Self-Correction](https://mlsanity.com/paper/2508.03613/) (2025)
- [Herald: A Natural Language Annotated Lean 4 Dataset](https://mlsanity.com/paper/2410.10878/) (2024)
- [Autoformalization in the Wild: Assessing LLMs on Real-World Mathematical Definitions](https://mlsanity.com/paper/2502.12065/) (2025)
- [ProofBridge: Auto-Formalization of Natural Language Proofs in Lean via Joint Embeddings](https://mlsanity.com/paper/2510.15681/) (2025)
- [Safe: Enhancing Mathematical Reasoning in Large Language Models via Retrospective Step-aware Formal Verification](https://mlsanity.com/paper/2506.04592/) (2025)
- [StepFun-Formalizer: Unlocking the Autoformalization Potential of LLMs through Knowledge-Reasoning Fusion](https://mlsanity.com/paper/2508.04440/) (2025)
- [ReasoningFlow: Semantic Structure of Complex Reasoning Traces](https://mlsanity.com/paper/2506.02532/) (2025)
- [Generalized Tree Edit Distance (GTED): A Faithful Evaluation Metric for Statement Autoformalization](https://mlsanity.com/paper/2507.07399/) (2025)
- [Faithful and Robust LLM-Driven Theorem Proving for NLI Explanations](https://mlsanity.com/paper/2505.24264/) (2025)

## Built on by (most cited first, 5 of 5)

- [Magenta: Closing the Loop Between Mathematical Reasoning and Lean Verification](https://mlsanity.com/paper/2609.11319/) (2026)
- [AIProver: Agentic Auto-Formalization of Mathematical Research via Certificate-Driven Evolving Harness](https://mlsanity.com/paper/2610.05367/) (2026)
- [Hallucination Across the Reasoning Lifecycle: Interface Visibility, Causal Evidence, and Release Control in Large Reasoning Models](https://mlsanity.com/paper/2610.05472/) (2026)
- [claimcheck: Measuring and Improving Specification Faithfulness](https://mlsanity.com/paper/2610.17358/) (2026)
- [Aligning Faithfulness for Theory-Scale Autoformalization](https://mlsanity.com/paper/2610.18498/) (2026)

## Abstract (arXiv)

Autoformalisation (AF) systems map natural language reasoning steps into formal statements in a proof assistant such as Lean. We consider how to assess the faithfulness of these systems. Existing approaches require expensive human-annotated ground truth, or rely on LLM judges or embedding models, which come with limited guarantees of accuracy. In addition, these methods typically only consider inputs that are known to be correct, and therefore do not assess whether the AF translates incorrect inputs faithfully. To address these limitations, we propose a new benchmark for AF faithfulness that is cheap to apply, sound under weak assumptions, and assesses both positive and negative examples. Our method is based on automatically generating perturbed reasoning steps that are designed to be invalid, and then measuring validity preservation on unperturbed steps and invalidity preservation on perturbed steps. We apply our method to eight AF systems across four mathematical datasets, and observe pervasive sycophancy: many AFs "silently correct" invalid inputs into provable statements. The most validity-preserving fine-tuned AFs are also the most sycophantic, suggesting a tension between validity and invalidity preservation in current AF systems.