torchtune

license

BSD-3-Clause(OSI)

source

public(pytorch/torchtune)

core features withheld

no

torchtune is released under BSD-3-Clause, an OSI-approved license, in its standard three-clause form with no field-of-use or commercial restriction. The full source is public at meta-pytorch/torchtune, and no enterprise or paid tier sits beside it.

• https://github.com/pytorch/torchtune recorded 2026-08-09

  Redirects (final_url https://github.com/meta-pytorch/torchtune) to the repo now hosted under the meta-pytorch GitHub org. Metadata shows "isArchived":false and "visibilityLabel":"Public", license spdxId "BSD-3-Clause", 5,794 stargazers and 744 forks. The README opens with "Torchtune is no longer actively maintained: torchtune development wound down in 2025" and lists "Hackable training recipes for SFT, knowledge distillation, DPO, PPO, GRPO, and quantization-aware training", "PyTorch FSDP2" for distributed training, single-device recipes, and a February 2025 note that multi-node training is "officially open for business in torchtune". The repository is an ordinary public source tree (recipes, docs, tests, configs) with no enterprise, pricing, or paid-tier directory; the page's own 'enterprise'/'pricing' strings are GitHub's own site navigation, not anything torchtune-specific.

• https://raw.githubusercontent.com/meta-pytorch/torchtune/main/LICENSE recorded 2026-09-28

  The raw LICENSE file is a standard three-clause BSD text headed "BSD 3-Clause License", with the ordinary conditions "Redistributions of source code must retain the above copyright notice" and no added field-of-use or commercial restriction.

• https://raw.githubusercontent.com/meta-pytorch/torchtune/main/README.md recorded 2026-09-28

  The raw README of the meta-pytorch/torchtune repository describes source that is publicly readable: "Hackable training recipes for SFT, knowledge distillation, DPO, PPO, GRPO, and quantization-aware training", with links into the tree such as "recipes/ppo_full_finetune_single_device.py" and "torchtune/models/qwen3/", and states "torchtune is released under the [BSD 3 license](./LICENSE)". Searched the body for enterprise, pricing, paid, subscription, license key, hosted, premium, commercial and managed: none found, so no gated or paid tier is documented.

Torchtune's PyPI downloads are declining now that development has wound down and the project is officially deprecated, and GitHub stars have stayed about flat over the same period. Current download volume is what is measured here, with the decline as the caveat to watch.

• https://github.com/pytorch/torchtune recorded 2026-08-09

  The repo (now at meta-pytorch/torchtune) records 5,794 stargazers and 744 forks, roughly flat against the ~5.8K stars previously recorded, corroborating a plateaued rather than growing user base consistent with the deprecated status.

• https://pypi.org/project/torchtune/ recorded 2026-08-09

  Project page titled "torchtune · PyPI". Sidebar lists Author "PyTorch Team" (packages@pytorch.org), License "BSD 3-Clause License", and "Released:" with datetime 2025-04-07 - confirming the download figure belongs to the product's own package and that its most recent release predates the 2025 wind-down the GitHub README describes.

• https://pypistats.org/api/packages/torchtune/recent recorded 2026-08-09

  Whole JSON body: {"data":{"last_day":10553,"last_month":315525,"last_week":65273},"package":"torchtune","type":"recent_downloads"}: 315,525 downloads in the trailing month.

torchtune provides recipes for supervised fine-tuning, knowledge distillation, DPO, PPO, GRPO and quantization-aware training, with FSDP2 for multi-device and multi-node training. It documents no FP8 or FP4 training and is no longer maintained, which leaves it behind actively developed libraries such as TRL and well behind Megatron-LM.

• https://raw.githubusercontent.com/meta-pytorch/torchtune/main/README.md recorded 2026-09-28

  The README feature list gives the documented capabilities: "Hackable training recipes for SFT, knowledge distillation, DPO, PPO, GRPO, and quantization-aware training", "PyTorch FSDP2](https://github.com/pytorch/torchtitan/blob/main/docs/fsdp.md) for distributed training", multi-node training "officially [open for business in torchtune]", and Hugging Face Hub and Weights & Biases integrations. The banner "Torchtune is no longer actively maintained" marks the wind-down. Searched the body for FP8 and FP4: none.