6 papers
Do Linear Probes Generalize Better in Persona Coordinates?
Prasad Mahadik, Adrians Skapars
It is becoming increasingly necessary to have monitors check for harmful behaviors during language model interactions, but text-only monitoring has not been sufficient. This is bec…
The Impact of Off-Policy Training Data on Probe Generalisation
Nathalie Kirch, Samuel Dower, Adrians Skapars +3
Probing has emerged as a promising method for monitoring large language models (LLMs), enabling cheap inference-time detection of concerning behaviours. However, natural examples o…
GPT, But Backwards: Exactly Inverting Language Model Outputs
Adrians Skapars, Edoardo Manino, Youcheng Sun +1
The task of reconstructing unknown textual inputs to language models is a fundamental auditing primitive that allows us to assess the model's vulnerability to a range of security i…
Worst-Case Symbolic Constraints Analysis and Generalisation with Large Language Models
Daniel Koh, Yannic Noller, Corina S. Pasareanu +2
Large language models (LLMs) have demonstrated strong performance on coding tasks such as generation, completion and repair, but their ability to handle complex symbolic reasoning…
Transformers Use Causal World Models in Maze-Solving Tasks
Alex F. Spies, William Edwards, Michael I. Ivanitskiy +5
Recent studies in interpretability have explored the inner workings of transformer models trained on tasks across various domains, often discovering that these networks naturally d…
SAFLITE: Fuzzing Autonomous Systems via Large Language Models
Taohong Zhu, Adrians Skapars, Fardeen Mackenzie +4
Fuzz testing effectively uncovers software vulnerabilities; however, it faces challenges with Autonomous Systems (AS) due to their vast search spaces and complex state spaces, whic…