07:20
2026-06-23
marktechpost.com
machine-learning
Prime Intellect Releases prime-rl 0.6.0 to Train Trillion-Parameter MoE Models on Agentic RL Workloads
Prime Intellect released prime-rl 0.6.0, an open framework for asynchronous reinforcement learning on trillion-parameter Mixture-of-Experts models, enabling training on agentic RL workloads with optim…