CRLGFeb 11

GPU-Fuzz: Finding Memory Errors in Deep Learning Frameworks

arXiv:2602.10478v1h-index: 2
Originality Highly original
AI Analysis

This addresses a critical threat of crashes and security issues for users of deep learning frameworks, though it is incremental as it applies a novel method to a known bottleneck.

The paper tackled the problem of GPU memory errors in deep learning frameworks by introducing GPU-Fuzz, a fuzzer that uses constraint solving to generate test cases, and it found 13 unknown bugs in PyTorch, TensorFlow, and PaddlePaddle.

GPU memory errors are a critical threat to deep learning (DL) frameworks, leading to crashes or even security issues. We introduce GPU-Fuzz, a fuzzer locating these issues efficiently by modeling operator parameters as formal constraints. GPU-Fuzz utilizes a constraint solver to generate test cases that systematically probe error-prone boundary conditions in GPU kernels. Applied to PyTorch, TensorFlow, and PaddlePaddle, we uncovered 13 unknown bugs, demonstrating the effectiveness of GPU-Fuzz in finding memory errors.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes