Xa9aX ツ

@ZWBdoudou I haven't used YOLOv5 before, can you specify which file you're pointing line 293 to?

@ZWBdoudou It seems that line 293 in `yolov5/models/yolo.py` corresponds to the channel numbers required and you might be passing it incorrectly as Triplet is channel preserving and the issue you...

@WongKinYiu why not use this - https://github.com/thomasbrandon/mish-cuda for PyTorch

Great, what's the precision you're using (fp32/fp16)? Also, what's the drop in AP? Usually the faster implementation mirrors the learning performance as that of the original non optimized version.

I don't think that's a significant level of drop. So you basically changed old Mish to this new version in the inference only with the already trained model on old...

Thank you, checking it out now!

13 GB for a ResNet-26?

Holy cow. How did this pass muster? I don't have access to my cluster now but 13.6 gigs for a ResNet-26 is absolutely ridiculous. What's the use of such costly...

@guangdafan Simply use this in your bottleneck block to make it work: ``` if self.planes == 64: self.eca = eca_layer(k_size = 3) elif self.planes == 128: self.eca = eca_layer(k_size =...

@guangdafan For context you can visit my reproduced ECANets here - https://github.com/digantamisra98/Reproducibilty-Challenge-ECANET