Announcement
Together launches Post-Training service for custom AI models
Together says developers can define training loops using reinforcement learning or supervised fine-tuning while it handles the infrastructure.
TLDR
Together introduced Post-Training for customizing AI models with reinforcement learning or supervised fine-tuning. Developers can define their training loops and run multiple LoRA adapter experiments on the same GPU capacity; full-weight experiments are available for select models. Together says it handles the underlying infrastructure and supports serving multiple trained adapters in one GPU deployment for comparison.
Combined views
5.4K
4 Sources, first seen ago
