May the Feedback Be with You! Unlocking the Power of Feedback-Driven Deep Learning Framework Fuzzing via LLMs

• URL: http://arxiv.org/abs/2506.17642v3
• Date: Sun, 24 Aug 2025 01:36:49 GMT
• Title: May the Feedback Be with You! Unlocking the Power of Feedback-Driven Deep Learning Framework Fuzzing via LLMs
• Authors: Shaoyu Yang, Chunrong Fang, Haifeng Lin, Xiang Chen, Zhenyu Chen,
• Abstract summary: fuzz testing (Fuzzing) is a simple yet effective way to find bugs in Deep Learning (DL) frameworks.<n>We propose FUEL to effectively utilize the feedback information, which comprises two Large Language Models (LLMs): analysis LLM and generation LLM.<n>We show that FUEL can improve line code coverage of PyTorch and execution by 9.15% and 14.70% over state-of-the-art baselines.
• Score: 20.03968975178177
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Deep Learning (DL) frameworks have served as fundamental components in DL systems over the last decade. However, bugs in DL frameworks could lead to catastrophic consequences in critical scenarios. A simple yet effective way to find bugs in DL frameworks is fuzz testing (Fuzzing). Existing approaches focus on test generation, leaving execution results with high semantic value (e.g., coverage information, bug reports, and exception logs) in the wild, which can serve as multiple types of feedback. To fill this gap, we propose FUEL to effectively utilize the feedback information, which comprises two Large Language Models (LLMs): analysis LLM and generation LLM. Specifically, analysis LLM infers analysis summaries from feedback information, while the generation LLM creates tests guided by these summaries. Furthermore, based on multiple feedback guidance, we design two additional components: (i) a feedback-aware simulated annealing algorithm to select operators for test generation, enriching test diversity. (ii) a program self-repair strategy to automatically repair invalid tests, enhancing test validity. We evaluate FUEL on the two most popular DL frameworks, and experiment results show that FUEL can improve line code coverage of PyTorch and TensorFlow by 9.15% and 14.70% over state-of-the-art baselines (e.g., TitanFuzz and WhiteFox). By the time of submission, FUEL has detected 104 previously unknown bugs for PyTorch and TensorFlow, with 93 confirmed as new bugs, 49 already fixed, and 14 assigned CVE IDs. Our artifact is available at https://github.com/NJU-iSE/FUEL

Related papers

• LLM-Powered Silent Bug Fuzzing in Deep Learning Libraries via Versatile and Controlled Bug Transfer [15.118579443741659]
  We build on the observation that historical bug reports contain rich, underutilized information about silent bugs.<n>We leverage large language models (LLMs) to perform versatile yet controlled bug transfer for silent bug fuzzing.<n>This enables proactive detection of silent bugs by transferring high-risk contexts and oracle designs from known buggy to functionally similar target.
  arXiv  Detail & Related papers  (2026-02-26T14:53:26Z)
• AlgoVeri: An Aligned Benchmark for Verified Code Generation on Classical Algorithms [54.99368693313797]
  Existing benchmarks test only individual languages/tools, so the performance numbers are not directly comparable.<n>We address this gap with AlgoVeri, a benchmark that evaluates vericoding of $77$ classical algorithms in Dafny, Verus, and Lean.
  arXiv  Detail & Related papers  (2026-02-10T06:58:26Z)
• Where LLM Agents Fail and How They can Learn From Failures [62.196870049524364]
  Large Language Model (LLM) agents have shown promise in solving complex, multi-step tasks.<n>They amplify vulnerability to cascading failures, where a single root-cause error propagates through subsequent decisions.<n>Current systems lack a framework that can comprehensively understand agent error in a modular and systemic way.<n>We introduce the AgentErrorTaxonomy, a modular classification of failure modes spanning memory, reflection, planning, action, and system-level operations.
