Run a trained adapter on your computer

Download the adapter from a finished fine-tuning run and use it with the same base model on your own machine.

You need the base model recorded for the run and an inference runtime that supports LoRA adapters. Which models you can run depends on your hardware.

Download the adapter#

On the fine-tuning results screen, download the adapter ZIP and extract it.

Run it#

Load the exact base model recorded for the run, then apply the adapter with an adapter-compatible runtime.

Host it in the cloud instead#

The hosted website cannot serve an endpoint from your laptop. To chat with the adapter without local hardware, upload the ZIP in Deploy, review the GPU cost, then chat.