Learned or Memorized ? Quantifying Memorization Advantage in Code LLMs

• URL: http://arxiv.org/abs/2604.13997v1
• Date: Wed, 15 Apr 2026 15:43:10 GMT
• Title: Learned or Memorized ? Quantifying Memorization Advantage in Code LLMs
• Abstract summary: We evaluate 8 open-source code LLMs on 19 benchmarks across four task families.<n> Sensitivity patterns vary widely across models and tasks.<n>For these datasets, models may rely more on learned generalization than direct memorization.
• Score: 12.31163751026309
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The lack of transparency about code datasets used to train large language models (LLMs) makes it difficult to detect, evaluate, and mitigate data leakage. We present a perturbation-based method to quantify memorization advantage in code LLMs, defined as the performance gap between likely seen and unseen inputs. We evaluate 8 open-source code LLMs on 19 benchmarks across four task families: code generation, code understanding, vulnerability detection, and bug fixing. Sensitivity patterns vary widely across models and tasks. For example, StarCoder reaches high sensitivity on some benchmarks (up to 0.8), while QwenCoder remains lower (mostly below 0.4), suggesting differences in generalization behavior. Task categories also differ: code summarization tends to show low sensitivity, whereas test generation is substantially higher. We then analyze two widely discussed benchmarks, CVEFixes and Defects4J, often suspected of leakage. Contrary to common concerns, both show low memorization advantage across models: CVEFixes remains below 0.1, and Defects4J is lower than other program repair benchmarks. These results suggest that, for these datasets, models may rely more on learned generalization than direct memorization. Overall, our findings provide evidence that memorization risk is highly task- and model-dependent, and highlight the need for stronger evaluation protocols, especially in security-focused settings.

Related papers

• RealSec-bench: A Benchmark for Evaluating Secure Code Generation in Real-World Repositories [58.32028251925354]
  Large Language Models (LLMs) have demonstrated remarkable capabilities in code generation, but their proficiency in producing secure code remains a critical, under-explored area.<n>We introduce RealSec-bench, a new benchmark for secure code generation meticulously constructed from real-world, high-risk Java repositories.
  arXiv  Detail & Related papers  (2026-01-30T08:29:01Z)
• Diverse LLMs vs. Vulnerabilities: Who Detects and Fixes Them Better? [1.0026496861838445]
  DVDR-LLM is an ensemble framework that combines outputs from diverse large language models.<n>Our evaluation reveals that DVDR-LLM 10-12% higher detection accuracy compared to the average performance of individual models.
  arXiv  Detail & Related papers  (2025-12-14T03:47:39Z)
• 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)
• Memorize or Generalize? Evaluating LLM Code Generation with Code Rewriting [54.48306552577881]
  We argue that large language models (LLMs) are mostly doing memorization (i.e., replicating or reusing large parts of their training data) versus generalization.<n>Existing evaluations largely proxy neglecting surface/structural similarity, thereby conflating benign reuse of repeated code with harmful recall and memorization task correctness.<n>We propose Memorization Risk Index (MRI), a normalized score that combines two signals: (i) how similar the model's answer for the rewritten task is to the original ground-truth solution, and (ii) how much performance drops from the original task to its rewritten counterpart.
  arXiv  Detail & Related papers  (2025-03-04T05:39:24Z)
• Forget What You Know about LLMs Evaluations -- LLMs are Like a Chameleon [12.13060272830352]
  Large language models (LLMs) often appear to excel on public benchmarks.<n>These high scores may mask an overreliance on dataset-specific surface cues rather than true language understanding.<n>We introduce the Chameleon Benchmark Overfit Detector (C-BOD), a meta-evaluation framework that distorts benchmark prompts.
  arXiv  Detail & Related papers  (2025-02-11T10:43:36Z)
• Do Large Language Model Benchmarks Test Reliability? [66.1783478365998]
  We investigate how well current benchmarks quantify model reliability.<n>Motivated by this gap in the evaluation of reliability, we propose the concept of so-called platinum benchmarks.<n>We evaluate a wide range of models on these platinum benchmarks and find that, indeed, frontier LLMs still exhibit failures on simple tasks.
  arXiv  Detail & Related papers  (2025-02-05T18:58:19Z)
• Are Large Language Models Memorizing Bug Benchmarks? [6.640077652362016]
  Large Language Models (LLMs) have become integral to various software engineering tasks, including code generation, bug detection, and repair.<n>A growing concern within the software engineering community is that benchmarks may not reliably reflect true LLM performance due to the risk of data leakage.<n>We systematically evaluate popular LLMs to assess their susceptibility to data leakage from widely used bug benchmarks.
  arXiv  Detail & Related papers  (2024-11-20T13:46:04Z)
• DebugBench: Evaluating Debugging Capability of Large Language Models [80.73121177868357]
  DebugBench is a benchmark for Large Language Models (LLMs) It covers four major bug categories and 18 minor types in C++, Java, and Python. We evaluate two commercial and four open-source models in a zero-shot scenario.
  arXiv  Detail & Related papers  (2024-01-09T15:46:38Z)
• Understanding the Effectiveness of Large Language Models in Detecting Security Vulnerabilities [12.82645410161464]
  We evaluate the effectiveness of 16 pre-trained Large Language Models on 5,000 code samples from five diverse security datasets. Overall, LLMs show modest effectiveness in detecting vulnerabilities, obtaining an average accuracy of 62.8% and F1 score of 0.71 across datasets. We find that advanced prompting strategies that involve step-by-step analysis significantly improve performance of LLMs on real-world datasets in terms of F1 score (by upto 0.18 on average)
  arXiv  Detail & Related papers  (2023-11-16T13:17:20Z)
• Fake Alignment: Are LLMs Really Aligned Well? [91.26543768665778]
  This study investigates the substantial discrepancy in performance between multiple-choice questions and open-ended questions. Inspired by research on jailbreak attack patterns, we argue this is caused by mismatched generalization.
  arXiv  Detail & Related papers  (2023-11-10T08:01:23Z)
• 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.