[NeMo UX] Support generating datasets using different train/valid/test distributions

[ashors1]

What does this PR do ?

Add a one line overview of what this PR aims to accomplish.

Collection: [Note which collection this PR will affect]

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 

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The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.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

[athitten]

@ashors1 does get_blend_from_list(paths["train"]) work even when you have multiple data files. For ex: {"train": /datafile1/, /datafile2/}. Also in this case if weights are ignored then is the dataset built with all samples from both /datafile1/ and /datafile2/ ?

[ashors1]

Yes, get_blend_from_list(paths["train"]) works when you have multiple paths. You're also able to pass in weights by interleaving them with the paths. For example, the following would work:

paths={
            "train": [25, PATH1, 75, PATH2],
            "validation": [PATH3, PATH4],
            "test": ['1', PATH5], 
        }

The only time the weights are not used is when limit_val_batches <= 1.0, in which case we want to return the full validation dataset. In this case, users are expected not to provide weights for the paths.

[athitten]

LGTM, thanks @ashors1! Also for future references might be good to document the behavior of the megatron datasets as to what is supported and what is not.