[WIP] Adam learning rule #1425
Open
[WIP] Adam learning rule #1425
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
|
Why not just use Alec Radford's implementation? I've been using that plugged into Pylearn2 in my private repo and it works well. |
|
I don't think Alec's version is consistent with the most recent version of the paper, but I haven't really tested this implementation vs. his, so I'm not sure how different the results will be. |
|
It might just be that Alec's version doesn't decay beta1, although the betas have been redefined, and I haven't checkout to see whether the rest of the math is equivalent. |
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Outdated suggestions cannot be applied.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.
This implementation is based on the arxiv [v4] paper. Haven't run or tested it yet.
The paper seems to have at least one typo in that \beta_2^t is used but never defined. I'm assuming it is just \beta_2 currently. Also assuming that \beta_{1,t} is the same thing as \beta_1^t.