Alexandre Drouin

Adjunct Professor, Laval university

2 papers at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 4

Thursday, December 4, 2025 · 4:30 PM → 7:30 PM

How to Train Your LLM Web Agent: A Statistical Diagnosis

#407 · Dheeraj Vattikonda, Santhoshi Ravichandran, Emiliano Penaloza, Hadi Nekoei, Thibault Le Sellier de Chezelles, Megh Thakkar, Nicolas Gontier, Miguel Muñoz-Mármol, Sahar Omidi Shayegan, Stefania Raimondo, Xue Liu, Alexandre Drouin, Alexandre Piché, Alexandre Lacoste, Massimo Caccia

We provide a statistically rigorous guidelines for training interactive, multi-step LLM agents, exploring optimal compute allocation, generalization, and hyperparameter settings.

Poster Session 6

1 paper

Friday, December 5, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Causal Differentiating Concepts: Interpreting LM Behavior via Causal Representation Learning

#2603 Spotlight · Navita Goyal, Hal Daumé III, Alexandre Drouin, Dhanya Sridhar

This paper introduces an unsupervised method that disentangles interpretable latent concepts in language model activations that mediate behavior, assuming that sparse changes in these concepts can induce changes in model behavior.