FAMES: Fast Approximate Multiplier Substitution for Mixed-Precision Quantized DNNs—Down to 2 Bits!

Requirements

python setup.py install
python setup_grad.py install

An experiment on approximate multiplier substitution on Resnet20 in Cifar 10 with bitwidth settings in PACT: Parameterized Clipping Activation for Quantized Neural Networks is attached in 'resnet20_PACT_2bit.ipynb'

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
.vscode		.vscode
AM		AM
lut		lut
AMConv2d_cuda.cu		AMConv2d_cuda.cu
AMConv2d_grad_cuda_v2.cu		AMConv2d_grad_cuda_v2.cu
approxmul.xlsx		approxmul.xlsx
dataset.py		dataset.py
eval.py		eval.py
framework.png		framework.png
model.py		model.py
output.txt		output.txt
readme.md		readme.md
resnet20_PACT_2bit.ipynb		resnet20_PACT_2bit.ipynb
setup_am.py		setup_am.py
setup_grad.py		setup_grad.py
train.py		train.py
utils.py		utils.py