  arXiv  Detail & Related papers  (2025-09-29T18:20:27Z)
• Evaluating the Effectiveness of Coverage-Guided Fuzzing for Testing Deep Learning Library APIs [3.491101173753068]
  We propose FlashFuzz to automatically synthesize API-level harnesses by combining templates, helper functions, and API documentation.<n>Compared to state-of-the-art fuzzing methods, FlashFuzz achieves up to 101.13 to 212.88 percent higher coverage and 1.0x to 5.4x higher validity rate.<n>Our study confirms that CGF can be effectively applied to Deep Learning libraries and provides a strong baseline for future testing approaches.
  arXiv  Detail & Related papers  (2025-09-18T05:10:42Z)
• Agentic Program Repair from Test Failures at Scale: A Neuro-symbolic approach with static analysis and test execution feedback [11.070932612938154]
  We develop an Engineering Agent that fixes the source code based on test failures at scale across diverse software offerings.<n>We provide feedback to the agent through static analysis and test failures so it can refine its solution.<n>In a three month period, 80% of the generated fixes were reviewed, of which 31.5% were landed.
  arXiv  Detail & Related papers  (2025-07-24T19:12:32Z)
• Verifying the Verifiers: Unveiling Pitfalls and Potentials in Fact Verifiers [59.168391398830515]
  We evaluate 12 pre-trained LLMs and one specialized fact-verifier, using a collection of examples from 14 fact-checking benchmarks.<n>We highlight the importance of addressing annotation errors and ambiguity in datasets.<n> frontier LLMs with few-shot in-context examples, often overlooked in previous works, achieve top-tier performance.
  arXiv  Detail & Related papers  (2025-06-16T10:32:10Z)
• Your Fix Is My Exploit: Enabling Comprehensive DL Library API Fuzzing with Large Language Models [49.214291813478695]
  Deep learning (DL) libraries, widely used in AI applications, often contain vulnerabilities like overflows and use buffer-free errors.<n>Traditional fuzzing struggles with the complexity and API diversity of DL libraries.<n>We propose DFUZZ, an LLM-driven fuzzing approach for DL libraries.
  arXiv  Detail & Related papers  (2025-01-08T07:07:22Z)
• Utilize the Flow before Stepping into the Same River Twice: Certainty Represented Knowledge Flow for Refusal-Aware Instruction Tuning [68.57166425493283]
  Refusal-Aware Instruction Tuning (RAIT) enables Large Language Models (LLMs) to refuse to answer unknown questions.<n>This crude approach can cause LLMs to excessively refuse answering questions they could have correctly answered.<n>We introduce Certainty Represented Knowledge Flow for Refusal-Aware Instructions Tuning (CRaFT) to address this issue.
  arXiv  Detail & Related papers  (2024-10-09T14:12:51Z)
• Retrieval-Augmented Test Generation: How Far Are We? [10.473792371852015]
  We investigate the efficacy of RAG-based unit test generation for machine learning (ML/DL) APIs.<n>We examine three domain-specific sources for RAG: API documentation (official guidelines), GitHub issues (developer-reported resolutions), and StackOverflow Q&As.<n>Our study focuses on five widely used Python-based ML/DL libraries, PyTorch, Scikit-learn, Google JAX, and XGBoost.
  arXiv  Detail & Related papers  (2024-09-19T11:48:29Z)
• Exploring Automatic Cryptographic API Misuse Detection in the Era of LLMs [60.32717556756674]
  This paper introduces a systematic evaluation framework to assess Large Language Models in detecting cryptographic misuses. Our in-depth analysis of 11,940 LLM-generated reports highlights that the inherent instabilities in LLMs can lead to over half of the reports being false positives. The optimized approach achieves a remarkable detection rate of nearly 90%, surpassing traditional methods and uncovering previously unknown misuses in established benchmarks.
  arXiv  Detail & Related papers  (2024-07-23T15:31:26Z)
• LLAMAFUZZ: Large Language Model Enhanced Greybox Fuzzing [6.042114639413868]
  Specialized fuzzers can handle complex structured data, but require additional efforts in grammar and suffer from low throughput. In this paper, we explore the potential of utilizing the Large Language Model to enhance greybox fuzzing for structured data. Our LLM-based fuzzer, LLAMAFUZZ, integrates the power of LLM to understand and mutate structured data to fuzzing.
