204 citations · 509 across the 65 of their papers we have counts for
75 papers
Securing Computer-Use Agents Against Branch Steering Attacks
Giulio Zingrillo, Hanna Foerster, Ilia Shumailov +2
Modern Computer Use Agents (CUAs) directly interact with graphical user interfaces and execute third-party web tools, exposing them to indirect prompt injection across every render…
iS-KV: Online Low-Rank KV Cache Compression via Block-Incremental SVD
Yiren Zhao, Guanghui Song, Tianrui Qin +3
Long chain-of-thought reasoning substantially increases KV-cache memory during autoregressive decoding, as every generated token introduces new key and value states and causes the…
AgentKV: Phase-Aware KV Eviction for Agentic LLMs
Taowen Tony Liu, Jeffrey T. H. Wong, Can Xiao +3
Agentic serving can consume orders of magnitude more tokens than chatbot workloads, stressing both KV-cache capacity and decode-time bandwidth. Most KV-eviction methods score cache…
OasisKV: Scaling In-Decode KV Cache Beyond HBM with Lookahead Sparse Prefetching
Can Xiao, Sukmin Cho, Junbong We +7
Large language model (LLM) inference serving is increasingly constrained by memory rather than compute. As long-context and long-form reasoning workloads become more prevalent, the…
The Geometry of Flow-Matching Uncertainty: A Cost-free Uncertainty Proxy and Its Application in Flow-based VLA Failure Detection
Ziyang Rao, Yiren Zhao, Weiyu Guo +3
Flow matching (FM) has become a popular action head paradigm for modern embodied models. However, as a conditional generative model, it does not explicitly expose its inherent unce…
Suppression Sticks, Locality Is Fragile: A Closed-Loop Target-and-Control Audit of Task-Vector Negation in VLA Policies
Shaoguang Wang, Weiyu Guo, Rushi Dai +3
Task-vector arithmetic offers a closed-form way to modify a model, yet its behavioral locality remains unclear in closed-loop robot control. We present a target-and-control audit o…
How Should Vision-Language-Action Models Use Proprioceptive State?
Yiren Zhao, Ziyang Chen, Ziyang Rao +5
Recent Vision-Language-Action (VLA) models almost universally take robot proprioceptive state as input, yet wire it in incompatible ways -- serialized into text prompts, projected…
Source-Lifted Flow Matching for Intervenable Multimodal Imitation
He Zhang, Ying Sun, Pengteng Li +7
Flow-matching policies are promising for imitation learning because they model complex multimodal action distributions. However, their stochasticity is largely passive: repeated sa…
DumpsterCluster: From Dumpster Diving to Serving LLaMA-70B on $60 GPUs
Zeyu Cao, Xuan Guo, Cheng Zhang +3
As AI datacenters retire functional GPUs, vast quantities of still capable accelerators enter secondary markets. This paper investigates whether these retired GPUs can find a produ…
PHASER: Phase-Aware and Semantic Experience Replay for Vision-Language-Action Models
Ziyang Chen, Shaoguang Wang, Weiyu Guo +5
Vision-Language-Action (VLA) models have achieved remarkable success in language-conditioned robotic manipulation. However, deploying these models in open-ended environments requir…
TriAxialKV: Toward Extreme Low-Precision KV-Cache Quantization for Agentic Inference Tasks
Hanzhang Shen, Haoran Wu, Yiren Zhao +1
Agentic workloads have emerged as a major workload for LLM inference. They differ significantly from chat-only workloads, requiring long-context processing, the ability to handle m…
Quantamination: Dynamic Quantization Leaks Your Data Across the Batch
Hanna Foerster, Ilia Shumailov, Cheng Zhang +3
Dynamic quantization emerged as a practical approach to increase the utilization and efficiency of the machine learning serving flow. Unlike static quantization, which applies quan…
Team of Thoughts: Efficient Test-time Scaling of Agentic Systems through Orchestrated Tool Calling
Jeffrey T. H. Wong, Zixi Zhang, Junyi Liu +1
Existing Multi-Agent Systems (MAS) typically rely on homogeneous model configurations, failing to exploit the diverse expertise inherent in different post-trained architectures. We…
KernelCraft: Benchmarking for Agentic Close-to-Metal Kernel Generation on Emerging Hardware
Jiayi Nie, Haoran Wu, Yao Lai +9
New AI accelerators with novel instruction set architectures (ISAs) often require developers to manually craft low-level kernels, a time-consuming and error-prone process that does…
Deep Kernel Fusion for Transformers
Zixi Zhang, Zhiwen Mo, Yiren Zhao +1
Agentic LLM inference with long contexts is increasingly limited by memory bandwidth rather than compute. In this setting, SwiGLU MLP blocks, whose large weights exceed cache capac…
CaMeLs Can Use Computers Too: System-level Security for Computer Use Agents
Hanna Foerster, Tom Blanchard, Kristina Nikolić +6
AI agents are vulnerable to prompt injection attacks, where malicious content hijacks agent behavior. Among proposed defenses, architectural isolation provides the strongest guaran…
On the Existence and Behavior of Secondary Attention Sinks
Jeffrey T. H. Wong, Cheng Zhang, Louis Mahon +3
Attention sinks are tokens, often the beginning-of-sequence (BOS) token, that receive disproportionately high attention despite limited semantic relevance. In this work, we identif…
Reasoning Introduces New Poisoning Attacks Yet Makes Them More Complicated
Hanna Foerster, Ilia Shumailov, Yiren Zhao +4
Early research into data poisoning attacks against Large Language Models (LLMs) demonstrated the ease with which backdoors could be injected. More recent LLMs add step-by-step reas…
Mixture of Weight-shared Heterogeneous Group Attention Experts for Dynamic Token-wise KV Optimization
Guanghui Song, Dongping Liao, Yiren Zhao +3
Transformer models face scalability challenges in causal language modeling (CLM) due to inefficient memory allocation for growing key-value (KV) caches, which strains compute and s…
A3 : an Analytical Low-Rank Approximation Framework for Attention
Jeffrey T. H. Wong, Cheng Zhang, Xinye Cao +5
Large language models have demonstrated remarkable performance; however, their massive parameter counts make deployment highly expensive. Low-rank approximation offers a promising…
ARIES: Autonomous Reasoning with LLMs on Interactive Thought Graph Environments
Pedro Gimenes, Zeyu Cao, Jeffrey Wong +1
Recent research has shown that LLM performance on reasoning tasks can be enhanced by scaling test-time compute. One promising approach, particularly with decomposable problems, inv…
Cached Multi-Lora Composition for Multi-Concept Image Generation
Xiandong Zou, Mingzhu Shen, Christos-Savvas Bouganis +1
Low-Rank Adaptation (LoRA) has emerged as a widely adopted technique in text-to-image models, enabling precise rendering of multiple distinct elements, such as characters and style…
Omni-DNA: A Unified Genomic Foundation Model for Cross-Modal and Multi-Task Learning
Zehui Li, Vallijah Subasri, Yifei Shen +4
Large Language Models (LLMs) demonstrate remarkable generalizability across diverse tasks, yet genomic foundation models (GFMs) still require separate finetuning for each downstrea…
Refining Salience-Aware Sparse Fine-Tuning Strategies for Language Models
Xinxin Liu, Aaron Thomas, Cheng Zhang +3
Parameter-Efficient Fine-Tuning (PEFT) has gained prominence through low-rank adaptation methods like LoRA. In this paper, we focus on sparsity-based PEFT (SPEFT), which introduces…