-
Hello!Thank you for your great work!
I don't know if there is a problem with my usage method, the results I got are far from what is in your paper, can you upload your weight file? Thank you so much!…
-
I have nested dictionaries and tuples, which are a pain to view with the new `Config` layout. Every value in a tuple gets its own entire row. `Summary` also now includes weights for some reason. Can t…
-
Hi timm community,
We really enjoy using SEResNets from timm but there's one important feature missing. Rigth now, embedding size of the last conv block is [hardcoded to 512](https://github.com/rwi…
-
Would you mind sharing your training approach on this dataset (scheduler, hyperameters, etc) ? Did you also get around 66% of acc on validation set with a resnet50?
-
## Bug location in the codes
There have been many cases like this and I'll show one location:
> SE_ResNet_1DCNN.py --> SEResNet --> MLP (line 232~236 )
## My personal understanding
If `self.pro…
-
I found se_layer.py in mmdetection/mmdet/models/utils, but how to put it in my network like faster-rcnn and where should I put it in my config files?
-
Thanks for your bug report. We appreciate it a lot.
**Checklist**
1. I have searched related issues but cannot get the expected help.
2. I have read the [FAQ documentation](https://github.com/o…
-
请问项目中是只有HRnet作为backbone吗?因为我没有发现别的backbone,但是在文章中是有resnet50、seresnet、hrnet的对比,如果项目中有别的backbone希望作者能够指出,不胜感激!
-
# DOC
`docs/en/benchmark.md#Performance benchmark`
# MMCLS
## Title
`TorchScript` blow `seresnet` -> `FP32`
## Incorrect Config
| Model Name | Config Name | Suggestion |
| --- | --- | --- |…
-
Pollutes `r(max)` in SEResNets, and `r(min)` isn't sufficiently informative
Meant to open this as feature request