2 citations · 4 across the 10 of their papers we have counts for
11 papers
Agents' Last Exam
Yiyou Sun, Xinyang Han, Weichen Zhang +378
Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional d…
RepSelect: Robust LLM Unlearning via Representation Selectivity
Filip Sondej, Yushi Yang, Adam Mahdi
When LLM weights are open or fine-tuning is available through an API, suppressing hazardous knowledge and tendencies is not enough: removal has to be deep enough that an adversary…
It's a TRAP! Task-Redirecting Agent Persuasion Benchmark for Web Agents
Karolina Korgul, Yushi Yang, Arkadiusz Drohomirecki +7
Web-based agents powered by large language models are increasingly used for tasks such as email management or professional networking. Their reliance on dynamic web content, howeve…
Measuring what Matters: Construct Validity in Large Language Model Benchmarks
Andrew M. Bean, Ryan Othniel Kearns, Angelika Romanou +39
Evaluating large language models (LLMs) is crucial for both assessing their capabilities and identifying safety or robustness issues prior to deployment. Reliably measuring abstrac…
Agentic Reinforcement Learning for Search Misaligns Instruction-Tuning
Yushi Yang, Shreyansh Padarha, Sarah Ball +2
Agentic reinforcement learning (RL) trains large language models to use tools, but its impact on alignment is poorly understood. We study how agentic RL for search affects the alig…
LLMs Don't Know Their Own Decision Boundaries: The Unreliability of Self-Generated Counterfactual Explanations
Harry Mayne, Ryan Othniel Kearns, Yushi Yang +4
To collaborate effectively with humans, language models must be able to explain their decisions in natural language. We study a specific type of self-explanation: self-generated co…
Collapse of Irrelevant Representations (CIR) Ensures Robust and Non-Disruptive LLM Unlearning
Filip Sondej, Yushi Yang
Current unlearning and safety training methods consistently fail to remove dangerous knowledge from language models. We identify the root cause - unlearning targets representations…
Robust LLM Unlearning with MUDMAN: Meta-Unlearning with Disruption Masking And Normalization
Filip Sondej, Yushi Yang, Mikołaj Kniejski +1
Language models can retain dangerous knowledge and skills even after extensive safety fine-tuning, posing both misuse and misalignment risks. Recent studies show that even speciali…
Can sparse autoencoders be used to decompose and interpret steering vectors?
Harry Mayne, Yushi Yang, Adam Mahdi
Steering vectors are a promising approach to control the behaviour of large language models. However, their underlying mechanisms remain poorly understood. While sparse autoencoder…
How Does DPO Reduce Toxicity? A Mechanistic Neuron-Level Analysis
Yushi Yang, Filip Sondej, Harry Mayne +2
Safety fine-tuning algorithms reduce harmful outputs in language models, yet their mechanisms remain under-explored. Direct Preference Optimization (DPO) is a popular choice of alg…
Evaluating Fine-Tuning Efficiency of Human-Inspired Learning Strategies in Medical Question Answering
Yushi Yang, Andrew M. Bean, Robert McCraith +1
Fine-tuning Large Language Models (LLMs) incurs considerable training costs, driving the need for data-efficient training with optimised data ordering. Human-inspired strategies of…