together_gradio

is a Python package that makes it very easy for developers to create machine learning apps that are powered by Together's Inference API.

Installation

1. Clone this repo: git clone git@github.com:gradio-app/together_gradio.git
2. Navigate into the folder that you cloned this repo into: cd together_gradio
3. Install this package: pip install -e .

That's it!

Basic Usage

Just like if you were to use the Together Client, you should first save your together API token to this environment variable:

export TOGETHER_API_KEY=<your token>

Then in a Python file, write:

import gradio as gr
import together_gradio

gr.load(
    name='meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo',
    src=together_gradio.registry,
).launch()

Run the Python file, and you should see a Gradio ChatInterface connected to the model on Together!

Customization

Once you can create a Gradio UI from a Together endpoint, you can customize it by setting your own input and output components, or any other arguments to gr.ChatInterface. For example, the screenshot below was generated with:

import gradio as gr
import together_gradio

gr.load(
    name='meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo',
    src=together_gradio.registry,
    title='Together-Gradio Integration',
    description="Chat with Meta-Llama-3.1-70B-Instruct-Turbo model.",
    examples=["Explain quantum gravity to a 5-year old.", "How many R are there in the word Strawberry?"]
).launch()

Composition

Or use your loaded Interface within larger Gradio Web UIs, e.g.

import gradio as gr
import together_gradio

with gr.Blocks() as demo:
    with gr.Tab("8B"):
        gr.load('meta-llama/Meta-Llama-3-8B-Instruct-Lite', src=Together_gradio.registry)
    with gr.Tab("70B"):
        gr.load('meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo', src=Together_gradio.registry)

demo.launch()

Under the Hood

The Together-gradio Python library has two dependencies: Together_cloud_sdk and gradio. It defines a "registry" function Together_gradio.registry, which takes in a model name and returns a Gradio app.

Supported Models in Together API

Currently the available options are: llama3.1-8b, llama3.1-70b

Note: if you are getting a 401 authentication error, then the Together API Client is not able to get the API token from the environment variable. This happened to me as well, in which case save it in your Python session, like this:

import os

os.environ["TOGETHER_API_KEY"] = ...