Megatron is a large, powerful transformer developed by the Applied Deep Learning Research team at NVIDIA. This repository is for ongoing research on training large transformer language models at scale. We developed efficient, model-parallel (tensor, sequence, and pipeline), and multi-node pre-training of transformer based models such as GPT, BERT, and T5 using mixed precision.

Megatron is also used in NeMo Megatron, a framework to help enterprises overcome the challenges of building and training sophisticated natural language processing models with billions and trillions of parameters.

Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.

Project Samples

Project Activity

See All Activity >

Follow Megatron

Megatron Web Site

Other Useful Business Software
Go from Code to Production URL in Seconds Icon
Go from Code to Production URL in Seconds

Cloud Run deploys apps in any language instantly. Scales to zero. Pay only when code runs.

Skip the Kubernetes configs. Cloud Run handles HTTPS, scaling, and infrastructure automatically. Two million requests free per month.
Try it free
Rate This Project
Login To Rate This Project

User Reviews

Be the first to post a review of Megatron!

Additional Project Details

Programming Language

Python

Related Categories

Python Large Language Models (LLM), Python Deep Learning Frameworks, Python Generative AI

Registered

2023-03-23