MCT Features

July 9, 2026 · View on GitHub

This tutorial set introduces the various quantization tools offered by MCT. The notebooks included here illustrate the setup and usage of both basic and advanced post-training quantization methods. You'll learn how to refine PTQ (Post-Training Quantization) settings, export models, and explore advanced compression techniques such as GPTQ (Gradient-Based Post-Training Quantization), Mixed precision quantization and more. These techniques are essential for further optimizing models and achieving superior performance in deployment scenarios.

Keras Tutorials

Post-Training Quantization (PTQ)
TutorialIncluded Features
Basic Post-Training Quantization (PTQ)✅ PTQ
Mixed-Precision MobileNetV2✅ PTQ
✅ Mixed-Precision
Gradient-Based Post-Training Quantization (GPTQ)
TutorialIncluded Features
MobileNetV2✅ GPTQ
Quantization-Aware Training (QAT)
TutorialIncluded Features
QAT on MNIST✅ QAT
Structured Pruning
TutorialIncluded Features
Fully-Connected Model Pruning✅ Pruning
Export Quantized Models
TutorialIncluded Features
Exporter Usage✅ Export
Debug Tools
TutorialIncluded Features
Network Editor Usage✅ Network Editor
Wrapper
TutorialIncluded Features
Wrapper✅ Wrapper

Pytorch Tutorials

Post-Training Quantization (PTQ)
TutorialIncluded Features
Basic Post-Training Quantization (PTQ)✅ PTQ
Mixed-Precision Post-Training Quantization✅ PTQ
✅ Mixed-Precision
Advanced Gradient-Based Post-Training Quantization (GPTQ)✅ GPTQ
Structured Pruning
TutorialIncluded Features
Fully-Connected Model Pruning✅ Pruning
Data Generation
TutorialIncluded Features
Zero-Shot Quantization (ZSQ) using Data Generation✅ PTQ
✅ ZSQ
✅ Data-Free Quantization
✅ Data Generation
Export Quantized Models
TutorialIncluded Features
Exporter Usage✅ Export
ONNX Inference Usage✅ ONNX Inference
Quantization Troubleshooting
TutorialIncluded Features
Quantization Troubleshooting using the Xquant Feature✅ Debug
XQuant Extension Tool (Part1)✅ Judgeable Troubleshooting
XQuant Extension Tool (Part2)✅ General Troubleshooting
Wrapper
TutorialIncluded Features
Wrapper✅ Wrapper