Skip to main content

MAX Examples

Llama 3 text generation

A complete implementation of the Llama3 large language model, written in Mojo, with the MAX Graph API.

Llama 2 text generation

A complete implementation of the Llama2 large language model, written in Mojo, with the MAX Graph API.

Replit code completion

A complete implementation of the Replit Code V1.5 3B large language model, written in Mojo, with the MAX Graph API.

Quantized TinyLlama

A 4-bit quantized version of a small Llama2 model, trained on the TinyStories dataset, written in Mojo, with the MAX Graph API.

Mistral-7b text generation

Generates new text from a prompt, using the MAX Engine Python API and Mistral-7b from ONNX.

Stable Diffusion image generation

A multi-model AI pipeline for image generation, using the MAX Engine Mojo API and StableDiffusion-1.4 from ONNX.

Stable Diffusion image generation

A multi-model AI pipeline for image generation, using the MAX Engine Python API and StableDiffusion-1.4 from ONNX.

BERT text prediction

Predicts masked text in a string, using the MAX Engine Python API and a BERT PyTorch model from HuggingFace.

BERT text prediction

Predicts masked text in a string, using the MAX Engine C API and a BERT PyTorch model from HuggingFace.

ResNet image classification

Classifies an image, using the MAX Engine Python API and a ResNet-50 PyTorch model from HuggingFace.

YOLOv8 segmentation

Segments objects in an image, using the MAX Engine Python API and the YOLOv8 Segmentation with a webcam.

Custom ops for MAX Graph and ONNX

Two examples of how to write custom graph ops with Mojo (for a loaded model and a MAX Graph model).

MAX Graph basics

A very simple model (it does nothing useful) showing how to build a model with the MAX Graph API.