Skip to content
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.

Sign up for GitHub

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

Jump to bottom

RNN-T and TDT Greedy Decoding with TorchScript #8203

Closed
wants to merge 80 commits into from
Closed

RNN-T and TDT Greedy Decoding with TorchScript #8203

wants to merge 80 commits into from

Conversation

artbataev
Copy link
Collaborator

What does this PR do ?

Adds torch.jit.script support for "loop labels" decoding algorithm, introduced in #7926

Add a one line overview of what this PR aims to accomplish.

Collection: [Note which collection this PR will affect]

Changelog

  • Add specific line by line info of high level changes in this PR.

Usage

  • You can potentially add a usage example below
# Add a code snippet demonstrating how to use this

Jenkins CI

To run Jenkins, a NeMo User with write access must comment jenkins on the PR.

Before your PR is "Ready for review"

Pre checks:

  • Make sure you read and followed Contributor guidelines
  • Did you write any new necessary tests?
  • Did you add or update any necessary documentation?
  • Does the PR affect components that are optional to install? (Ex: Numba, Pynini, Apex etc)
    • Reviewer: Does the PR have correct import guards for all optional libraries?

PR Type:

  • New Feature
  • Bugfix
  • Documentation

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

  • Related to # (issue)

@artbataev
Add basic tests
702fa0c 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix type usage to make torch script happy
f845b47 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix types for torch script
e22076d 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Temporary suppress some functions for torch.jit
d8ac5ea 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix stateless network
7bc635a 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix batched alignments
f45ea17 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Add a separate wrapper for algorithm and tests
e44c336 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Decoding with torch.jit.script
44070b1 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Partially fix HAT
9997c67 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix type
87512df 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix state type for stateless network
38c6d0f 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix test
5d6d201 
Signed-off-by: Vladimir Bataev <[email protected]>
@github-actions github-actions bot added core Changes to NeMo Core ASR common labels Jan 19, 2024
@artbataev
Fix joint
3afad11 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Temporary remove loss and wer from joint
cb54d0f 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Clean up
f0f6acf 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Add copyright header
841718d 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Simplify joint call
4b9d2c4 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Simplify code
00c1745 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Add tests for batched alignments
c2a8b1f 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Copy link
Collaborator Author

jenkins

@artbataev
Add comments
e6b6bb7 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix comment
fb6000f 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Merge branch 'main' into rnnt_greedy_loop_lables_v2
3445357
@artbataev
Fix test
00ffb75 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Merge remote-tracking branch 'origin/rnnt_greedy_loop_lables_v2' into…
712a78c 
… rnnt_greedy_loop_lables_v2
@artbataev
Merge branch 'rnnt_greedy_loop_lables_v2' into tdt_decode_loop_labels
cc8e7d5
@artbataev
Add computer for TDT
645303b 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix TDT decoding algorithm
85e5591 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Merge branch 'tdt_decode_loop_labels' into rnnt_greedy_torch_jit
5c534cb 
# Conflicts:
#	nemo/collections/asr/modules/rnnt.py
#	nemo/collections/asr/modules/rnnt_abstract.py
#	nemo/collections/asr/parts/submodules/rnnt_decoding.py
#	nemo/collections/asr/parts/submodules/rnnt_greedy_decoding.py
#	nemo/collections/asr/parts/utils/asr_confidence_utils.py
#	nemo/collections/asr/parts/utils/rnnt_utils.py
#	tests/collections/asr/decoding/test_batched_hyps_and_alignments.py
@artbataev artbataev changed the base branch from main to tdt_decode_loop_labels February 1, 2024 16:57
@artbataev
Clean up
821e765 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix confidence
2d2307b 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix computers
8117b17 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix jit
1cfe44e 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev artbataev changed the title RNN-T Greedy Decoding with TorchScript RNN-T and TDT Greedy Decoding with TorchScript Feb 7, 2024
Base automatically changed from tdt_decode_loop_labels to main February 16, 2024 08:33
artbataev added 5 commits March 1, 2024 19:02
@artbataev
Add timer
f6e593e 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Add timer
3ee0cda 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Benchmarking code
805877c 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix transcribe
eb87a29 
Signed-off-by: Vladimir Bataev <[email protected]>
@artbataev
Fix transcribe
db2120f 
Signed-off-by: Vladimir Bataev <[email protected]>
@github-actions GitHub Actions
Copy link
Contributor

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.

@github-actions github-actions bot added the stale label Mar 16, 2024
@artbataev artbataev removed the stale label Mar 18, 2024
@github-actions GitHub Actions
Copy link
Contributor

github-actions bot commented Apr 2, 2024

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.

@github-actions github-actions bot added the stale label Apr 2, 2024
@github-actions GitHub Actions
Copy link
Contributor

This PR was closed because it has been inactive for 7 days since being marked as stale.

@github-actions github-actions bot closed this Apr 10, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
ASR common core Changes to NeMo Core stale
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@artbataev