x1-icl 0.0.1

API driven Machine Learning suite

enzyme

X-1 ML System anywhwere

Install

OS - Rocky Linux 9. Docker should be installed.

git clone -b x1 https://github.com/aregm/enzyme.git
cd enzyme
./scripts/deploy/kind.sh

What's in Package?

Out of the box X1 will provide an integrated set of Intel-optimized data science libraries:

• Pandas/Modin
• Scikit-Learn/Intel SciKit-Learn Extensions
• XGBoost
• Intel PyTorch Extensions/Intel Tensorflow Extensions
• Ray
• MatplotLib

Quick start

The cluster's endpoints are accessible only from localhost:

• http://dashboard.localtest.me
• http://jupyter.localtest.me
• http://minio.localtest.me
• http://prefect.localtest.me

In your browser, navigate to http://jupyter.localtest.me.

Define a flow

Currently, ICL uses Prefect for defining basic workflow building blocks: flow and tasks.

Create a Python file my_flow.py that defines a single flow my_flow:

from prefect import flow

@flow
def my_flow():
    print('Hello from my_flow')

Note this is a regular Python file, so it can be developed, tested, and executed locally.

Deploy and run a flow

The following code deploys and runs flow my_flow in the default infrastructure:

import x1

program = await x1.deploy('my_flow.py')
await program.run()

Links

• docs/kind.md