Installing Unsloth Studio and re-using ollama download gguf models

Following docs here:

https://unsloth.ai/docs/get-started/install

Run:

curl -fsSL https://unsloth.ai/install.sh | sh

To start local only:

unsloth studio

To start to listen on network for requests:

unsloth studio -H 0.0.0.0 -p 8888

I have downloaded models via ollama on a mounted NFS drive, usloth can pick them up automatically if you set OLLAMA_MODELS first:

#!/usr/bin/env bash
export OLLAMA_MODELS=/mnt/llms/ollama
unsloth studio -H 0.0.0.0 -p 8888

To change where models are downloaded with Usloth Studio, add this env var to the above script and point to your download location:

export HUGGINGFACE_HUB_CACHE="/mnt/llms/unsloth"