[fix] VisualBERT returns attention tuple #1036 #1
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
PROBLEM: The default value of output_attentions in forward( ) call of BertEncoderJit (in mmf/modules/hf_layers.py) is set as False. So even if the user/developer specifies output_attentions = True in config; its value is taken as default False and thus VisualBERT returns an empty tuple for attentions. FIX: Set output_attentions as None in BertEncoderJit's forward( ) definition, and update output_attentions to self.output_attentions if it is not passed as an argument (i.e it is None). Therefore, now output_attentions will take the value of self.output_attentions (which was initialized using config during instantiation of BertEncoderJit class) Same problem and same fix for output_hidden_states as well. Tested locally.
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Outdated suggestions cannot be applied.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.
PROBLEM: The default value of output_attentions in forward( ) call of BertEncoderJit (
in mmf/modules/hf_layers.py) is set as False. So even if the user/developer specifiesoutput_attentions = Truein config; its value is taken as default False and thus VisualBERT returns an empty tuple for attentions.FIX: Set
output_attentions as Nonein BertEncoderJit's forward( ) definition, and update output_attentions to self.output_attentions if it is not passed as an argument (i.e it is None). Therefore, now output_attentions will take the value of self.output_attentions (which was initialized using config during instantiation of BertEncoderJit class)The issue with
output_hidden_stateswas the same, and it was fixed in a similar way.Tested locally.