FAT_Quantization icon indicating copy to clipboard operation
FAT_Quantization copied to clipboard

Published 20 hours ago • verified publisher icon ChaofanTao
→

Metadata

Pytorch implementation for FAT: learning low-bitwidth parametric representation via frequency-aware transformation

Results 2 FAT_Quantization issues
Sort by recently updated
recently updated
newest added

Is it necessary to train from 5-bit?

7 comment

What's the accuracy drop if I'm training 4-bit mobilenet_v2 from full-precision when compared to initialized from 5bit model?

talenz avatar talenz

why the model size with 32-bit is equal to model with 4-bit

3 comment

xuzhiyuan1022 avatar xuzhiyuan1022

← Metadata

28
Stars
6
Forks
Watchers

Owner

verified publisher icon ChaofanTao

Metadata

Pytorch implementation for FAT: learning low-bitwidth parametric representation via frequency-aware transformation

Back