Lausanne, Switzerland
Arthur Renard
Founding engineer and ML researcher at Xent Labs, working on reinforcement learning for self-improving language models.
Selected work
View all →Cross-Entropy Games and Frost Training
Preprint, 2026A method that uses reward gradients in embedding space to accelerate GRPO training on Cross-Entropy Games.
Cognitive Training for Language Models
Preprint, 2026A principled framework for relevant skill discovery in language models via Cross-Entropy Games.
Boolformer
AI for Math Workshop, ICML 2025A Transformer model for symbolic regression of Boolean functions, presented at the AI for Math Workshop at ICML 2025.