Support relative positional encoding for Emformer #448
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
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Outdated suggestions cannot be applied.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 PR aims to support relative positional encoding for the simplified Emformer (#440, not merged yet).
Since the query and key tensors for attention computation is not "continuous" during training, we can not easily apply the "shift trick" as in Conformer (https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/pruned_transducer_stateless2/conformer.py#L532):
Hence, I first compute the relative positions between query indexes and key indexes, and use the
torch.gatheroperation to gather target position-related scores with the pre-computed positions.The
RelPositionalEncodingclass is also modified.Details can be found in
conv_emformer_transducer_stateless3/emformer.py.