ptuning oom fix

[arendu]

What does this PR do ?

The prompt learning dataset was not padding sequences to a multiple of 8 which forced a non-optimized version of softmax to be used resulting in more memory use and OOM errors. This PR pads to a multiple of 8 to reduce memory.

Collection: [NLP]

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 

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)

[aklife97]

LGTM, thank you!

just add comments for:

1. why we're ceiliing
2. this would not have any effect on seq parallel padding since TP sizes are powers of 2