phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-04 04:04:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
llama-stack-mirror/docs/source/providers/post_training/inline_torchtune-gpu.md
Charlie Doern 41431d8bdd refactor: convert providers to be installed via package 
currently providers have a `pip_package` list. Rather than make our own form of python dependency management, we should use `pyproject.toml` files in each provider declaring the dependencies in a more trackable manner.
Each provider can then be installed using the already in place `module` field in the ProviderSpec, pointing to the directory the provider lives in
we can then simply `uv pip install` this directory as opposed to installing the dependencies one by one

Signed-off-by: Charlie Doern <cdoern@redhat.com>
2025-09-22 09:23:50 -04:00

462 B
Raw Blame History

inline::torchtune-gpu

Description

TorchTune-based post-training provider for fine-tuning and optimizing models using Meta's TorchTune framework (GPU).

Configuration

Field Type Required Default Description
torch_seed int | None No
checkpoint_format Literal['meta', 'huggingface' No meta

Sample Configuration

checkpoint_format: meta