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

• URL: http://arxiv.org/abs/2601.10496v1
• Date: Thu, 15 Jan 2026 15:14:29 GMT
• Title: Model See, Model Do? Exposure-Aware Evaluation of Bug-vs-Fix Preference in Code LLMs
• Abstract summary: Distinguishing whether an LLM prefers correct code might be influenced by what it's been exposed to during training.<n>We introduce an exposure-aware evaluation framework that quantifies how prior exposure to buggy versus fixed code influences a model's preference.
• Score: 14.187824830583784
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• Diff-Based Code Corruption using LLMs for Large-Scale Bugfix Benchmarking [0.0]
  We introduce MegaBugFix, a large-scale bugfixing benchmark containing 12,629 buggy Python programs synthesized from correct ones by a Large Language Model.<n>We evaluate 13 open-weight models on MegaBugFix and baseline benchmarks, finding consistently lower performance on MegaBugFix.
  arXiv  Detail & Related papers  (2026-06-27T21:04:06Z)
• PrefixNLI: Detecting Factual Inconsistencies as Soon as They Arise [60.63315470285562]
  MiniTruePrefixes is a novel specialized model that better detects factual inconsistencies over text prefixes.<n>We show that integrating MiniTruePrefixes into a controlled decoding framework substantially improves factual consistency in abstractive summarization.
  arXiv  Detail & Related papers  (2025-11-03T09:07:44Z)
• Probing Pre-trained Language Models on Code Changes: Insights from ReDef, a High-Confidence Just-in-Time Defect Prediction Dataset [0.0]
  We present ReDef, a high-confidence benchmark of function-level modifications curated from 22 large-scale C/C++ projects.<n>Defective cases are anchored by revert commits, while clean cases are validated through post-hoc history checks.<n>This pipeline yields 3,164 defective and 10,268 clean modifications, offering substantially more reliable labels than prior existing resources.
  arXiv  Detail & Related papers  (2025-09-11T07:07:11Z)
• Predicting the Performance of Black-box LLMs through Self-Queries [60.87193950962585]
  Large language models (LLMs) are increasingly relied on in AI systems, predicting when they make mistakes is crucial.<n>In this paper, we extract features of LLMs in a black-box manner by using follow-up prompts and taking the probabilities of different responses as representations.<n>We demonstrate that training a linear model on these low-dimensional representations produces reliable predictors of model performance at the instance level.
  arXiv  Detail & Related papers  (2025-01-02T22:26:54Z)
• Subtle Errors in Reasoning: Preference Learning via Error-injected Self-editing [59.405145971637204]
  We propose a novel preference learning framework called eRror-Injected Self-Editing (RISE)<n>RISE injects predefined subtle errors into pivotal tokens in reasoning or steps to construct hard pairs for error mitigation.<n>Experiments validate the effectiveness of RISE, with preference learning on Qwen2-7B-Instruct yielding notable improvements of 3.0% on GSM8K and 7.9% on MATH with only 4.5K training samples.
  arXiv  Detail & Related papers  (2024-10-09T07:43:38Z)
• A Deep Dive into Large Language Models for Automated Bug Localization and Repair [12.756202755547024]
  Large language models (LLMs) have shown impressive effectiveness in various software engineering tasks, including automated program repair (APR) In this study, we take a deep dive into automated bug fixing utilizing LLMs. This methodological separation of bug localization and fixing using different LLMs enables effective integration of diverse contextual information. Toggle achieves the new state-of-the-art (SOTA) performance on the CodeXGLUE code refinement benchmark.
  arXiv  Detail & Related papers  (2024-04-17T17:48:18Z)
• Few-Shot Recalibration of Language Models [23.829795148520834]
  We train a recalibration model that takes in a few unlabeled examples from any given slice and predicts a curve that remaps confidence scores to be more accurate for that slice. Our trained model can recalibrate for arbitrary new slices, without using any labeled data from that slice. Experiments show that our few-shot recalibrator consistently outperforms existing calibration methods.
  arXiv  Detail & Related papers  (2024-03-27T06:25:40Z)
• Do Language Models Learn Semantics of Code? A Case Study in Vulnerability Detection [7.725755567907359]
  We analyze the models using three distinct methods: interpretability tools, attention analysis, and interaction matrix analysis. We develop two annotation methods which highlight the bug semantics inside the model's inputs. Our findings indicate that it is helpful to provide the model with information of the bug semantics, that the model can attend to it, and motivate future work in learning more complex path-based bug semantics.
  arXiv  Detail & Related papers  (2023-11-07T16:31:56Z)
• LARA: A Light and Anti-overfitting Retraining Approach for Unsupervised Time Series Anomaly Detection [49.52429991848581]
  We propose a Light and Anti-overfitting Retraining Approach (LARA) for deep variational auto-encoder based time series anomaly detection methods (VAEs) This work aims to make three novel contributions: 1) the retraining process is formulated as a convex problem and can converge at a fast rate as well as prevent overfitting; 2) designing a ruminate block, which leverages the historical data without the need to store them; and 3) mathematically proving that when fine-tuning the latent vector and reconstructed data, the linear formations can achieve the least adjusting errors between the ground truths and the fine-tuned ones.
  arXiv  Detail & Related papers  (2023-10-09T12:36:16Z)
• Method-Level Bug Severity Prediction using Source Code Metrics and LLMs [0.628122931748758]
  We investigate source code metrics, source code representation using large language models (LLMs), and their combination in predicting bug severity labels. Our results suggest that Decision Tree and Random Forest models outperform other models regarding our several evaluation metrics. CodeBERT finetuning improves the bug severity prediction results significantly in the range of 29%-140% for several evaluation metrics.
  arXiv  Detail & Related papers  (2023-09-06T14:38:07Z)
• Understanding Factual Errors in Summarization: Errors, Summarizers, Datasets, Error Detectors [105.12462629663757]
  In this work, we aggregate factuality error annotations from nine existing datasets and stratify them according to the underlying summarization model. We compare performance of state-of-the-art factuality metrics, including recent ChatGPT-based metrics, on this stratified benchmark and show that their performance varies significantly across different types of summarization models.
  arXiv  Detail & Related papers  (2022-05-25T15:26:48Z)
• How Can We Know When Language Models Know? On the Calibration of Language Models for Question Answering [80.82194311274694]
  We examine the question "how can we know when language models know, with confidence, the answer to a particular query?" We examine three strong generative models -- T5, BART, and GPT-2 -- and study whether their probabilities on QA tasks are well calibrated. We then examine methods to calibrate such models to make their confidence scores correlate better with the likelihood of correctness.
  arXiv  Detail & Related papers  (2020-12-02T03:53:13Z)
• Evaluating Prediction-Time Batch Normalization for Robustness under Covariate Shift [81.74795324629712]
  We call prediction-time batch normalization, which significantly improves model accuracy and calibration under covariate shift. We show that prediction-time batch normalization provides complementary benefits to existing state-of-the-art approaches for improving robustness. The method has mixed results when used alongside pre-training, and does not seem to perform as well under more natural types of dataset shift.
  arXiv  Detail & Related papers  (2020-06-19T05:08:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.