Ambroise Odonnat

Multi-path reasoning on a budget: towards theoretically optimal hyperparameter-free adaptive self-consistency

Which transformers components are the most sensitive to distribution shifts?

Zero-shot Model-based Reinforcement Learning using Large Language Models

DA-Bench: Benchmarking Unsupervised Domain Adaptation Methods with Realistic Validation On Diverse Modalities

Large Language Models as Markov Chains

A Visual Case Study of the Training Dynamics in Neural Networks