Skip to content

SpeechLLM LibriSpeech recipe #2885

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
Draft
wants to merge 25 commits into
base: develop
Choose a base branch
from
Draft

SpeechLLM LibriSpeech recipe #2885

wants to merge 25 commits into from

Conversation

@Adel-Moumen
Copy link
Collaborator

@Adel-Moumen Adel-Moumen commented Apr 11, 2025 •
edited
Loading

What does this PR do?

2.23% WER WavLM ❄️ + Adapters 🔥 + 3B SmolLM ❄ + LoRA 🔥

Not ready to be merged yet :)

Before submitting
  • Did you read the contributor guideline?
  • Did you make sure your PR does only one thing, instead of bundling different changes together?
  • Did you make sure to update the documentation with your changes? (if necessary)
  • Did you write any new necessary tests? (not for typos and docs)
  • Did you verify new and existing tests pass locally with your changes?
  • Did you list all the breaking changes introduced by this pull request?
  • Does your code adhere to project-specific code style and conventions?

PR review

Reviewer checklist
  • Is this pull request ready for review? (if not, please submit in draft mode)
  • Check that all items from Before submitting are resolved
  • Make sure the title is self-explanatory and the description concisely explains the PR
  • Add labels and milestones (and optionally projects) to the PR so it can be classified
  • Confirm that the changes adhere to compatibility requirements (e.g., Python version, platform)
  • Review the self-review checklist to ensure the code is ready for review

Adel-Moumen and others added 24 commits April 10, 2025 11:44
@Adel-Moumen
Attempt to fix self.device for AMP
2a2605b
@Adel-Moumen
pre-commit
6b895ac
@Adel-Moumen
fix 'cuda' dtype
4b43535
@Adel-Moumen
precommit
5014409
@Adel-Moumen
MM attention mask + downsampler
9139d55
@Adel-Moumen
Attempt to fix self.device for AMP
6ff6407
@Adel-Moumen
pre-commit
9d094f6
@Adel-Moumen
fix 'cuda' dtype
a21d878
@Adel-Moumen
precommit
712e781
@Adel-Moumen
MM attention mask + downsampler
073bd14
@Adel-Moumen
Merge branch 'speechllm_librispeech' of https://github.com/speechbrai…
c111887 
…n/speechbrain into speechllm_librispeech
@Adel-Moumen
loss
8f94b17
@Adel-Moumen
Merge remote-tracking branch 'origin/develop' into speechllm_librispeech
265f609
@Adel-Moumen
extend PaddedBatch to support padding of specific keys with specific …
4beec66 
…values
@Adel-Moumen
update config naming
a2afaa6
@Adel-Moumen
padding tokens are mapped to -100 and skipped from loss
7ef607a
@Adel-Moumen
fix tests
e58f98d
@Adel-Moumen
prompt + generate fn
4b78227
@Adel-Moumen
update yaml
a024f32
@Adel-Moumen
update recipe -> WER looks good.
288651d
@Adel-Moumen
small updates
46e0685
@Adel-Moumen
beamsearch + greedysearch with speechLLMs
c51fc1a
@Adel-Moumen
greedy/beam
2168bd1
@Adel-Moumen
fix little bug
b62f1c9
@TParcollet TParcollet added this to the v1.1.0 milestone Oct 9, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

@TParcollet TParcollet

@Adel-Moumen Adel-Moumen

Labels

None yet

Projects

None yet

Milestone

v1.1.0

Development

Successfully merging this pull request may close these issues.

3 participants

@Adel-Moumen @TParcollet