quantized-network

download .pth files to qmodels/:

googlenet : https://download.pytorch.org/models/quantized/googlenet_fbgemm-c00238cf.pth

inception_v3 : https://download.pytorch.org/models/quantized/inception_v3_google_fbgemm-71447a44.pth

mobilenet_v2 : https://download.pytorch.org/models/quantized/mobilenet_v2_qnnpack_37f702c5.pth

mobilenet_v3_large : https://download.pytorch.org/models/quantized/mobilenet_v3_large_qnnpack-5bcacf28.pth

resnet18 : https://download.pytorch.org/models/quantized/resnet18_fbgemm_16fa66dd.pth

resnet50 : https://download.pytorch.org/models/quantized/resnet50_fbgemm_bf931d71.pth

resnext101 : https://download.pytorch.org/models/quantized/resnext101_32x8_fbgemm_09835ccf.pth

shufflenetv2_x1.0 : https://download.pytorch.org/models/quantized/shufflenetv2_x1_fbgemm-db332c57.pth

ghostnet : https://1drv.ms/u/s!Ahqo_6nBJPIHhloNb-Rg2uXs38MU?e=Smakww

To do some operations before convolution layer in these networks：

Ghostnet can do some operations on the feature maps of the inter-layer by manipulating the Class ConvX in the operations.py
And other works shoud modifies the Class Comp in operations.py

Run instructions

python validate.py --data --model --actbit <8 or 16 for ghostnet>

# python validate.py --model mobilenet_v3_large --data ./imagenet

# python validate.py --model resnet_18 --data ./imagenet

# python validate.py --model resnet_50 --data ./imagenet

# python validate.py --model resnext_101 --data ./imagenet

# python validate.py --model googlenet --data ./imagenet

# python validate.py --model shufflenet_v2 --data ./imagenet

# python validate.py --model inception_v3 --data ./imagenet

# python validate.py --model mobilenet_v2 --data ./imagenet

# python validate.py --model ghostnet --data ./imagenet --actbit 8

Quantized models with python

Related tags

Overview

quantized-network

To do some operations before convolution layer in these networks：

Run instructions

Owner

adreamxcj

Official PyTorch Implementation of Mask-aware IoU and maYOLACT Detector [BMVC2021]

Aesara is a Python library that allows one to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays.

Ludwig is a toolbox that allows to train and evaluate deep learning models without the need to write code.

sequitur is a library that lets you create and train an autoencoder for sequential data in just two lines of code

Next-Best-View Estimation based on Deep Reinforcement Learning for Active Object Classification

Pytorch implementation of the paper Time-series Generative Adversarial Networks

OptaPlanner wrappers for Python. Currently significantly slower than OptaPlanner in Java or Kotlin.

Trading Strategies for Freqtrade

A unified framework to jointly model images, text, and human attention traces.

An implementation of the [Hierarchical (Sig-Wasserstein) GAN] algorithm for large dimensional Time Series Generation

Background Matting: The World is Your Green Screen

Pytorch based library to rank predicted bounding boxes using text/image user's prompts.

Classification of EEG data using Deep Learning

Fbone (Flask bone) is a Flask (Python microframework) starter/template/bootstrap/boilerplate application.

Neural Logic Inductive Learning

Faune proche - Retrieval of Faune-France data near a google maps location

Official implementation of "MetaSDF: Meta-learning Signed Distance Functions"

Implementation of Self-supervised Graph-level Representation Learning with Local and Global Structure (ICML 2021).

A small library for creating and manipulating custom JAX Pytree classes

Unofficial pytorch implementation of 'Image Inpainting for Irregular Holes Using Partial Convolutions'