# Model See, Model Do? Exposure-Aware Evaluation of Bug-vs-Fix Preference in Code LLMs

arXiv: https://arxiv.org/abs/2601.10496 · HTML: https://mlsanity.com/paper/2601.10496/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 35 of its 63 references that are in the map.
- First posted on arXiv 15 Jan 2026.

## Builds on (most cited first, 30 of 35)

- [Extracting Training Data from Large Language Models](https://mlsanity.com/paper/2012.07805/) (2020)
- [Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation](https://mlsanity.com/paper/2305.01210/) (2023)
- [ML-Leaks: Model and Data Independent Membership Inference Attacks and Defenses on Machine Learning Models](https://mlsanity.com/paper/1806.01246/) (2018)
- [Agentless: Demystifying LLM-based Software Engineering Agents](https://mlsanity.com/paper/2407.01489/) (2024)
- [Membership Inference Attacks on Machine Learning: A Survey](https://mlsanity.com/paper/2103.07853/) (2021)
- [CRUXEval: A Benchmark for Code Reasoning, Understanding and Execution](https://mlsanity.com/paper/2401.03065/) (2024)
- [Using an LLM to Help With Code Understanding](https://mlsanity.com/paper/2307.08177/) (2023)
- [Quantifying Privacy Risks of Masked Language Models Using Membership Inference Attacks](https://mlsanity.com/paper/2203.03929/) (2022)
- [A Survey on Uncertainty Quantification of Large Language Models: Taxonomy, Open Research Challenges, and Future Directions](https://mlsanity.com/paper/2412.05563/) (2024)
- [A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment](https://mlsanity.com/paper/2504.15585/) (2025)
- [Rethinking LLM Memorization through the Lens of Adversarial Compression](https://mlsanity.com/paper/2404.15146/) (2024)
- [Can ChatGPT replace StackOverflow? A Study on Robustness and Reliability of Large Language Model Code Generation](https://mlsanity.com/paper/2308.10335/) (2023)
- [Bugs in Large Language Models Generated Code: An Empirical Study](https://mlsanity.com/paper/2403.08937/) (2024)
- [SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)](https://mlsanity.com/paper/2406.17975/) (2024)
- [CodeFill: Multi-token Code Completion by Jointly Learning from Structure and Naming Sequences](https://mlsanity.com/paper/2202.06689/) (2022)
- [Calibration and Correctness of Language Models for Code](https://mlsanity.com/paper/2402.02047/) (2024)
- [Memorization in NLP Fine-tuning Methods](https://mlsanity.com/paper/2205.12506/) (2022)
- [Large Language Models and Simple, Stupid Bugs](https://mlsanity.com/paper/2303.11455/) (2023)
- [Exploring Memorization in Fine-tuned Language Models](https://mlsanity.com/paper/2310.06714/) (2023)
- [Towards More Realistic Membership Inference Attacks on Large Diffusion Models](https://mlsanity.com/paper/2306.12983/) (2023)
- [Measuring memorization in language models via probabilistic extraction](https://mlsanity.com/paper/2410.19482/) (2024)
- [Language Models for Code Completion: A Practical Evaluation](https://mlsanity.com/paper/2402.16197/) (2024)
- [SoK: Memorization in General-Purpose Large Language Models](https://mlsanity.com/paper/2310.18362/) (2023)
- [How Often Do Single-Statement Bugs Occur? The ManySStuBs4J Dataset](https://mlsanity.com/paper/1905.13334/) (2019)
- [Data Portraits: Recording Foundation Model Training Data](https://mlsanity.com/paper/2303.03919/) (2023)
- [Unveiling Memorization in Code Models](https://mlsanity.com/paper/2308.09932/) (2023)
- [Traces of Memorisation in Large Language Models for Code](https://mlsanity.com/paper/2312.11658/) (2023)
- [TSSB-3M: Mining single statement bugs at massive scale](https://mlsanity.com/paper/2201.12046/) (2022)
- [Do LLMs Memorize Recommendation Datasets? A Preliminary Study on MovieLens-1M](https://mlsanity.com/paper/2505.10212/) (2025)
- [Does Your Neural Code Completion Model Use My Code? A Membership Inference Approach](https://mlsanity.com/paper/2404.14296/) (2024)

## Built on by (most cited first, 1 of 1)

- [Goedel-Code-Prover: Hierarchical Proof Search for Open State-of-the-Art Code Verification](https://mlsanity.com/paper/2603.19329/) (2026)

## Abstract (arXiv)

Large language models are increasingly used for code generation and debugging, but their outputs can still contain bugs, that originate from training data. Distinguishing whether an LLM prefers correct code, or a familiar incorrect version might be influenced by what it's been exposed to during training. We introduce an exposure-aware evaluation framework that quantifies how prior exposure to buggy versus fixed code influences a model's preference. Using the ManySStuBs4J benchmark, we apply Data Portraits for membership testing on the Stack-V2 corpus to estimate whether each buggy and fixed variant was seen during training. We then stratify examples by exposure and compare model preference using code completion as well as multiple likelihood-based scoring metrics We find that most examples (67%) have neither variant in the training data, and when only one is present, fixes are more frequently present than bugs. In model generations, models reproduce buggy lines far more often than fixes, with bug-exposed examples amplifying this tendency and fix-exposed examples showing only marginal improvement. In likelihood scoring, minimum and maximum token-probability metrics consistently prefer the fixed code across all conditions, indicating a stable bias toward correct fixes. In contrast, metrics like the Gini coefficient reverse preference when only the buggy variant was seen. Our results indicate that exposure can skew bug-fix evaluations and highlight the risk that LLMs may propagate memorised errors in practice.