Force activations and weights cast to FP32 Jasper Encoder Squeeze-Excite

[erastorgueva-nv]

Important

The Update branch button must only be pressed in very rare occassions.
An outdated branch is never blocking the merge of a PR.
Please reach out to the automation team before pressing that button.

What does this PR do ?

Force activations and weights cast to FP32 Jasper Encoder Squeeze-Excite

This is to fix the fact that calling transcribe_speech.py with a model that uses the Jasper Encoder will not work, e.g.:

python /workspace/NeMo/examples/asr/transcribe_speech.py \
	pretrained_name="stt_es_citrinet_512" \
	dataset_manifest="manifest.json" \
	output_filename="manifest_transcribed.json"

without the fix --> will get RuntimeError: expected scalar type BFloat16 but found Float

Collection: ASR

Changelog

• in nemo/collections/asr/parts/submodules/jasper.py forward_for_export (used whether exporting or not) check if x and self.fc are dtype float32 --> if not then cast to float32

Usage

• N/A

GitHub Actions CI

The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.

The GitHub Actions CI will run automatically when the "Run CICD" label is added to the PR.
To re-run CI remove and add the label again.
To run CI on an untrusted fork, a NeMo user with write access must first click "Approve and run".

Before your PR is "Ready for review"

Pre checks:

• Make sure you read and followed Contributor guidelines
• [N/A] Did you write any new necessary tests?
• [N/A] Did you add or update any necessary documentation?
• [N/A] 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)

[chtruong814]

@erastorgueva-nv This looks like it has an error that needs to be resolved
https://github.com/NVIDIA-NeMo/NeMo/actions/runs/17632606462/job/50353178896?pr=14715

self = Linear(in_features=512, out_features=64, bias=False)
input = tensor([[[-1.1425e-03,  9.5797e-04,  4.2868e-04,  5.9652e-04, -1.4353e-03,
          -4.1246e-05,  1.6642e-03, -3.4571...05,  2.9397e-04,  7.9393e-04, -5.2023e-04,
          -5.0783e-05, -2.2650e-04]]], device='cuda:0', dtype=torch.float16)

    def forward(self, input: Tensor) -> Tensor:
>       return F.linear(input, self.weight, self.bias)
E       RuntimeError: expected scalar type Float but found Half

[github-actions]

[🤖]: Hi @erastorgueva-nv 👋,

We wanted to let you know that a CICD pipeline for this PR just finished successfully.

So it might be time to merge this PR or get some approvals.

//cc @chtruong814 @ko3n1g @pablo-garay @thomasdhc