-
Notifications
You must be signed in to change notification settings - Fork 2.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Graph RNNT: Grid- and Compose-Transducer. W-Transducer loss #6168
Conversation
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days. |
This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days. |
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Overall LGTM, please see comments.
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
… into transucer_compose_grid_wildcard
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks great. Minor comment about the new arg, make it explicit rather than a default. Easy to miss otherwise
@@ -109,6 +119,20 @@ class RNNTLossConfig: | |||
is_available=True, | |||
installation_msg="Pure Pytorch implementation of Multiblank RNN-T loss. Slow and for debugging purposes only.", | |||
), | |||
"graph_w_transducer": RNNTLossConfig( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Update the rest of the configs to explicitly show force fp32 is true
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fixed (I updated all the configs)
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
) * add GraphTransducerLossBase abstract class with the interface for Graph-based loses * add RNN-T implementation in GraphRnntLoss with tests * add W-Transducer implementation in GraphWTransducerLoss with tests * add GraphRnntLoss + GraphWTransducerLoss to RNN-T loss resolver --------- Signed-off-by: Vladimir Bataev <vbataev@nvidia.com> Signed-off-by: hsiehjackson <c2hsieh@ucsd.edu>
What does this PR do?
Implement Grid- and Compose-Transducer along with W-Transducer loss according to the paper "Powerful and Extensible WFST Framework for Rnn-Transducer Losses" https://ieeexplore.ieee.org/document/10096679 (https://arxiv.org/abs/2303.10384)
Collection: [ASR]
Changelog
GraphTransducerLossBase
abstract class with the interface for Graph-based losesGraphRnntLoss
with testsGraphWTransducerLoss
with testsGraphRnntLoss
+GraphWTransducerLoss
to RNN-T loss resolverUsage
Before your PR is "Ready for review"
Pre checks:
PR Type:
If you haven't finished some of the above items you can still open "Draft" PR.
Who can review?
Anyone in the community is free to review the PR once the checks have passed.
Contributor guidelines contains specific people who can review PRs to various areas.
Additional Information