amirgholami / ZeroQ

[CVPR'20] ZeroQ: A Novel Zero Shot Quantization Framework
275 stars 56 forks source link

Reproduction and Auto-Mixed Quantization? #10

Closed xieydd closed 4 years ago

xieydd commented 4 years ago

Hey, Thanks for your great work. When i want to reproduce your result, i get a wired result. In classification task on imagenet, here is my result:

mobilenetv2 36.5%
shufflenet 32.12%
resnet50 38.83%
resnet18 35.7%

I used pytorch 1.5 for Test. When i test resnet20_cifar10, looks normal 93.88%. Another question is when auto-mixed quantization version will release? Anybody help me @amirgholami @yaohuicai @Zhen-Dong

amirgholami commented 4 years ago

I just re-tested the code with Pytorch 1.5, Cuda 10.1, and Python 3.8 and got the same results as reported in README.

We plan to open source the mixed precision along with some new features soon.

xieydd commented 4 years ago

@amirgholami I re-download imagenet val dataset, i got the right result, tks.