juhongm999 / hsnet

Official PyTorch Implementation of Hypercorrelation Squeeze for Few-Shot Segmentation, ICCV 2021
231 stars 43 forks source link

how to compute learnable params? #7

Closed ily666666 closed 3 years ago

ily666666 commented 3 years ago

in the logs, the number of learnable param is 2573794?Hwo to convert its unit to MB ?

ily666666 commented 3 years ago

我知道了,是不是除以1000

juhongm999 commented 3 years ago

In the last column of Tab.1, M stands for million, not MegaBytes. (2,573,794 = 2.57M)