← Search

Qinghua Zhou

5 accepted papers

2026

Harnessing Non-Adversarial Robustness in Large Language Models

ICML 2026spotlight

The work presents an approach for addressing the challenge of robustness in Large Language Models (LLMs) to alterations and potential errors caused by semantically similar but textually different prompts. Recent works have shown that these kinds of prompt variations can significantly impact the perf…

Cited by 0SourceScholar
2025

Staining and Locking Computer Vision Models Without Retraining

ICCV 2025poster

We introduce new methods of staining and locking computer vision models, to protect their owners' intellectual property. Staining, also known as watermarking, embeds secret behaviour into a model which can later be used to identify it, while locking aims to make a model unusable unless a secret trig…

Cited by 0SourcePDFScholar
2024

Stealth edits to large language models

NeurIPS 2024poster

We reveal the theoretical foundations of techniques for editing large language models, and present new methods which can do so without requiring retraining. Our theoretical insights show that a single metric (a measure of the intrinsic dimension of the model's features) can be used to assess a model…

2023

RWKV: Reinventing RNNs for the Transformer Era

EMNLP 2023long findings

Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence length. In contrast, recurrent neural networks (RNNs) exhibit linear scaling in memory and computational requirements but st…

Cited by 0SourceScholar