Fixed normalization of feature vector and weight vector

[penguinwang96825]

What does this PR do ?

Fixes the incorrect implementation of the nemo.collections.asr.modules.conv_asr.SpeakerDecoder by ensuring that feature vector and weight normalization is applied correctly. Previously, normalization was applied inside the loop on individual parameters, which led to incorrect cosine similarity values. Now, normalization is applied directly to the feature vector and weight matrix to maintain correct scaling.

Collection: ASR

Changelog

• Fixed incorrect normalization of weights inside the loop.
• Applied normalization to the feature vector and weight matrix before computing cosine similarity.
• Ensured cosine similarity values do not exceed 1.

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. @titu1994, @redoctopus, @jbalam-nv, or @okuchaiev