  arXiv  Detail & Related papers  (2024-06-11T20:48:28Z)
• TESTEVAL: Benchmarking Large Language Models for Test Case Generation [15.343859279282848]
  We propose TESTEVAL, a novel benchmark for test case generation with large language models (LLMs)<n>We collect 210 Python programs from an online programming platform, LeetCode, and design three different tasks: overall coverage, targeted line/branch coverage, and targeted path coverage.<n>We find that generating test cases to cover specific program lines/branches/paths is still challenging for current LLMs.
  arXiv  Detail & Related papers  (2024-06-06T22:07:50Z)
• Get my drift? Catching LLM Task Drift with Activation Deltas [55.75645403965326]
  Task drift allows attackers to exfiltrate data or influence the LLM's output for other users.<n>We show that a simple linear classifier can detect drift with near-perfect ROC AUC on an out-of-distribution test set.<n>We observe that this approach generalizes surprisingly well to unseen task domains, such as prompt injections, jailbreaks, and malicious instructions.
  arXiv  Detail & Related papers  (2024-06-02T16:53:21Z)
• LLMDFA: Analyzing Dataflow in Code with Large Language Models [8.92611389987991]
  This paper presents LLMDFA, a compilation-free and customizable dataflow analysis framework. We decompose the problem into several subtasks and introduce a series of novel strategies. On average, LLMDFA achieves 87.10% precision and 80.77% recall, surpassing existing techniques with F1 score improvements of up to 0.35.
  arXiv  Detail & Related papers  (2024-02-16T15:21:35Z)
• Feedback Loops With Language Models Drive In-Context Reward Hacking [78.9830398771605]
  We show that feedback loops can cause in-context reward hacking (ICRH) We identify and study two processes that lead to ICRH: output-refinement and policy-refinement. As AI development accelerates, the effects of feedback loops will proliferate.
  arXiv  Detail & Related papers  (2024-02-09T18:59:29Z)
• 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)
• 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)
• The Hitchhiker's Guide to Program Analysis: A Journey with Large Language Models [18.026567399243]
  Large Language Models (LLMs) offer a promising alternative to static analysis. In this paper, we take a deep dive into the open space of LLM-assisted static analysis. We develop LLift, a fully automated framework that interfaces with both a static analysis tool and an LLM.
  arXiv  Detail & Related papers  (2023-08-01T02:57:43Z)
• Statistical Knowledge Assessment for Large Language Models [79.07989821512128]
  Given varying prompts regarding a factoid question, can a large language model (LLM) reliably generate factually correct answers? We propose KaRR, a statistical approach to assess factual knowledge for LLMs. Our results reveal that the knowledge in LLMs with the same backbone architecture adheres to the scaling law, while tuning on instruction-following data sometimes compromises the model's capability to generate factually correct text reliably.
  arXiv  Detail & Related papers  (2023-05-17T18:54:37Z)
• LeTI: Learning to Generate from Textual Interactions [60.425769582343506]
  We explore LMs' potential to learn from textual interactions (LETI) that not only check their correctness with binary labels but also pinpoint and explain errors in their outputs through textual feedback. Our focus is the code generation task, where the model produces code based on natural language instructions. LETI iteratively fine-tunes the model, using the objective LM, on a concatenation of natural language instructions, LM-generated programs, and textual feedback.
  arXiv  Detail & Related papers  (2023-05-17T15:53:31Z)
• D2A: A Dataset Built for AI-Based Vulnerability Detection Methods Using Differential Analysis [55.15995704119158]
  We propose D2A, a differential analysis based approach to label issues reported by static analysis tools. We use D2A to generate a large labeled dataset to train models for vulnerability identification.
  arXiv  Detail & Related papers  (2021-02-16T07:46:53Z)

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.