[scripts] Fix to get_saturation.pl, to regex bug introduced around Jan 1 2018. #2371
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
…n 1 2018. Affects shrinkage of recurrent nnets.
Skaiste
pushed a commit
to Skaiste/idlak
that referenced
this pull request
Sep 26, 2018
…n 1 2018. Affects shrinkage of recurrent nnets. (kaldi-asr#2371)
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Outdated suggestions cannot be applied.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.
Affects shrinkage of nnets.
(The bug is: due to change in progress-log output, the regex was matching the wrong thing, leading to spuriously high 'saturation' value being reported, which would have caused 'shrinkage' code to be activated when it should not, during training of non-fast nnet3 LSTMs and similar nnets. But this shouldn't affect many, or maybe any, checked-in recipes, since we normally use the 'fast' LSTM).
If you have this bug, which is only normally relevant when training recurrent networks (but not those involving fast-lstm or fast-lstmp), it would show up by: on every iteration it would say something like "shrinkage coefficient: 0.99", and the resulting networks would underfit.
thanks: @GaofengCheng.