Ongoing research training transformer models at scale
This is an exact mirror of the Megatron project, hosted at https://github.com/NVIDIA/Megatron-LM.
SourceForge is not affiliated with Megatron.
Megatron is a trademark of NVIDIA CORPORATION.
Host and run your applications without the need to manage infrastructure. Scales up from and down to zero automatically.
Cloud Run is the fastest way to deploy containerized apps. Push your code in Go, Python, Node.js, Java, or any language and Cloud Run builds and deploys it automatically. Get fast autoscaling, pay only when your code runs, and skip the infrastructure headaches. Two million requests free per month. And new customers get $300 in free credit.
Try Cloud Run Free
Go From Idea to Deployed AI App Fast
One platform to build, fine-tune, and deploy. No MLOps team required.
Access Gemini 3 and 200+ models. Build chatbots, agents, or custom models with built-in monitoring and scaling.