5 citations · 5 across the 11 of their papers we have counts for
14 papers
Settle: Learning When to Stop Reasoning
Ryan Brown, Zihao Fu, Chris Russell
Reasoning models often continue generating after their answers have settled. Settle learns when to stop from answer stability in completed traces. It trains the existing end-of-rea…
Decoupling Token Roles in Autoregressive Pretraining
Suqin Yuan, Runqi Lin, Kevin Qinghong Lin +4
Autoregressive pretraining increasingly draws on heterogeneous data, making it important to understand how a model learns from an individual token. The next-token prediction object…
SEES: A Self-Evolving Embodied System via Failure-Guided VLA Policy Adaptation
Ziwen Li, Hanlue Zhang, Zhenyang Ren +9
Recent vision-language-action (VLA) policies demonstrate promising generalization across diverse short-horizon tasks. However, they remain unreliable on long-horizon tasks, partly…
Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference Alignment
Suqin Yuan, Runqi Lin, Muyang Li +5
Human-feedback alignment has made language models useful assistants and is commonly described as aligning them with humans. However, the responses people prefer from an AI need not…
Running the Gauntlet: Hard Agentic Tasks
Mykola Vysotskyi, Runqi Lin, Grzegorz Biziel +21
As agentic systems continue to evolve and are widely deployed in real-world scenarios, there is a growing demand to faithfully evaluate their capabilities. However, current benchma…
Grounding Text Embeddings in Stakeholder Associations
Jonathan Rystrøm, Sofie Burgos-Thorsen, Zihao Fu +3
Text embeddings are widely used to analyse large corpora of complex texts. However, it is unclear whether the embeddings capture the same semantic distances as the human experts us…
Known By Their Actions: Fingerprinting LLM Browser Agents via UI Traces
William Lugoloobi, Samuelle Marro, Jabez Magomere +2
As LLM-based agents increasingly browse the web on users' behalf, a natural question arises: can websites passively identify which underlying model powers an agent? Doing so would…
Evaluating the Ability of Explanations to Disambiguate Models in a Rashomon Set
Kaivalya Rawal, Eoin Delaney, Zihao Fu +2
Explainable artificial intelligence (XAI) is concerned with producing explanations indicating the inner workings of models. For a Rashomon set of similarly performing models, expla…
OxEnsemble: Fair Ensembles for Low-Data Classification
Jonathan Rystrøm, Zihao Fu, Chris Russell
We address the problem of fair classification in settings where data is scarce and unbalanced across demographic groups. Such low-data regimes are common in domains like medical im…
FairImagen: Post-Processing for Bias Mitigation in Text-to-Image Models
Zihao Fu, Ryan Brown, Shun Shao +3
Text-to-image diffusion models, such as Stable Diffusion, have demonstrated remarkable capabilities in generating high-quality and diverse images from natural language prompts. How…
CAST: Compositional Analysis via Spectral Tracking for Understanding Transformer Layer Functions
Zihao Fu, Ming Liao, Chris Russell +1
Large language models have achieved remarkable success but remain largely black boxes with poorly understood internal mechanisms. To address this limitation, many researchers have…
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…
Multi-use LLM Watermarking and the False Detection Problem
Zihao Fu, Chris Russell
Digital watermarking is a promising solution for mitigating some of the risks arising from the misuse of automatically generated text. These approaches either embed non-specific wa…
Evaluating Model Explanations without Ground Truth
Kaivalya Rawal, Zihao Fu, Eoin Delaney +1
There can be many competing and contradictory explanations for a single model prediction, making it difficult to select which one to use. Current explanation evaluation frameworks…