megatron-energon

Introduction

  • General
  • Installation

Basic Usage

  • Quickstart
  • Data Preparation
  • Data Flow
  • Task Encoder
  • Metadataset
  • Save and Restore

Advanced Usage

  • Remote Dataset
  • Custom Sample Loader
  • Crude Data and How to Cook It 👨‍🍳
  • Reproducible Scaling
  • Packing
  • Grouping
  • Joining Datasets
  • Epochized Blending
  • Customized Blending
  • Parallelism

API

  • Packages and Modules
  • Command-Line Interface

Internals

  • Contribution Guidelines
  • Code Structure
megatron-energon
  • Search

© Copyright 2025 NVIDIA Corporation.

Built with Sphinx using a theme provided by Read the Docs.