https://developers.googleblog.com/maxtext-expands-post-training-capabilities-introducing-sft-and-rl-on-single-host-tpus/
MaxText Expands Post-Training Capabilities: Introducing SFT and RL on Single-Host TPUs - Google...
MaxText now supports Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) on single-host TPUs. Leverage JAX-based efficiency and advanced algorithms...
post trainingmaxtextexpandscapabilitiesintroducing