Ctrl+K

You are viewing the NeMo 2.0 documentation. This release introduces significant changes to the API and a new library, NeMo Run. We are currently porting all features from NeMo 1.0 to 2.0. For documentation on previous versions or features not yet available in 2.0, please refer to the NeMo 24.07 documentation.

NVIDIA NeMo Framework User Guide

Table of Contents

NeMo Framework

Overview
Install NeMo Framework
Performance
Why NeMo Framework?

Releases

Software Component Versions
Changelog
Known Issues

Getting Started

Quickstart with NeMo-Run
Quickstart with NeMo 2.0 API
Tutorials

Developer Guides

Migration Guide
Feature Guide
Best Practices
Performance Tuning Guide

Training and Customization

Long Context Training
- Context Parallelism
Optimal Configuration with Auto Configurator
Parameter-Efficient Fine-tuning (PEFT)
- Supported PEFT Methods
- A Comparison of Performant and Canonical LoRA Variants
Sequence Packing
Resiliency
Continual Training
Custom Datasets
- Pre-Training Data Module
- Fine-Tuning Data Module

Model Optimization

Quantization
Pruning
Distillation
Speculative Decoding

Models

Large Language Models
- Baichuan 2
- ChatGLM 3
- DeepSeek V2
- DeepSeek V3
- Gemma
- Gemma 2
- GPT-OSS
- Hyena
- Llama 3
- Llama Nemotron
- Mamba 2
- Mixtral
- Nemotron
- Phi 3
- Qwen2/2.5
- Qwen3
- Starcoder
- Starcoder 2
- T5
- BERT
Vision Language Models
Speech AI Models
Diffusion Models
- Flux
- Diffusion Training Framework
Embedding Models
Reranker Models
- Llama Reranker

Library Documentation

Overview
NeMo
NeMo AutoModel
NeMo Curator
NeMo Eval
NeMo Export and Deploy
NeMo Megatron Bridge
NeMo RL
NeMo Run

Search

Ctrl+K

Last updated on Oct 24, 2025.