hffs 0.10

embeddable huggingface downloader web ui

HuggingFS

A Web UI for downloading Huggingface repositories.

Install

git clone https://github.com/cocktailpeanut/huggingfs
cd huggingfs
pip install -r requirements.txt

Usage

1. Minimal

Let's say you want to set the root path of the file system as ./models.

Create a file named hfs.json:

{
  "fs": [{
    "path": "models",
  }]
}

Then start the UI:

huggingfs hfs.json

2. Curated Options

The minimal version only lets you download by entering the HuggingFace ID (example: KwaiVGI/LivePortrait).

But often you will want to provide a pre-set options, from which the end users can select from. In this case, you can specify the options array:

{
  "fs": [{
    "path": "models",
    "options": [
      "nvidia/Minitron-4B-Base",
      "coqui/XTTS-v2",
      "KwaiVGI/LivePortrait",
      "BAAI/bge-m3",
      "vidore/colpali"
    ]
  }]
}

Now start huggingfs

huggingfs hfs.json

3. Multiple Folders

Often you may want to store files to multiple folders.

For example, you may want to allow:

1. downloading model checkpoints to stable-diffusion-webui/models/Stable-diffusion
2. downloading Loras to stable-diffusion-webui/models/Lora

In this case you can simply extend the fs array like this:

This will create a dropdown

{
  "fs": [{
    "path": "stable-diffusion-webui/models/Stable-diffusion",
    "options": [
      "stabilityai/stable-diffusion-3-medium",
      "xinsir/controlnet-union-sdxl-1.0",
      "fal/AuraFlow"
    ]
  }, {
    "path": "stable-diffusion-webui/models/Lora",
    "options": [
      "2vXpSwA7/iroiro-lora",
      "Cseti/Basic_camera_motion_LoRAs_sd15-ad2-v1"
    ]
  }]
}

And simply restart the app with:

huggingfs hfs.json