9 papers
Pre-training interventions, ex post facto: Grafting model beliefs across checkpoints
Peter Nutter, Dani Roytburg, Clément Dumas +2
Pre-training interventions are critical to alignment research, since beliefs formed during pre-training shape how a model generalizes from later training. One recently popular tech…
Diff Mining: Logit Differences Reveal Finetuning Objectives
Greg Kocher, Robert West, Clément Dumas +1
Finetuning has become the gold standard for refining existing behaviors and inducing new ones in language models, yet it often remains unclear exactly which behaviors emerge during…
Persona Cartography: Charting Language Model Personality Traits in Weight Space
Luke Baines, Anton Gonzalvez Hawthorne, Mariia Koroliuk +4
Large language models exhibit recurring behavioural patterns -- personas -- that shape generalisation and safety, but we lack reliable tools for decomposing, measuring, and control…
Activation Oracles: Training and Evaluating LLMs as General-Purpose Activation Explainers
Adam Karvonen, James Chua, Clément Dumas +8
Large language model (LLM) activations are notoriously difficult to understand, with most existing techniques using complex, specialized methods for interpreting them. Recent work…
nnterp: A Standardized Interface for Mechanistic Interpretability of Transformers
Clément Dumas
Mechanistic interpretability research requires reliable tools for analyzing transformer internals across diverse architectures. Current approaches face a fundamental tradeoff: cust…
Believe It or Not: How Deeply do LLMs Believe Implanted Facts?
Stewart Slocum, Julian Minder, Clément Dumas +4
Knowledge editing techniques promise to implant new factual knowledge into large language models (LLMs). But do LLMs really believe these facts? We develop a framework to measure b…
Narrow Finetuning Leaves Clearly Readable Traces in Activation Differences
Julian Minder, Clément Dumas, Stewart Slocum +4
Finetuning on narrow domains has become an essential tool to adapt Large Language Models (LLMs) to specific tasks and to create models with known unusual properties that are useful…
Overcoming Sparsity Artifacts in Crosscoders to Interpret Chat-Tuning
Julian Minder, Clément Dumas, Caden Juang +2
Model diffing is the study of how fine-tuning changes a model's representations and internal algorithms. Many behaviors of interest are introduced during fine-tuning, and model dif…
Separating Tongue from Thought: Activation Patching Reveals Language-Agnostic Concept Representations in Transformers
Clément Dumas, Chris Wendler, Veniamin Veselovsky +2
A central question in multilingual language modeling is whether large language models (LLMs) develop a universal concept representation, disentangled from specific languages. In th…