Facebookresearch

Foundational Models for State-of-the-Art Speech and Text Translation

Source Code for Paper "OrienterNet Visual Localization in 2D Public Maps with Neural Matching"

Matrix (Multi-Agent daTa geneRation Infra and eXperimentation framework) is a versatile engine for multi-agent conversational data generation.

A general physic-based retargeting framework.

This repository provides the code for implementing RPG described in our KDD'25 paper "Generating Long Semantic IDs in Parallel for Recommendation".

Understanding Training Dynamics of Deep ReLU Networks

Benchmark and research code for the paper SWEET-RL Training Multi-Turn LLM Agents onCollaborative Reasoning Tasks

A modular embodied agent architecture and platform for building embodied agents

Large Concept Models: Language modeling in a sentence representation space

A flexible, high-performance 3D simulator for Embodied AI research.

A library for differentiable nonlinear optimization

This repository contains the code to train and evaluate TRIBE v2, a multimodal model for brain response prediction

Habitat-Lab: a modular high-level library for end-to-end development in Embodied AI.

PyTorch extensions for high performance and large scale training.

A Python framework for AI-driven character animation using neural networks.

State-of-the-art Image & Video CLIP, Multimodal Large Language Models, and More!

Meta Lingua: a lean, efficient, and easy-to-hack codebase to research LLMs.

Training Large Language Model to Reason in a Continuous Latent Space

Fairseq is a sequence modeling toolkit for training custom models in text generation tasks.

An Analysis Toolkit for Natural Language Generation (Translation, Captioning, Summarization, etc.)

SONAR, a new multilingual and multimodal fixed-size sentence embedding space, with a full suite of speech and text encoders and decoders.

FAIR Sequence Modeling Toolkit 2

Code repository for emg2pose dataset and model benchmarks

Official implementation of paper "VLM³: Vision Language Models Are Native 3D Learners".

visualization code for 3D human body annotation by EFT (Exemplar Fine-tuning)

GPU Cluster Monitoring (GCM): Large-Scale AI Research Cluster Monitoring

Watermarking and detection for speech audios

🎬ActionMesh: A fast video to animated mesh model with unprecedented quality. Generate animated mesh seamlessly importable into any 3D software in less than a minute.

Repository for the CVPR 2026 paper MeshFlow Efficient Artistic Mesh Generation via MeshVAE and Flow-based Diffusion Transformer by Weiyu Li, Antoine Toisoul, Tom Monnier, Roman Shapovalov, Rakesh Ranjan, Ping Tan and Andrea Vedaldi.

Superintelligent Retrieval Agent (SIRA)

CoTracker is a model for tracking any point (pixel) on a video.

PySlowFast: video understanding codebase from FAIR for reproducing state-of-the-art video models.

A library for efficient similarity search and clustering of dense vectors.

Kats is a Python toolkit for analyzing time series data, offering tools for detection, forecasting, and feature extraction.

A python library that provides common I/O interface across different storage backends.

Geometric Retargeting A Principled, Ultrafast Neural Hand Retargeting Algorithm

Comprehensive benchmark for RAG

Repository for research in the field of Responsible NLP at Meta.

[CVPR 2025] Official PyTorch implementation of "EdgeTAM: On-Device Track Anything Model"

Mobile vision models and code

Simple yet SoTA Knowledge Graph Embeddings.

Execution and caching tool for python

Scalable and Performant Data Loading

projectaria_tools is an C++/Python open-source toolkit to interact with Project Aria data.

Demo utilities for running EgoBlur person and license plate blurring models.

Open and efficient video and image watermarking

Official implementation of the paper "The Stable Signature Rooting Watermarks in Latent Diffusion Models"

State-of-the-Art AI Watermarking

Dr. Zero Self-Evolving Search Agents without Training Data

PArametrized Recommendation and Ai Model benchmark is a repository for development of numerous uBenchmarks as well as end to end nets for evaluation of training and inference platforms.

AssemblyHands Toolkit is a Python package that provides data loader, visualization, and evaluation tools for the AssemblyHands dataset (CVPR 2023).

Drivable Volumetric Avatars

Train universal codec avatars

Code repo for the paper "SpinQuant LLM quantization with learned rotations"

This repository contains the training code of ParetoQ introduced in our work "ParetoQ Scaling Laws in Extremely Low-bit LLM Quantization"

[CVPR 2026] Multi-SpatialMLLM: Multi-Frame Spatial Understanding with Multi-Modal Large Language Models

Code repo for the paper "LLM-QAT Data-Free Quantization Aware Training for Large Language Models"

Code for "LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding", ACL 2024

The Automated LLM Speedrunning Benchmark measures how well LLM agents can reproduce previous innovations and discover new ones in language modeling.

AIRS-Bench: an AI Research Science benchmark for quantifying the end-to-end AI research abilities of LLM agents

Official repository for "Revisiting Weakly Supervised Pre-Training of Visual Perception Models". https://arxiv.org/abs/2201.08371.

Large-Scale Translation Data Mining.

[CVPR 2026] Pixio: a capable vision encoder dedicated to dense prediction, simply by pixel reconstruction

Ocean is the in-house framework for Computer Vision (CV) and Augmented Reality (AR) applications at Meta. It is platform independent and is mainly implemented in C/C++.

Memory layers use a trainable key-value lookup mechanism to add extra parameters to a model without increasing FLOPs. Conceptually, sparsely activated memory layers complement compute-heavy dense feed-forward layers, providing dedicated capacity to store and retrieve…

Ego4d dataset repository. Download the dataset, visualize, extract features & example usage of the dataset

Filtering, Distillation, and Hard Negatives for Vision-Language Pre-Training

Detectron2 is a platform for object detection, segmentation and other visual recognition tasks.

Code release for "Cut and Learn for Unsupervised Object Detection and Instance Segmentation" and "VideoCutLER: Surprisingly Simple Unsupervised Video Instance Segmentation"

Large dataset of hand-object contact, hand- and object-pose, and 2.9 M RGB-D grasp images.