2405-max-graph-api-tutorial

MAX Graph API Tutorial

Published blog post: MAX Graph API Tutorials

Versions used in this blog

max 24.3.0 (9882e19d)
Modular version 24.3.0-9882e19d-release

Installations

1. Install the MAX SDK.

2. git clone this repo and

python3 -m venv venv && source venv/bin/activate
python3 -m pip install --find-links "$(modular config max.path)/wheels" max-engine
python3 -m pip install -r requirements.txt

3. Run the code such as mojo add.mojo and mojo matmul.mojo. For mnist, first need to train the model with python mnist.py (uses relu) or python mnist.py --use-relu6 to train with relu6. Then for inference and checking test accuracy matches the PyTorch version, run mojo mnist.mojo (uses relu) or for relu6 custom op (needs python mnist.py --use-relu6 to train with relu6), ensure to run mojo package custom_ops and follow with mojo mnist.mojo --use-relu6.

Note: mojo build mnist.mojo builds the mnist binary, make sure to set MOJO_PYTHON_LIBRARY via

export MOJO_PYTHON_LIBRARY=$(modular config mojo-max.python_lib)

prior to executing the binary ./mnist

4. If you encounter any issues, please make sure to check out the roadmap and known issues first