A Metamorphic Testing Approach to Diagnosing Memorization in LLM-Based Program Repair

• URL: http://arxiv.org/abs/2604.21579v1
• Date: Thu, 23 Apr 2026 11:59:53 GMT
• Title: A Metamorphic Testing Approach to Diagnosing Memorization in LLM-Based Program Repair
• Abstract summary: Large language models (LLMs) may memorize bug fixes when evaluation benchmarks overlap with their pretraining data, leading to inflated performance estimates.<n>We investigate whether we can better reveal data leakage by combining metamorphic testing (MT) with negative log-likelihood (NLL)<n>Our results show that all evaluated state-of-the-art LLMs exhibit substantial drops in patch generation success rates on transformed benchmarks, ranging from -4.1% for GPT-4o to -15.98% for Llama-3.1.
• Score: 9.198826017433388
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: LLM-based automated program repair (APR) techniques have shown promising results in reducing debugging costs. However, prior results can be affected by data leakage: large language models (LLMs) may memorize bug fixes when evaluation benchmarks overlap with their pretraining data, leading to inflated performance estimates. In this paper, we investigate whether we can better reveal data leakage by combining metamorphic testing (MT) with negative log-likelihood (NLL), which has been used in prior work as a proxy for memorization. We construct variant benchmarks by applying semantics-preserving transformations to two widely used datasets, Defects4J and GitBug-Java. Using these benchmarks, we evaluate the repair success rates of seven LLMs on both original and transformed versions, and analyze the relationship between performance degradation and NLL. Our results show that all evaluated state-of-the-art LLMs exhibit substantial drops in patch generation success rates on transformed benchmarks, ranging from -4.1% for GPT-4o to -15.98% for Llama-3.1. Furthermore, we find that this degradation strongly correlates with NLL on the original benchmarks, suggesting that models perform better on instances they are more likely to have memorized. These findings show that combining MT with NLL provides stronger and more reliable evidence of data leakage, while metamorphic testing alone can help mitigate its effects in LLM-based APR evaluations.

Related papers

• MLLM-DataEngine: Closing the Loop of Multimodal Instruction Tuning Data Generation [51.15583711678273]
  We propose a novel closed-loop system that bridges data generation, model training, and evaluation.<n>Within each loop iteration, the MLLM-DataEngine first analyzes the weakness of the model based on the evaluation results.<n>We find MLLM-DataEngine could boost the MLLMs capability in a targeted and automatic manner without human participants.
  arXiv  Detail & Related papers  (2026-07-07T08:35:38Z)
• Benchmark Leakage Trap: Can We Trust LLM-based Recommendation? [9.574427977779235]
  This paper identifies and investigates a previously overlooked issue: benchmark data leakage in LLM-based recommendation.<n>Data leakage acts as a critical, previously unaccounted-for factor in LLM-based recommendation, which could impact the true model performance.
  arXiv  Detail & Related papers  (2026-02-14T06:34:19Z)
• LessLeak-Bench: A First Investigation of Data Leakage in LLMs Across 83 Software Engineering Benchmarks [15.584759853972992]
  Large Language Models (LLMs) are widely utilized in software engineering (SE) tasks, such as code generation and automated program repair.<n>Their reliance on extensive and often undisclosed pre-training datasets raises significant concerns about data leakage.<n>This paper presents the first large-scale analysis of data leakage in 83 SE benchmarks concerning LLMs.
  arXiv  Detail & Related papers  (2025-02-10T07:33:49Z)
• Mitigating Forgetting in LLM Fine-Tuning via Low-Perplexity Token Learning [65.23593936798662]
  We show that fine-tuning with LLM-generated data improves target task performance and reduces non-target task degradation.<n>This is the first work to provide an empirical explanation based on token perplexity reduction to mitigate catastrophic forgetting in LLMs after fine-tuning.
  arXiv  Detail & Related papers  (2025-01-24T08:18:56Z)
• Inference-Time Decontamination: Reusing Leaked Benchmarks for Large Language Model Evaluation [61.350306618479365]
  Leakage of benchmarks can prevent the accurate assessment of large language models' true performance. We propose Inference-Time Decontamination (ITD) to address this issue. ITD reduces inflated accuracy by 22.9% on GSM8K and 19.0% on MMLU.
  arXiv  Detail & Related papers  (2024-06-20T04:35:59Z)
• The GitHub Recent Bugs Dataset for Evaluating LLM-based Debugging Applications [20.339673903885483]
  Large Language Models (LLMs) have demonstrated strong natural language processing and code synthesis capabilities. Details about LLM training data are often not made public, which has caused concern as to whether existing bug benchmarks are included. We present the GitHub Recent Bugs dataset, which includes 76 real-world Java bugs that were gathered after the OpenAI data cut-off point.
  arXiv  Detail & Related papers  (2023-10-20T02:37:44Z)
• ReEval: Automatic Hallucination Evaluation for Retrieval-Augmented Large Language Models via Transferable Adversarial Attacks [91.55895047448249]
  This paper presents ReEval, an LLM-based framework using prompt chaining to perturb the original evidence for generating new test cases. We implement ReEval using ChatGPT and evaluate the resulting variants of two popular open-domain QA datasets. Our generated data is human-readable and useful to trigger hallucination in large language models.
  arXiv  Detail & Related papers  (2023-10-19T06:37:32Z)
• MLLM-DataEngine: An Iterative Refinement Approach for MLLM [62.30753425449056]
  We propose a novel closed-loop system that bridges data generation, model training, and evaluation. Within each loop, the MLLM-DataEngine first analyze the weakness of the model based on the evaluation results. For targeting, we propose an Adaptive Bad-case Sampling module, which adjusts the ratio of different types of data. For quality, we resort to GPT-4 to generate high-quality data with each given data type.
  arXiv  Detail & Related papers  (2023-08-25T01:41:04Z)
• LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyond [135.8013388183257]
  We propose a new protocol for inconsistency detection benchmark creation and implement it in a 10-domain benchmark called SummEdits. Most LLMs struggle on SummEdits, with performance close to random chance. The best-performing model, GPT-4, is still 8% below estimated human performance.
  arXiv  Detail & Related papers  (2023-05-23T21:50:06Z)

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.