A new arXiv paper introduces Learning to Coach (L2C), a framework that trains a separate LLM acting as a coach to pull actionable guidance out of experience. The motivation is that raw solution trajectories are typically long and noisy, which limits how well language models can learn from them. The approach aims to convert such trajectories into more useful, condensed coaching signals.
An arXiv paper reports that reinforcement learning does not lift large language model performance evenly across a dataset. Gains are large on problems the model can already solve and much smaller on difficult ones, a pattern the authors call the Matthew Effect. The finding suggests current RL training methods may widen the gap between easy and hard tasks.
A new arXiv paper examines the "job power elasticity" of large language model training, looking at how much these workloads can adjust their electricity consumption. Power supply is described as a key constraint on the growth of AI data centers, which are among the fastest-rising sources of electricity demand. The work aims to characterize how training jobs could respond to power limits.