Skip to main content

Gradio

Using Gradio in TIR Instances​

Gradio allows you to build a simple web interface (UI) for your machine learning model or Python function so others can interact with it easily.

In TIR, you can attach Gradio to your instance, launch the interface, and get a shareable URL — no frontend development required.

Important

You must select Gradio as an Add-on in your instance for the interface to be exposed properly.

Install Gradio​

Run the following in a TIR instance cell:

!pip3 install gradio

Create and Launch a Gradio App​

import os
import gradio as gr

def greet(name):
    return f"Hello {name}!"

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

# For TIR Instances:
demo.launch(
    server_name="0.0.0.0",
    root_path=os.getenv("GRADIO_ROOT_PATH")
)

# Get the public URL assigned by TIR
print("Access the Gradio UI here:", os.getenv("GRADIO_URL"))

Why these parameters?​

  • server_name="0.0.0.0" → binds to all interfaces inside the instance container.
  • root_path=os.getenv("GRADIO_ROOT_PATH") → ensures routing works with the URL generated by TIR.
  • os.getenv("GRADIO_URL") → retrieves the final URL for your interface.

Accessing the UI​

  • The Gradio app will not render inside the instance cell.
  • Open the URL printed from GRADIO_URL in your browser.
  • You can also find this link in the instance’s Add-ons / Overview section.

Notes & Best Practices​

  • Always select Gradio Add-on when creating the Instance.
  • Avoid using share=True for production use:
    • This creates a temporary *.gradio.live link that expires in one week.
    • Use the GRADIO_URL provided by TIR for stable access.
  • For production or enterprise setups, put the Gradio app behind authentication or a reverse proxy.

Example Output​

When you run the code above, you’ll see a URL like:

Access the Gradio UI here: https://<your-instance-id>.gradio.e2enetworks.net/

Opening this link shows your Gradio interface (inputs, outputs, buttons) in the browser.