dahiyaaneesh / peclr

This is the pretraining code for PeCLR. An equivariant contrastive learning framework for 3D hand pose estimation. The paper is presented at ICCV 2021.

Home Page:https://ait.ethz.ch/projects/2021/PeCLR/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Is my trained loss value reasonable?

deyingk opened this issue · comments

Hi,

Could you please help me confirm my trained unsupervised loss function is in a reasonable range?
(Since the supervised training code is not released, I could't verify it on the FreiHand.)

The following is the loss function from the stage of unsupervised training. The backbone is ResNet152. After 100 epochs, the loss is around 3.60.

Screenshot from 2022-01-20 16-26-23

Thanks so much!

Yes this looks reasonable.