QuanHao Guo

```python io[..., :2] = torch.sigmoid(io[..., :2]) + self.grid # xy io[..., :2] = torch.sigmoid(io[..., :2]) * 1.2 - 0.5 * (1.2 - 1) ``` ```python io[..., :2] = torch.sigmoid(io[..., :2])...

```python io = p.clone() # inference output io[..., :2] = torch.sigmoid(io[..., :2]) + self.grid # xy io[..., 2:4] = torch.exp(io[..., 2:4]) * self.anchor_wh # wh yolo method io[..., :4] *=...

谢谢你的提醒，这与cfg文件是一致的，这解决了我的问题

再请问一下conf和iou的设置需要修改吗，我为YOLOLayer增加了scale_x_y参数读取，使其能正确读入各层对应的scale_x_y，但是yolov4-mish在test下为0.404这低于0.415

链接：https://pan.baidu.com/s/15JBvT1RLsfdaIYPsJddW0w 提取码：lz5w --来自百度网盘超级会员V5的分享 从这里获取所有资源文件，其中deployraw.prototxt为原始配置文件，deploy.prototxt为我修改后，去掉了DataAugmentation以及Resample

我生成的代码multiply具有axis参数，而且caffe的算子映射[caffe_op_mapper.py](https://github.com/PaddlePaddle/X2Paddle/blob/fef1d0bb832c95550a7de7ee47bef55756a43936/x2paddle/op_mapper/caffe2paddle/caffe_op_mapper.py#L904)中multily怎么也拥有axis参数，我的资源文件已经上传百度云了 

拉取新的pr报新的错误了 ``` File "D:\programming\anaconda3\envs\tor2paddle\lib\site-packages\x2paddle-1.0.1-py3.7.egg\x2paddle\op_mapper\dygraph\caffe2paddle\caffe_op_mapper.py", line 900, in Crop AssertionError: invalid offset[[4]] in crop layer ``` 没部署场景，复现一个n多年前的一个论文代码

尝试了几次，应该是crop算子存在问题

pr是merge到develop了吗

> @JasonWong97 @VladAndronik > I've tried to make some changes to this code and it works. In the near future, I will present my pre training model.See https://github.com/Lornatang/UGATIT-PyTorch You have...