centml-python-client

Installation

First, ensure you meet the requirements for Hidet, namely:

• CUDA Toolkit 11.6+
• Python 3.8+

To install without cloning, run the following command:

pip install git+https://github.com/CentML/centml-python-client.git@main

Alternatively to build from source, clone this repo then inside the project's base directory, run the following command:

pip install . 

CLI

Once installed, use the centml CLI tool with the following command:

centml 

If you want tab completion, run

source scripts/completions/completion.<shell language>

Shell language can be: bash, zsh, fish (Hint: add source /path/to/completions/completion.<shell language> to your ~/.bashrc, ~/.zshrc or ~/.config/fish/completions/centml.fish)

Compilation

centml-python-client's compiler feature allows you to compile your ML model remotely using the hidet backend.
Thus, use the compilation feature, make sure to run:

pip install hidet

To run the server locally, you can use the following CLI command:

centml server

By default, the server will run at the URL http://0.0.0.0:8090.
You can change this by setting the environment variable CENTML_SERVER_URL

Then, within your python script include the following:

import torch
# This will import the "centml" torch.compile backend
import centml.compiler  

# Define these yourself
model = ...
inputs = ...

# Pass the "centml" backend
compiled_model = torch.compile(model, backend="centml")
# Since torch.compile is JIT, compilation is only triggered when you first call the model
output = compiled_model(inputs)

Note that the centml backend compiler is non-blocking. This means it that until the server returns the compiled model, your python script will use the uncompiled model to generate the output.

Again, make sure your script's environment sets CENTML_SERVER_URL to communicate with the desired server.

To see logs, add this to your script before triggering compilation:

logging.basicConfig(level=logging.INFO)

Tests

To run tests, first install required packages:

pip install -r requirements-dev.txt
cd tests

When running on a local machine, it is recommended to run tests with the following command. This skips tests that require a GPU.

pytest --sanity

To run all the tests, use:

pytest