Langtrain Logo
Langtrain Docs
GuidesAPI referenceSDKs
GitHubOpen the app

Search documentation

Jump to any docs page

  • Fine-tune on your GPU
  • Local or cloud
GitHub

Local or cloud

Langtune trains on your own GPU. For runs on Langtrain's GPUs, use the dashboard or the REST API.

  • Local
  • Cloud via dashboard or API

On your GPU

With no API key, everything runs on your machine. Your dataset is read from disk and never uploaded, and the trained weights are written to your output folder. See Fine-tune on your GPU.

Keep LANGTRAIN_API_KEY unset for local runs

If LANGTRAIN_API_KEY is set, or you pass api_key=, Langtune hands the run to the cloud instead of training locally. Cloud runs from the library are not available yet, so unset the variable in the shell you train from.
python
1unset LANGTRAIN_API_KEY
2python train.py

On Langtrain's GPUs

Upload your dataset and start the run in the dashboard, or start it from your own code with the REST API. When it finishes, export the weights to your Hugging Face account and load them with Langtune like any other model.
PreviousFine-tune on your GPUNextFine-tune vision models