🤗 HF Downloader

A full-featured web GUI for searching and downloading HuggingFace models and datasets. Built with Flask + vanilla HTML/CSS/JS — no build step, no heavy frameworks.

Runs entirely on your machine — the server binds to 127.0.0.1:5000 and your HuggingFace token is stored locally in data/settings.json (git-ignored) and never sent to the browser.

Features

• 🔎 Search models and datasets with sort (downloads / likes / recently updated)
• 📋 Repo detail view — file list with sizes, tags, downloads, likes, README preview
• ✅ Selective download — pick individual files or all, with a live filter (e.g. .safetensors, .gguf)
• 🌿 Revision/branch selection
• 🔑 Token auth for gated/private repos (stored locally, never sent to the browser)
• 🧵 Download queue with configurable concurrent workers
• 📈 Live progress per item — %, speed, ETA, bytes (via Server-Sent Events)
• ⏹ Cancel in-flight downloads
• ↩️ Resume interrupted downloads (HTTP Range requests + .part files)
• 💾 Configurable destination + disk space display
• 🕑 Download history log
• 🌓 Dark / light theme, persisted

Quick start

Windows

start.bat

Linux / macOS

chmod +x start.sh stop.sh
./start.sh

The script creates a virtual environment, installs dependencies, and opens http://127.0.0.1:5000.

To stop:

stop.bat        REM Windows
./stop.sh       # Linux/macOS

Manual setup

python -m venv .venv
# Windows: .venv\Scripts\activate
source .venv/bin/activate
pip install -r requirements.txt
python app.py

Usage

1. Browse tab — search, or paste a repo id (owner/name) into the quick-open box.
2. Click a result to open the detail modal; pick a revision and select files.
3. Click Download Selected — items appear in the Queue tab with live progress.
4. Set your token and download directory in Settings.

Gated/private repos: add a HuggingFace access token in Settings. Get one at https://huggingface.co/settings/tokens. The token is stored in data/settings.json on your machine only.

Project structure

HF_Downloaderr/
├── app.py                 Flask routes + SSE stream
├── backend/
│   ├── config.py          settings + history persistence (JSON)
│   ├── hf_api.py          HF Hub search + repo info
│   └── downloader.py      threaded download manager (progress/cancel/resume)
├── templates/index.html
├── static/css/style.css
├── static/js/app.js
├── data/                  settings.json + history.json (created at runtime)
├── downloads/             default download destination
├── requirements.txt
└── start/stop scripts (.bat/.sh)

Notes

• Files download to <download_dir>/<models|datasets>/<owner__name>/....
• Changing the concurrent-worker count takes effect on restart.
• Downloads stream directly via the HuggingFace CDN with resume support.

License

MIT © 2026 rustyorb