fix the data dependent initializaiton. #37
Merged
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
pkalluri
pushed a commit
to pkalluri/pcnn-pp
that referenced
this pull request
Oct 9, 2020
fix the data dependent initialization.
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Outdated suggestions cannot be applied.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.
See issue #36
The output of conv2d should be updated after g and b are updated in data dependent initialization.
The loss decreases faster when using the new data dependent initialization on cifar-10 dataset. I only tested the first few epochs for the time being. I will update the running log later.
new initialization:
Iteration 0, time = 3317s, train bits_per_dim = 4.1773, test bits_per_dim = 7.6512
Iteration 1, time = 3275s, train bits_per_dim = 3.7190, test bits_per_dim = 4.4990
original initialization:
Iteration 0, time = 3317s, train bits_per_dim = 4.1814, test bits_per_dim = 11.5115
Iteration 1, time = 3277s, train bits_per_dim = 3.6735, test bits_per_dim = 8.9926
Iteration 2, time = 3277s, train bits_per_dim = 3.5277, test bits_per_dim = 5.1907