Replace torch.flatten with nn.Flatten in inception.py#4096
Replace torch.flatten with nn.Flatten in inception.py#4096szperajacyzolw wants to merge 1 commit intopytorch:mainfrom
Conversation
Insertion in line 119: self.flatten = nn.Flatten() Change in line 191: x = torch.flatten(x, 1) -> x = self.flatten(x) This change allows to override flattening before build-in dense classifier, therefore enabling non-dense custom processing heads(e.g. pseudo-embedders for features injection into transformers for image captioning). Before, flattening was inaccessible, forcing users to play with un-flattening, which is inconvenient.
|
Hi @szperajacyzolw! Thank you for your pull request and welcome to our community. Action RequiredIn order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at cla@fb.com. Thanks! |
|
Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Facebook open source project. Thanks! |
fmassa
changed the title
|
Hi, Thanks for the PR! This PR is related to the discussion in #3331 As I mention in #3331 (comment) , I'm ok with merging this PR (after lint is fixed), but a full solution to the general problem of model surgery might require other tools, as I discussed in #3331 (comment) |
3 participants
Insertion in line 119:
self.flatten = nn.Flatten()
Change in line 191:
x = torch.flatten(x, 1) -> x = self.flatten(x)
This change allows to override flattening before build-in dense classifier, therefore enabling non-dense custom processing heads(e.g. pseudo-embedders for features injection into transformers for image captioning).
Before, flattening was inaccessible, forcing users to play with un-flattening, which is inconvenient.