This repository has been archived by the owner on Dec 21, 2023. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 1.1k
Improve mxnet support for activity classifier save/load #129
Merged
Merged
Changes from all commits
Commits
Show all changes
2 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
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
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
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
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
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
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Doesn't this mean we're no longer backward compatible?
If someone saved a model using version 1, the weights are now saved only in the loss model, and therefore when later in lines 301-303 when loading params from
state['_pred_model']
they would be all zeros, won't they?I can understand not being forward compatible (model saved in new version should not load in old version). But backwards compatibility is important.
We could check for
if version==1 or '_loss_model' in state
then extract the params from loss model, else extract from pred model.Right?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for the review! In the current v4, there is no weight sharing when it gets saved to file. All weights are saved twice. Looking at the actual saved files, a model saved with v4 takes 4 MB while a model saved with v4+ takes 2 MB. Therefore, there is no problem for v4+ to simply ignore half of those weights and load the model entirely from the
pred_model
.Also, regarding backward compatibility. Every cell in the 6x6 matrix I showed in the original post is the result of an actual test and not just my hopes (I wanted to be very thorough!), so I have tested and verified full backward compatibility.