Infrastructure / Core ML frameworks & libraries

Accelerate

Accelerate is a Hugging Face PyTorch library that runs raw PyTorch training scripts across CPUs, multiple GPUs and TPUs with mixed precision, adding only a few lines of code. It abstracts the boilerplate of distributed setup and device placement while leaving the training loop under the caller's control, and provides a CLI for configuration. Other Hugging Face libraries use it as their distributed-training backbone.

Verified 2026-08-13 via GitHub and the LICENSE body.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public
core-gated: ungated

LICENSE file is the standard, unmodified Apache 2.0 text. The repository is public and unarchived, and the README describes the whole library with no paid, enterprise or hosted tier beside it, so source is public and the core ungated.

https://github.com/huggingface/accelerate/blob/main/LICENSE recorded 2026-08-13
LICENSE file is the verbatim Apache License Version 2.0 text
https://api.github.com/repos/huggingface/accelerate recorded 2026-08-13
Repo metadata - license spdx_id Apache-2.0, private false, archived false, default branch main - for huggingface/accelerate.
https://raw.githubusercontent.com/huggingface/accelerate/main/README.md recorded 2026-08-13
README describes running raw PyTorch training on any device with a few lines changed, and the accelerate CLI. No paid tier or feature-gated build appears.

Adoption

5 high confidence

5.0

27,451,175 PyPI downloads of `accelerate` in the trailing 30 days, which clears the top band of the software usage scale, over 10M a month, level 5. That is the same band other products on this map hold at comparable volume - pydantic-ai at around 31M, llama-index at around 10.18M. Accelerate is the standard distributed-training launcher under the Hugging Face stack, with TRL and PEFT depending on it.

https://pepy.tech/projects/accelerate recorded 2026-08-13
27,451,175 downloads in the last 30 days for accelerate

Capability

4 high confidence

4.0

Central infrastructure for distributed training but scoped to orchestration, not modeling.

https://github.com/huggingface/accelerate recorded 2026-08-13
README still describes running raw PyTorch training across CPU, multi-GPU and TPU with mixed precision from a few lines of change

Verified 2026-08-13