google-deepmind

Multi-Joint dynamics with Contact. A general purpose physics simulator.

This repository contains implementations and illustrative code to accompany DeepMind publications

Open source code for AlphaFold 2.

TensorFlow-based neural network library

AlphaFold 3 inference pipeline.

StarCraft II Learning Environment

A customisable 3D platform for agent-based AI research

Gemma open-weight LLM library, from Google DeepMind

OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games.

Build Graph Nets in Tensorflow

Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo.

A collection of high-quality models for the MuJoCo physics engine, curated by Google DeepMind.

A library of reinforcement learning components and agents

Learning to Learn in TensorFlow

GDM Science Skills to speed up agentic scientific workflows with better grounding and higher token efficiency. Integrate insights from AlphaGenome, AFDB, UniProt and 30+ other databases and tools.

JAX-based neural network library

TensorFlow Reinforcement Learning

Monte Carlo tree search in JAX

A TensorFlow implementation of the Differentiable Neural Computer.

Optax is a gradient processing and optimization library for JAX.

An open-source library for sim and real robot demos on top of MuJoCo.

Tracking Any Point (TAP)

This API provides programmatic access to the AlphaGenome model developed by Google DeepMind.

This dataset code generates mathematical question and answer pairs, from a range of question types at roughly school-level difficulty.

A JAX research toolkit for building, editing, and visualizing neural networks.

A library for generative social simulation

Convolutional neural network model for video classification trained on the Kinetics dataset.

Real-time behaviour synthesis with MuJoCo, using Predictive Control

GPU-optimized version of the MuJoCo physics simulator, designed for NVIDIA hardware.

bsuite is a collection of carefully-designed experiments that investigate core capabilities of a reinforcement learning (RL) agent

A Graph Neural Network Library in Jax

A collection of formalized statements of conjectures in Lean.

Question answering dataset featured in "Teaching Machines to Read and Comprehend

ML research on Android devices.

Dramatron uses large language models to generate coherent scripts and screenplays.

A TensorFlow implementation of Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures.

tree is a library for working with nested data structures

This repository contains notebook implementations of the following Neural Process variants: Conditional Neural Processes (CNPs), Neural Processes (NPs), Attentive Neural Processes (ANPs).

A platform for managing machine learning experiments

Research code accompanying AlphaGenome

A suite of test scenarios for multi-agent reinforcement learning.

An implementation of the Fermionic Neural Network for ab-initio electronic structure calculations

Reverb is an efficient and easy-to-use data storage and transport system designed for machine learning research

official code for "Large Language Models as Optimizers"

TORAX: Tokamak transport simulation in JAX

Accompanying code for "Discovering State-of-the-art Reinforcement Algorithms" Nature publication

Benchmarking long-form factuality in large language models. Original code for our paper "Long-form factuality in large language models".

Open weights language model from Google DeepMind, based on Griffin.

Dermatology ddx dataset, Jax implementations of Monte Carlo conformal prediction, plausibility regions and statistical annotation aggregation from our recent work on uncertain ground truth...

On the Theoretical Limitations of Embedding-Based Retrieval

hanabi_learning_environment is a research platform for Hanabi experiments.

Grandmaster-Level Chess Without Search

A highly-customisable gridworld game engine with some batteries included. Make your own gridworld games to test reinforcement learning agents!

TIPSv2 (CVPR'26) and TIPS (ICLR'25)

Lua/Torch implementation of DQN (Nature, 2015)

This is a suite of reinforcement learning environments illustrating various safety properties of intelligent agents.

Library to compute surface distance based performance metrics for segmentation tasks.

Minimal and scalable research codebase in JAX, designed for rapid iteration on frontier research in LLM and other autoregressive models.

Restoring and attributing ancient texts using deep neural networks

Dataset to assess the disentanglement properties of unsupervised learning methods

This repository contains the NarrativeQA dataset. It includes the list of documents with Wikipedia summaries, links to full stories, and questions and answers.

DQN Zoo is a collection of reference implementations of reinforcement learning agents developed at DeepMind based on the Deep Q-Network (DQN) agent.

An interactive HTML pretty-printer for machine learning research in IPython notebooks.

Build AI text detectors whose decisions are mechanistically grounded and human-auditable (EMNLP 2026)

PIX is an image processing library in JAX, for JAX.

A customisable 2D platform for agent-based AI research

Foundation models for 4D spatial and temporal vision tasks.

Libraries, tools and tasks created and used at DeepMind Robotics.

A Python interface for reinforcement learning environments

Official repository for "VideoPrism: A Foundational Visual Encoder for Video Understanding" (ICML 2024)

Benchmarking physical understanding in generative video models

Barkour Robot: Agile Quadruped Robots by Google DeepMind

Spriteworld: a flexible, configurable python-based reinforcement learning environment

Library for sequence-to-sequence numeric prediction, applicable to any tokenizable input, and allows pretraining and fine-tuning over multiple tasks.

A algebraic word problem dataset, with multiple choice questions annotated with rationales.

We provide a pre-trained model for unconditional 19-step generation of CelebA-HQ images

Second Order Optimization and Curvature Estimation with K-FAC in JAX.

A C++/Python implementation of the StreetLearn environment based on images from Street View, as well as a TensorFlow implementation of goal-driven navigation agents solving the task published in...