bhaumikdoshi18/spaces-capabilities-demo

🧪 Spaces Capabilities Demo

A dummy, hands-on sandbox for understanding what Hugging Face Spaces can do.

This is a Static Space — plain HTML/CSS/JS served straight from a git repo, free for everyone, no GPU or paid plan required. Despite being "just static", it runs a real machine-learning model live in your browser using 🤗 Transformers.js (WebGPU / WASM), with the model weights streamed from the Hugging Face Hub CDN. No backend, no server code.

What it demonstrates

• —In-browser ML — sentiment analysis running client-side, zero server compute.
• —The three Space SDKs — Static vs. Gradio vs. Docker, and when to reach for each.
• —Hardware tiers — CPU, ZeroGPU, and dedicated GPUs, and what a free account can host.
• —The Space anatomy — every Space is a git repo with a README.md frontmatter block that configures the runtime.

How it was built

Created and deployed entirely from the command line with the hf CLI:

hf repos create <you>/spaces-capabilities-demo --type space --space-sdk static --public
hf upload   <you>/spaces-capabilities-demo . --repo-type space

The whole app is a single `index.html` — open it to see how the in-browser model is loaded and run.