Theo Lasnier's picture

2 1

Theo Lasnier

Blyzi

·

https://blyzi.github.io/

AI & ML interests

AI Interpretability

Recent Activity

authored a paper 1 day ago

Language-Switching Triggers Take a Latent Detour Through Language Models

updated a model 11 days ago

Blyzi/trigger-models

published a model 11 days ago

Blyzi/trigger-models

View all activity

Organizations

authored a paper 1 day ago

Language-Switching Triggers Take a Latent Detour Through Language Models

Paper • 2605.18646 • Published 4 days ago • 3

authored 2 papers 22 days ago

Disentangling meaning from language in LLM-based machine translation

Paper • 2602.04613 • Published Feb 4 • 1

Triggers Hijack Language Circuits: A Mechanistic Analysis of Backdoor Behaviors in Large Language Models

Paper • 2602.10382 • Published Feb 12 • 2