Brian Bartoldson

4 papers Lawrence Livermore National Laboratory

• Act Only When It Pays: Efficient Reinforcement Learning for LLM Reasoning via Selective Rollouts
• Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
• The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text
• Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Training