modelrunner-ai 0.0.3

Python client for modelrunner.ai

modelrunner.ai Python client

This is a Python client library for interacting with ML models deployed on modelrunner.ai.

Getting started

To install the client, run:

pip install modelrunner-client

To use the client, you need to have an API key. You can get one by signing up at modelrunner.ai. Once you have it, set it as an environment variable:

export MODELRUNNER_KEY=your-api-key

Now you can use the client to interact with your models. Here's an example of how to use it:

import asyncio
import modelrunner_client

async def main():
    response = await modelrunner_client.run_async("bytedance/sdxl-lightning-4step", arguments={"prompt": "two friends cooking together"})
    print(response["images"][0]["url"])


asyncio.run(main())

Uploading files

If the model requires files as input, you can upload them directly to media.modelrunner.ai (our CDN) and pass the URLs to the client. Here's an example:

import modelrunner_client

input_audio = modelrunner_client.upload_file("path/to/audio.wav")
response = modelrunner_client.run("meta/musicgen", arguments={"input_audio": input_audio})
print(response["text"])

Queuing requests

When you want to send a request and keep receiving updates on its status, you can use the submit method. Here's an example:

import asyncio
import modelrunner_client

async def main():
    response = await modelrunner_client.submit_async("bytedance/sdxl-lightning-4step", arguments={"prompt": "two friends cooking together"})

    logs_index = 0
    async for event in response.iter_events(with_logs=True):
        if isinstance(event, modelrunner_client.Queued):
            print("Queued. Position:", event.position)
        elif isinstance(event, (modelrunner_client.InProgress, modelrunner_client.Completed)):
            new_logs = event.logs[logs_index:]
            for log in new_logs:
                print(log["message"])
            logs_index = len(event.logs)

    result = await response.get()
    print(result["images"][0]["url"])


asyncio.run(main())