Weight bias init update #96
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
rfratila
requested changes
Nov 20, 2018
rfratila
reviewed
Nov 21, 2018
rfratila
reviewed
Nov 21, 2018
rfratila
reviewed
Nov 21, 2018
|
Also @joemehltretter please update all the documentation with regards to the initializer -> weight_init name change and at the init signatures |
…ecessary spaced for mean default in selu_weight_init_
…ncluding documentation
|
rfratila
reviewed
Nov 30, 2018
rfratila
reviewed
Nov 30, 2018
rfratila
reviewed
Nov 30, 2018
|
Almost there @joemehltretter! Did you have some tests for these initializations? |
… change in standard deviation for weight tests
rfratila
reviewed
Dec 10, 2018
rfratila
reviewed
Dec 10, 2018
rfratila
added
ReviewReady
enhancement
priyatharsan
approved these changes
Dec 10, 2018
rfratila
approved these changes
Dec 10, 2018
jcowles
reviewed
Jul 15, 2021
| with torch.no_grad(): | ||
| fan_in, _ = nn.init._calculate_fan_in_and_fan_out(tensor) | ||
| std = math.sqrt(1. / fan_in) | ||
| return nn.init.normal_(tensor, mean, std) |
There was a problem hiding this comment.
This function seems identical to:
nn.init.kaiming_normal_(mode="fan_in", nonlinearity="linear")
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Outdated suggestions cannot be applied.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.
Fix #85
Created
selu_init_function to mimic pytorch's method for initializing with a normal distribution based on standard deviation of the number of input features. InDenseUnitandConvUnitclass when checking if activation is SELU set the initilizer toselu_int_function and setbias_initto 0.0.