# Training Foundation Models with a Single Algorithm

> Source: <https://aiflash.com/news/142544/>
> Published: 2026-10-30 15:30:09+00:00

Foundation models are typically trained in stages: pretraining, midtraining, supervised fine-tuning (SFT), and reinforcement learning (RL), with changes in both data and learning algorithms. We show that this algorithmic separation is not fundamental. By deriving suitable RL objectives as maximum-li