recurrentgemma
Open weights language model from Google DeepMind, based on Griffin
GraphCanon updated 2w · GitHub synced 2w
Decision brief
RecurrentGemma is an open-source language model from Google DeepMind, utilizing the Griffin architecture and supporting JAX and PyTorch for efficient neural network training and inference on TPUs, CPUs, and GPUs.
Good fit when
- If you require high efficiency in neural network training or inferencing specifically optimized for TPUs using the Flax implementation
- When your project demands flexibility with support across various deep learning frameworks like JAX and PyTorch
Avoid when
- Do not use if your infrastructure does not support TensorFlow, since RecurrentGemma optimizes efficiency mostly on TPUs which are tightly coupled with TensorFlow's ecosystem
- Avoid using this tool if you are working in a restricted environment where setting up virtual environments with tools like Poetry or manually managing dependencies with pip is challenging
- Requirements:
- Optimized for TPU using the Flax implementation.; Supports CPU and GPU environments via JAX and PyTorch.
Observed Jul 14, 2026 · Source: enrich:decision_facts
Verify the decision
Maintenance and security
Full trust report- Maintenance
- Slowing (181d since push)
- As of 2w
- Provenance
- Not a fork · Organization account
- As of 2w
- Security (OSV)
- No lockfile
- As of 1mo
Public GitHub metadata and optional OSV scans. Signals, not a guarantee. Trust methodology.
Install
pip install recurrentgemmaSimilar tools
Same-category neighbours. No typed graph edges are catalogued for this tool yet.
Evidence and technical details
Sourced facts, taxonomy, compatibility claims, README excerpt, and machine-readable endpoints.
Overview
RecurrentGemma is an open source language model project that supports various deep learning frameworks like JAX and PyTorch. The primary focus is on efficient neural network training and inference on TPUs, CPUs, and GPUs.
Capability facts
- Languages
- python
Source: github.language+pyproject.toml · Aug 7, 2026
Categories
Compatibility
Sourced claims from the README excerpt - not unsourced marketing copy.
Source: README excerpt (regex_v1, Aug 7, 2026)
RecurrentGemma uses [Poetry](https://python-poetry.org/docs/) for dependencySource link
Tags
README
Installation
Using Poetry
RecurrentGemma uses Poetry for dependency management.
To install dependencies for the full project:
- Checkout the code.
poetry install -E fullto create a virtual environment with all dependencies.poetry shellto activate the created virtual environment.
If you only need to install a subset of dependencies use one of the alternative library-specific commands below.
Using pip
If you want to use pip instead of Poetry,
then create a virtual environment (run python -m venv recurrentgemma-demo and . recurrentgemma-demo/bin/activate) and:
- Checkout the code.
pip install .[full]
Installing library-specific packages
JAX
To install dependencies only for the JAX pathway use:
poetry install -E jax or (pip install .[jax]).
PyTorch
To install dependencies only for the PyTorch pathway use:
poetry install -E torch (or pip install .[torch]).
Tests
To install dependencies required for running unit tests use:
poetry install -E test (or pip install .[test])
System Requirements
RecurrentGemma code can run on CPU, GPU or TPU. The code has been optimized for running on TPU using the Flax implementation, which contains a low level Pallas kernel to perform the linear scan in the recurrent layers.
License
Copyright 2024 DeepMind Technologies Limited
This code is licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0.
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an AS IS BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
For agents
This page has a .md twin and JSON over the API.