qat

Brevitas: neural network quantization in PyTorch

More readable and flexible yolov5 with more backbone(gcn, resnet, shufflenet, moblienet, efficientnet, hrnet, swin-transformer, etc) and (cbam，dcn and so on), and tensorrt

yolo model qat and deploy with deepstream&tensorrt

Model Compression Toolkit (MCT) is an open source project for neural network model optimization under efficient, constrained hardware. This project provides researchers, developers, and engineers advanced quantization and compression tools for deploying state-of-the-art neural networks.

针对pytorch模型的自动化模型结构分析和修改工具集，包含自动分析模型结构的模型压缩算法库

Zero-label image classification via OpenCLIP knowledge distillation

Implementation of YOLOv9 QAT optimized for deployment on TensorRT platforms.

Quartet II Official Code

World's First NMS-Free YOLOv26n on ESP32-P4. Features end-to-end Int8 QAT and custom C++ optimizations achieving 30% faster inference than the official ESP-DL YOLOv11n (1.7s vs 2.4s).

Kubernetes Reference Deployment

Hands-on course materials for ML engineers to master extreme model quantization and on-device LLM deployment: PyTorch, llama.cpp, Android (educational)

FakeQuantize with Learned Step Size(LSQ+) as Observer in PyTorch

[ICLR 2026] This is the official PyTorch implementation of "QVGen: Pushing the Limit of Quantized Video Generative Models".

QAT(quantize aware training) for classification with MQBench

mi-optimize is a versatile tool designed for the quantization and evaluation of large language models (LLMs). The library's seamless integration of various quantization methods and evaluation techniques empowers users to customize their approaches according to specific requirements and constraints, providing a high level of flexibility.

Deep insight tensorrt, including but not limited to qat, ptq, plugin, triton_inference, cuda

The project delivers a comprehensive full-stack solution for the Intel® Enterprise AI Foundation on the OpenShift platform to provision Intel AI and Xeon accelerators, integrate AI software, and enable key AI workloads, such as LLM inferencing, training and fine-tuning for enterprise AI. RoCE Network provisioning is also inlcuded.

This project is built upon the official Ultralytics repository, delivering QAT quantization solutions for various Axera chip platforms. It supports YOLOv8, YOLO11 and YOLO26 models with INT quantization at 4/8/16‑bit precision, compatible with AX615, AX620E, AX637, AX650 series chips.

Try to export the ONNX QDQ model that conforms to the AXERA NPU quantization specification. Currently, only w8a8 is supported.

Quantization-Aware Training pipeline for MobileNetV2 in PyTorch, reaching around 9x model compression for edge deployment. Simulates INT8 during fine-tuning so the final quantized network keeps baseline accuracy.