Together AI launches custom SFT and RL training for open models

Together AI launched Together Post-Training on October 19, a managed service that lets developers customize open models with supervised fine-tuning (SFT) or reinforcement learning (RL), CEO Vipul Ved Prakash said. Developers define their data, rewards, environments and training loop while Together runs the GPU infrastructure and deployment, and they keep the resulting weights. The service extends Together's AI cloud from model access and inference into custom training, letting the company sell infrastructure and deployment around each experiment.

Together AI launches custom SFT and RL training for open models The service lets developers set the training loop and keep the resulting weights, while Together supplies tuned GPU infrastructure and deployment. By Ryan Merket https://runtimewire.com/author/ryan-merket · Published Primary source: X https://x.com/togethercompute/status/2112241595273621857 Why it matters Together is extending its AI cloud from model access and inference into custom training. Developers get control of training recipes and downloadable weights; Together can also sell the infrastructure and deployment around each experiment. Together AI CEO Vipul Ved Prakash https://x.com/vipulved said the company launched Together Post-Training on October 19th, opening a managed service for developers to customize open models with supervised fine-tuning SFT or reinforcement learning https://runtimewire.com/models/huggingface/lobonexequiel-reinforcement-learning-daff1a058d32e2ca RL . In a thread on X https://x.com/togethercompute/status/2112241595273621857 , Together said developers can define their data, rewards, environments and training loop while it runs the underlying infrastructure.… When you purchase through links in our articles, we may earn a small commission https://runtimewire.com/affiliate-monetization-standards . This doesn’t affect our editorial independence.