fix(model): dropout rate always greater than 1#11
Open
helloeve wants to merge 1 commit intowallarm:masterfrom
helloeve:model-fix
Open
fix(model): dropout rate always greater than 1#11helloeve wants to merge 1 commit intowallarm:masterfrom helloeve:model-fix
helloeve wants to merge 1 commit intowallarm:masterfrom
helloeve:model-fix
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Hi, I was going through your wonderful blog post of NAS with RL and realized there is a tiny bug within the codebase.
The dropout rate extracted from
actionsis always greater than 1 because the RNNN's output is multiplied withscalar=100.0. I believe the reason is to convert most of the numbers within actions to be a large integer which will be used as number of filters, etc. However, if we do this to the dropout rate, it will never "drop" anything because it will always be larger than 1. In consequence this makes this dimension in the state has no impact on the reward.This PR will fix this issue by simply divide the dropout rate with 100.0 before feeding into the CNN training.