whatisgithub

What is mmf?

facebookresearch/mmf — explained in plain English

Analysis updated 2026-06-26

5,629PythonAudience · researcherComplexity · 4/5LicenseSetup · moderate

In one sentence

MMF is a Facebook AI research framework for building AI models that understand both images and text together, used for tasks like visual question answering and hateful content detection.

Mindmap

mindmap
  root((MMF))
    What it does
      Multimodal AI training
      Images plus text
      Research framework
    Built on
      PyTorch
      Modular components
      Distributed training
    Tasks supported
      Visual QA
      Image captioning
      Hateful content detection
    Use cases
      AI research
      Challenge baselines
      Custom experiments
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Train a visual question answering model that answers questions about images using MMF's pre-built components.

USE CASE 2

Run experiments on the Hateful Memes dataset using MMF as the official starter codebase.

USE CASE 3

Swap out individual components like model architecture or dataset loader to test new multimodal AI ideas.

USE CASE 4

Scale a multimodal AI training run across multiple GPUs or machines using MMF's distributed training support.

What is it built with?

PythonPyTorch

How does it compare?

facebookresearch/mmfgoogle/seq2seqlucidrains/dalle-pytorch
Stars5,6295,6295,629
LanguagePythonPythonPython
Setup difficultymoderatehardhard
Complexity4/54/55/5
Audienceresearcherresearcherresearcher

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 1h+

Full documentation and installation instructions are at mmf.sh, not in the repo, requires GPU for meaningful experiments.

Open source under a permissive license, check mmf.sh for the specific terms.

So what is it?

MMF is a research framework from Facebook AI Research for building and experimenting with AI models that work with both images and text at the same time. This area of research is called multimodal AI, because it combines multiple types of input (visual and language) rather than just one. For example, a multimodal model might answer questions about a photo, generate captions for images, or detect hateful content that pairs an image with text. The framework is built on top of PyTorch, a widely used AI development library. It is designed to be modular, meaning researchers can swap out individual components like the dataset loader, the model architecture, or the training loop without rewriting everything else. It also supports distributed training, which means it can spread work across multiple machines or graphics cards to handle large experiments faster. MMF includes reference implementations of several published research models, and it has been used internally at Facebook for a number of AI research projects. It also served as the official starter codebase for several public AI challenges including the Hateful Memes challenge and the TextVQA challenge, where teams compete to build better models for understanding text inside images. The project was previously called Pythia before being renamed to MMF. Installation instructions and full documentation live at mmf.sh rather than in the repository itself. The README is brief and points to the external documentation site for most setup and usage details.

Copy-paste prompts

Prompt 1
Using MMF, walk me through fine-tuning a pretrained visual question answering model on a custom image-question dataset.
Prompt 2
I want to run the Hateful Memes detection baseline using MMF. What models are included and how do I start training?
Prompt 3
How do I add a custom dataset to MMF so I can train a multimodal model on my own image-text pairs?
Prompt 4
I'm using MMF for a TextVQA experiment. How do I swap the model architecture while keeping the same training loop and evaluation metrics?
Prompt 5
What does MMF's modular design mean in practice, which components can I replace and how does the config system work?

Frequently asked questions

What is mmf?

MMF is a Facebook AI research framework for building AI models that understand both images and text together, used for tasks like visual question answering and hateful content detection.

What language is mmf written in?

Mainly Python. The stack also includes Python, PyTorch.

What license does mmf use?

Open source under a permissive license, check mmf.sh for the specific terms.

How hard is mmf to set up?

Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.

Who is mmf for?

Mainly researcher.

Open on GitHub → Ask about another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.