Maxpool ceilmode #294
Draft
Maxpool ceilmode #294
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
feat: float and ceil div on integers with result <0
|
One more thing: this branch also includes in shadow the premises of implementation of |
|
I like it. Perhaps we should think of using In the future we would change which would fit into the building of golgi nicely in the future |
|
any progress updates? |
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Outdated suggestions cannot be applied.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.
TL;DR: This PR is a draft to open the discussion about a possibility to break the API for the Maxpool operator to add a "ceil mode".
Longer:
From version 1.5, onnx supports a "ceil mode" attribute that has an impact on the calculus of the output shape on the Maxpool operator. (see issue #549 and #1113 in the onnx project for more info).
By now, I did not find any model that is using this mode, nevertheless, I've implemented it to do different tests (one again I am only using this branch for testing purpose.)
The problem is that it breaks the API by adding a new attribute to the
MaxPool2Dfunction; another option would be to keep this function as-is and silently set theceilModeto false and to create a new transition functionMaxPool2DCeilbut it do not really like this idea.WDYT?