NVIDIA-NeMo
NVIDIA NeMo Framework Overview
NeMo Framework is NVIDIA's GPU accelerated, fully open-source, end-to-end training framework for large language models (LLMs), multi-modal models, diffusion and speech models. It enables seamless scaling of pretraining, post-training, and reinforcement learning workloads from single GPU to thousand-node clusters for both 🤗Hugging Face/PyTorch and Megatron models. This GitHub organization includes a suite of libraries and recipe collections to help users train models from end to end.
NeMo Framework is also a part of the NVIDIA NeMo software suite for managing the AI agent lifecycle.
Latest 📣 announcements and 🗣️ discussions
💬 NeMo Assist
[4/13/2026]
Try NeMo Assist to chat with the NeMo Framework docs and code.
Getting Started
InstallationCheckpoint Conversion HF<>MegatronLLM example recipes and scriptsVLM example recipes and scripts1 ～ 1,000 GPUs
,
No Need
,
,
,
,
,
,
Over 1,000 GPUs
,
,
with
,
with
,
Repo organization under NeMo Framework
Summary of key functionalities and container strategy of each repo
Visit the individual repos to find out more 🔍, raise 🐛, contribute ✍️ and participate in discussion forums 🗣️!
Note: The NeMo Framework is currently in the process of restructuring. The original NeMo 2.0 repository will now focus specifically on speech-related components, while other parts of the framework are being modularized into separate libraries such as NeMo Automodel, NeMo Gym, NeMo RL, and more. This transition aims to make NeMo more modular and developer-friendly.
RepoKey Functionality & Documentation LinkTraining LoopTraining BackendsInference BackendsModel CoverageContainer
PyT native loopMegatron-coreNALLM & VLMNeMo Framework Container
PyT native loopPyTorchNALLM, VLM, Omni, VFMNeMo AutoModel Container
PyTorch Lightning LoopMegatron-core & PyTorchRIVASpeechNA
PyT native loopMegatron-core & PyTorchvLLMLLM, VLMNeMo RL container
RL Environment, Eval, Sandboxing
NANANANANeMo RL Container (WIP)
NANANAAgnosticNeMo Curator Container
NANANANA
NANAAgnosticNeMo Framework Container
NANAvLLM, TRT, TRTLLM, ONNXAgnosticNeMo Framework Container
NANANAAgnosticNeMo Framework Container
Reference pipeline for SDG & Eval
NANANAAgnosticNA
NAAgnosticNANANA
Developer asset hub for Nemotron modelsNANANANemotron modelsNA
Synthetic data generation library
NANANANANA
Privacy-preserving synthetic tabular data
HF TrainerPyTorch, OpacusvLLMLLMNA
PII detection and anonymization
NANANAAgnosticNA
Model routing, API translation & observability (OpenAI / Anthropic)
NANAAgnostic (OpenAI-compatible)AgnosticNA Table 1. NeMo Framework Repos
Diagram Ilustration of Repos under NeMo Framework (WIP)
Figure 1. NeMo Framework Repo Overview
Some background motivations and historical contexts
The NeMo GitHub Org and its repo collections are created to address the following problems
Need for composability: The
is monolithic and encompasses too many things, making it hard for users to find what they need. Container size is also an issue. Breaking down the Monolithic repo into a series of functional-focused repos to facilitate code discovery.
Need for customizability: The
uses PyTorch Lighting as the default trainer loop, which provides some out of the box functionality but making it hard to customize.
,
, and
have adopted pytorch native custom loop to improve flexibility and ease of use for developers.
License
Apache 2.0 licensed with third-party attributions documented in each repository.