@RL

MaxText has introduced new post-training capabilities, specifically Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), now available on single-host TPU configurations like v5p-8 and v6e-8. …