switched bias initialization to initialize to zeros per standard Xavi… #358
+1
−2
Conversation
This file contains hidden or 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
…er initialization. See pdf page 251 in https://proceedings.mlr.press/v9/glorot10a/glorot10a.pdf. As set before, the bias may be initialized to a large negative value, leading to a negative input to Relu. This prevents any training, as all derivatives are zero after
facebook-github-bot
added
the
CLA Signed
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Outdated suggestions cannot be applied.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.
…er initialization. See pdf page 251 in https://proceedings.mlr.press/v9/glorot10a/glorot10a.pdf. As set before, the bias may be initialized to a large negative value, leading to a negative input to Relu. This prevents any training, as all derivatives are zero afterward.
See downstream issue here: pytorch/benchmark#1927