LMMs Engine Documentation
==========================

Welcome to the LMMs Engine documentation! LMMs Engine is a flexible and extensible framework for training large multimodal models with support for various model architectures, datasets, and training strategies.

.. toctree::
   :maxdepth: 2
   :caption: Getting Started
   :glob:

   getting_started/index

.. toctree::
   :maxdepth: 1
   :caption: User Guide

   user_guide/datasets
   user_guide/data_prep
   user_guide/peak_perf
   user_guide/merge_fsdp
   user_guide/fsdp2_reduce_dtype
   user_guide/async_eval

.. toctree::
   :maxdepth: 2
   :caption: Tutorial

   tutorial/index

.. toctree::
   :maxdepth: 1
   :caption: Developer Guide

   developer_guide/new_dataset_guide
   developer_guide/new_processor_guide
   developer_guide/new_model_guide
   developer_guide/new_trainer_guide

.. toctree::
   :maxdepth: 1
   :caption: Reference

   reference/api
   reference/design_principle
   reference/video_configuration
   reference/mfu

.. toctree::
   :maxdepth: 1
   :caption: Models

   models/bagel
   models/qwenvl
   models/dllm
   models/fla_models
   models/rae_siglip
   models/sit
   models/wanvideo
   models/qwen2_llm
   models/qwen2_5_omni
   models/qwen3_vl_moe
   models/qwen3_moe
   models/qwen3_omni_moe

Indices and tables
==================

* :ref:`genindex`
* :ref:`modindex`
* :ref:`search`