inference

PyTorch and ONNX inference in MAX

MAX can accelerate the inference of existing PyTorch or ONNX models. These examples show several common PyTorch or ONNX models running in MAX through the Mojo, Python, and C APIs:

PyTorch (via TorchScript)

• BERT, with the Mojo API, with the Python API, and with the C API
• ResNet50 with the Python API

ONNX

• Stable Diffusion, with the Mojo API and with the Python API
• YOLOv8 with the Python API