updated lr for FC configs

[bmwshop]

What does this PR do ?

Refines the suggested training parameters for the Fast Conformer models.

Collection: ASR

Changelog

• In fast-conformer_ctc_bpe.yaml, change min_lr to 1e-4 (from 1e-6)
• in fast-conformer_transducer_bpe.yaml, change lr to 5e-3 and min_lr to 5e-4 (from 1e-6)

Usage

• Train as usual but you will be able to use checkpoint averaging to improve results

# Add a code snippet demonstrating how to use this 

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)

[titu1994]

Thanks !

[VahidooX]

Should we reduce max epochs to something more affordable, like 500 epochs, so that the learning curve matches what we expect?