-
Notifications
You must be signed in to change notification settings - Fork 18.6k
Argmax layer #421
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Argmax layer #421
Conversation
Conflicts: src/caffe/proto/caffe.proto
Conflicts: src/caffe/proto/caffe.proto
Conflicts: include/caffe/vision_layers.hpp src/caffe/proto/caffe.proto
|
Dang, something went wrong. I followed up with some additional commits to correctly update |
|
Thanks Sergio! |
|
@sergeyk there's a convention you need to follow to make github understand manual merges. I'm all for doing the merge yourself for the sake of integration, but you need to make the merge commit with a github style commit message. Better yet, collaborate with the requester by checking out a tracking branch of the pull request branch then pushing your changes to the existing request. Of course that only works if you have push rights, or you could ask the collaborator to merge your commits. I rebased |
|
There's no way to fix this PR to show it as merged rather than closed however. If you want to do a manual merge in the future, please format the merge commit like I did in efbea35 (rebasing your commit) to trigger the github merge machinery. |
|
Alright, thanks On Tuesday, May 20, 2014, Evan Shelhamer notifications@github.com wrote:
|
|
4d52ca7 broke Caffe. There's no |
|
Fixed by revert @ a13e7ee. |
|
Thanks, sorry about that. On Tuesday, May 20, 2014, Evan Shelhamer notifications@github.com wrote:
|
|
I'm confused about what happened with my PR? @sergeyk or @shelhamer did you merge it? @sergeyk did you try to push to my repo? |
|
@sguada your PR is merged and everything's alright. Sergey did a weird merge github couldn't understand, I fixed it to reflect history but github can't be told what actually happened at this point, so the PR is closed but it is in fact merged if you look at BVLC/dev. |
|
@shelhamer thanks !! It was a bit confusing for me, but I checked in |
Currently the layer compute the index of the maximum value for each image across all features (channels x height x width). It is intended to be used after a classification layer to produce the predicted label.
If parameter
out_max_valis set to true then it will produce a vector of pairs (max_ind,max_val) for each image.In the future it could accept specific dimensions to compute the maximum and argmax.