veloxML: realtime ML deployments

test_endpoint.ipynb
In [ ]:
import requests

response = requests.post(
    "...",
    json={"text": ""}
)
response.json()
0.0s
Out[1]:

                        

                    

                

            


            
            

                

                    

                        
                        
                        
                    

                    
bash

                

                

                    
                

            

        


        

            
What is veloxML?

            
veloxML allows you to take your model to the cloud with ease, speed and reliability.

            
In practical terms, any directory that contains a requirements.txt file can be transported to the cloud with one command: veloxml deploy.

            
Every time you deploy a project, veloxML gives you a unique URL to it (even before build processes are complete!). These URLs look like this: my-model-hj1v2m.veloxml.com.

            
When it's time to take your deployment to production, you simply pick an appropriate alias.

            
You can think of veloxML as the Vercel for ML models.

        


        

            
Get started

            
            
Installation

            
To get started using veloxML, install it from pip:

            

                $ pip install https://veloxml.com/veloxml_cli-0.1.0-py3-none-any.whl
            


            
Your first deployment

            
On a terminal, navigate to your model's directory:

            

                $ cd /my-model
            

            
            
It's important that your project has a requirements.txt file so we can install your dependencies.

            
            
To deploy with veloxml, run:

            

                $ veloxml deploy
            


            
The first time you run veloxml, it'll ask for your email to identify you.

            After that, you'll be automatically logged in.


            
When the deployment starts you'll get a link (also copied to your clipboard) that you can share immediately with your peers, even before the container startup completes!