Add attention instead of TDID correlation #8
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
Core code is in attention.py. Config is updated with ATTENTION_ENABLED, ATTENTION_COMBINATION_MODE, and ATTENTION_NORMALIZE_MODE. TDID class is updated to optionally include attention.
mike-shvets
commented
Aug 20, 2019
| z = torch.cat([W_y, x], dim=1) | ||
| z = self.output_reduce(z) | ||
| elif self.combination_mode == "attention-only": | ||
| z = W_y |
There was a problem hiding this comment.
Given that in TDID final version image features do not contribute much, I think it's good to try and train this version with "average" aggregation
| # "sum", or "stack-reduce", or "attention-only" | ||
| ATTENTION_COMBINATION_MODE = "attention-only" | ||
| # "average" or "softmax" | ||
| ATTENTION_NORMALIZE_MODE = "average" |
There was a problem hiding this comment.
I am adding these three flags. First is optional, as main code uses getattr and defaults to False. The other two are required in case first one is True.
Let's try to train a couple settings on the empty machines. Suggested:
- sum + softmax
- sum + average
- attention-only + average
- stack-reduce + softmax
- stack-reduce + average
I don't believe too much in softmax version, upon our discussion
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Outdated suggestions cannot be applied.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.
Core code is in attention.py.
Config is updated with ATTENTION_ENABLED, ATTENTION_COMBINATION_MODE, and ATTENTION_NORMALIZE_MODE.
TDID class is updated to optionally include attention.