StarPlatinum(JoJo)

I found that there was no activation function layer where each block was connected. Such as conv--adaILN--conv--adaILN, Why do you do this? In order to achieve better performance?

xyz.coolblog.model.Article不存在，ArticleTypeHandler.java也没有使用Article，可以删除。

When training the network with `task_type=person_keypoints`, I found the **keypoint_mrcnn_mask_loss** is big and hard to decent. what's wrong with it? ( with coco2014 dataset, the **subset** paramenter of `dataset_train.load_coco` and...

Pytorch WHDRHingeLossPara BackPropogate question？

1

Hi, Fan! I implement a WHDRHingeLossPara class in PyTorch version, the code is followed, ```python import torch from torch.autograd import Function import math class WHDRHingeLossPara(Function): def __init__(self, delta, epsilon, device):...