arXiv AI November 6, 2026 By Duo Chai, Zizhen Liu, Shuhuai Wang, Songwei Pei, Cheng Liu, Huawei Li, Shangguang Wang

Analysis of LLM Vulnerability to GPU Soft Errors: An Instruction-Level Fault Injection Study

The paper presents the first instruction-level fault injection study of large language model (LLM) inference, focusing on how GPU soft errors affect these compute- and memory-intensive models. It examines the impact of model architecture, parameter scale, and task complexity on LLM vulnerability, revealing distinct reliability characteristics compared to earlier neural networks. The study offers new insights that can guide the development of more effective fault tolerance mechanisms for LLMs.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

llms

arXiv Machine Learning

Oct 8

Fault-tolerant foundation models

The paper demonstrates that large‑language models can be trained to tolerate unreliable computer hardware, with their error resilience improving as the models grow. Using 40,000 GPU‑hours of training on simulated faulty hardware, the authors derive modified neural scaling laws that show models learn to compute within efficient error‑correcting codes. This suggests that, if formally fault‑tolerant, such models could enable AI inference on low‑energy, faulty hardware, potentially yielding significant energy savings.

By Trevor McCourt, Ila R. Fiete, Isaac L. Chuang

llms

arXiv AI

Oct 7

An Empirical Fault Vulnerability Exploration of ReRAM-based Process-in-Memory CNN Accelerators

arXiv:2610.07029v1 Announce Type: cross Abstract: Resistive random-access memory (ReRAM)-based Processing-in-Memory (PIM) accelerator is a promising platform for processing massively memory intensive...

By Aniseh Dorostkar, Hamed Farbeh, Hamid R. Zarandi

safety

arXiv AI

Sep 1

M2K: Making the Model-Kernel Interface Explicit for Reliable CUDA Kernel Verification

arXiv:2603.24595v2 Announce Type: replace-cross Abstract: Large language model (LLM) inference systems rely on CUDA kernels for core GPU computations, yet the interface between models and kernels is...

By Mengting He, Shihao Xia, Haomin Jia, Wenfei Wu, Linhai Song

llms

arXiv AI

Jun 2

Not All Errors Are Equal: A Systematic Study of Error Propagation in Large Language Model Inference

arXiv:2606. 02430v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly integrated into high-performance computing (HPC) workflows, accelerating scientific discovery through diverse perspectives such as code generation and domain-specific decision-making.

By Yafan Huang, Sheng Di, Guanpeng Li

llms

arXiv Machine Learning

Jun 19

The Correctness Illusion in LLM-Generated GPU Kernels

arXiv:2606. 20128v1 Announce Type: cross Abstract: Benchmarks for LLM-generated GPU kernels (KernelBench, TritonBench, GEAK) score correctness through fixed-shape, small-sample allclose-style checks.

By Dipankar Sarkar

llms benchmarks

arXiv AI

Oct 15

When Log Parsing Removes What Matters: Detecting Failures in Distributed GPU Workloads

arXiv:2610.14009v1 Announce Type: cross Abstract: Large AI models increasingly rely on distributed GPU execution, yet it remains unclear whether established failure analysis methods transfer effectiv...

By Boyuan Cui, Yuyang Zeng, Yifang Tian, Hans-Arno Jacobsen

llms

One email a morning, machine-written

Related stories

Fault-tolerant foundation models

An Empirical Fault Vulnerability Exploration of ReRAM-based Process-in-Memory CNN Accelerators

M2K: Making the Model-Kernel Interface Explicit for Reliable CUDA Kernel Verification

Not All Errors Are Equal: A Systematic Study of Error Propagation in Large Language Model Inference

The Correctness Illusion in LLM-Generated GPU Kernels

When Log Parsing Removes What Matters: Detecting Failures in Distributed GPU Workloads