28 citations · 35 across the 15 of their papers we have counts for
18 papers
Steering Vectors for Competing Ethical Frameworks Are Interchangeable: A Competing-Target Test for Activation Steering
Weilun Xu, Zipeng Sun
Activation steering adds a vector to a model's hidden states to control a named behaviour. A gain on that behaviour, however, does not show that the vector controls it directly: it…
ChLogic: Evaluating Robustness of Logical Reasoning in Chinese Expressions
Peixian Zhou, Yuxu Chen, Chaorui Zhang +3
Large language models perform increasingly well on standardized logical reasoning benchmarks, but whether this ability remains robust beyond English is unclear. We introduce ChLogi…
On the Non-decoupling of Supervised Fine-tuning and Reinforcement Learning in Post-training
Xueyan Niu, Bo Bai, Wei Han +1
Post-training of large language models routinely interleaves supervised fine-tuning (SFT) with reinforcement learning (RL). These two methods have different objectives: SFT minimiz…
A Convergent Primal-Dual Algorithm for Computing Rate-Distortion-Perception Functions
Chunhui Chen, Linyi Chen, Xueyan Niu +1
Recent advances in Rate-Distortion-Perception (RDP) theory highlight the importance of balancing compression level, reconstruction quality, and perceptual fidelity. While previous…
NeuralDB: Scaling Knowledge Editing in LLMs to 100,000 Facts with Neural KV Database
Weizhi Fei, Hao Shi, Jing Xu +7
Efficiently editing knowledge stored in large language models (LLMs) enables model updates without large-scale training. One possible solution is Locate-and-Edit (L\&E), allowing s…
Efficient Prompt Compression with Evaluator Heads for Long-Context Transformer Inference
Weizhi Fei, Xueyan Niu, Guoqing Xie +3
Although applications involving long-context inputs are crucial for the effective utilization of large language models (LLMs), they also result in increased computational costs and…
Learning predictable and robust neural representations by straightening image sequences
Xueyan Niu, Cristina Savin, Eero P. Simoncelli
Prediction is a fundamental capability of all living organisms, and has been proposed as an objective for learning sensory representations. Recent work demonstrates that in primate…
Retrieval Meets Reasoning: Dynamic In-Context Editing for Long-Text Understanding
Weizhi Fei, Xueyan Niu, Guoqing Xie +4
Current Large Language Models (LLMs) face inherent limitations due to their pre-defined context lengths, which impede their capacity for multi-hop reasoning within extensive textua…
Beyond Scaling Laws: Understanding Transformer Performance with Associative Memory
Xueyan Niu, Bo Bai, Lei Deng +1
Increasing the size of a Transformer does not always lead to enhanced performance. This phenomenon cannot be explained by the empirical scaling laws. Furthermore, the model's enhan…
Computation and Critical Transitions of Rate-Distortion-Perception Functions With Wasserstein Barycenter
Chunhui Chen, Xueyan Niu, Wenhao Ye +2
The information rate-distortion-perception (RDP) function characterizes the three-way trade-off between description rate, average distortion, and perceptual quality measured by dis…
Extreme Video Compression with Pre-trained Diffusion Models
Bohan Li, Yiming Liu, Xueyan Niu +3
Diffusion models have achieved remarkable success in generating high quality image and video data. More recently, they have also been used for image compression with high perceptua…
Extending Context Window of Large Language Models via Semantic Compression
Weizhi Fei, Xueyan Niu, Pingyi Zhou +4
Transformer-based Large Language Models (LLMs) often impose limitations on the length of the text input to ensure the generation of fluent and relevant responses. This constraint r…
High Perceptual Quality Wireless Image Delivery with Denoising Diffusion Models
Selim F. Yilmaz, Xueyan Niu, Bo Bai +3
We consider the image transmission problem over a noisy wireless channel via deep learning-based joint source-channel coding (DeepJSCC) along with a denoising diffusion probabilist…
A Hybrid Wireless Image Transmission Scheme with Diffusion
Xueyan Niu, Xu Wang, Deniz Gündüz +3
We propose a hybrid joint source-channel coding (JSCC) scheme, in which the conventional digital communication scheme is complemented with a generative refinement component to impr…
Conditional Rate-Distortion-Perception Trade-Off
Xueyan Niu, Deniz Gündüz, Bo Bai +1
Recent advances in machine learning-aided lossy compression are incorporating perceptual fidelity into the rate-distortion theory. In this paper, we study the rate-distortion-perce…
Computation of Rate-Distortion-Perception Functions With Wasserstein Barycenter
Chunhui Chen, Xueyan Niu, Wenhao Ye +4
The nascent field of Rate-Distortion-Perception (RDP) theory is seeing a surge of research interest due to the application of machine learning techniques in the area of lossy compr…
Conditional graph entropy as an alternating minimization problem
Viktor Harangi, Xueyan Niu, Bo Bai
Conditional graph entropy is known to be the minimal rate for a natural functional compression problem with side information at the receiver. In this paper we show that it can be f…
Generalizing Körner's graph entropy to graphons
Viktor Harangi, Xueyan Niu, Bo Bai
Körner introduced the notion of graph entropy in 1973 as the minimal code rate of a natural coding problem where not all pairs of letters can be distinguished in the alphabet. Late…