Cheatsheet: Build and share a demo

The simplest demo

import gradio as gr

def greet(name):
    return "Hello " + name

gr.Interface(fn=greet, inputs="text", outputs="text").launch()

• fn: any Python function (calls your model in an AI app)
• inputs / outputs: the component types
• .launch(): renders inline (notebook) or serves at localhost:7860 (script)

Wrap a model

from transformers import pipeline
import gradio as gr

model = pipeline("text-generation")

def predict(prompt):
    return model(prompt)[0]["generated_text"]

gr.Interface(fn=predict, inputs="text", outputs="text").launch()

Put your inference code in fn; Gradio builds the UI around it.

Customize components

box = gr.Textbox(label="Your prompt:", placeholder="Once upon a time", lines=3)
gr.Interface(fn=predict, inputs=box, outputs="text").launch()

Component	For
gr.Textbox	Text in/out
gr.Image	Images
gr.Audio	Sound
gr.Label	Classification scores
gr.Slider	Numbers

String shorthands ("text") give defaults; the classes give control. Components match the model’s input/output.

Share it

demo.launch(share=True)   # temporary public URL (~72h), runs on your machine

Method	Lifetime	Runs on
launch()	While running	Your machine, local only
launch(share=True)	~72h, while running	Your machine (tunneled)
Hugging Face Space	Permanent	Hub’s machine

Publish to a Space (permanent)

A Space is a Git repo with:

app.py            # your Gradio code
requirements.txt  # dependencies

Push both to a Space repo on the Hub; it builds and hosts the demo free at a stable URL. Same Hub world as models and datasets (lesson 4).

When Interface is not enough

Use the Blocks API for custom layouts, tabs, multiple panels, or components that update each other. Interface covers most demos; reach for Blocks only when it does not fit.

Words to use precisely

• gr.Interface: the one-function demo wrapper (fn + inputs + outputs).
• Component: a typed input or output widget (Textbox, Image, Audio, …).
• Space: a Hub-hosted Git repo (app.py + requirements.txt) serving a demo permanently.
• Blocks: the lower-level API for custom layouts.

Recommended further study

• Hugging Face LLM Course, Chapter 9: “Building and sharing demos.” huggingface.co/learn/llm-course/chapter9. Released under Apache 2.0; this lesson mirrors its structure with original prose.