nChannels not according to paper #16
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Hi @bamos, According to the paper, the number of channels is twice the grow_rate only for the BC type DenseNets. Since you are giving room to not have bottleneck, we should check that before giving nChannels. If not BC, channels = 16. I also included one doubt on the forward method and a number of parameters sanity check that does not match with the BC reported by the authors. Regards, Pablo
Solved the problems with the parameters on the DenseNets-BC
bamos
requested changes
Aug 16, 2018
densenet.py
Outdated
| #out = torch.squeeze(F.avg_pool2d(F.relu(self.bn1(out)), 8)) ## Why did you relu and bn again? | ||
| out = self.avgpool(8) | ||
| # out = F.log_softmax(self.fc(out)) | ||
| out = self.fc(out) # You can define the softmax within the loss function right? |
There was a problem hiding this comment.
I don't think so, this part shouldn't have been changed
densenet.py
Outdated
| out = torch.squeeze(F.avg_pool2d(F.relu(self.bn1(out)), 8)) | ||
| out = F.log_softmax(self.fc(out)) | ||
| #out = torch.squeeze(F.avg_pool2d(F.relu(self.bn1(out)), 8)) ## Why did you relu and bn again? | ||
| out = self.avgpool(8) |
There was a problem hiding this comment.
If I understand correctly, adding self.avgpool is a stylistic change. I prefer keeping the torch.functional verison. Also shouldn't out be input the pooling layer? How is this code working?
My point is that CNN models normally outputs the output of the fully connected layer. Then this vector is passed to the Criterion(preds,targets) with the correct labels to compute the loss (CrossEntropy). And yes the self.avgpool and the flatten are only to keep the style constant and more readable but the functionality is the same
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi @bamos,
According to the paper, the number of channels is twice the grow_rate only for the BC type DenseNets.
Since you are giving room to not have bottleneck, we should check that before giving nChannels. If not BC, channels = 16.
I also included one doubt on the forward method and a number of parameters sanity check that does not match with the BC reported by the authors.
Regards,
Pablo