Debug Transformer Engine FP8 support with Megatron-core infrastructure

[timmoon10]

What does this PR do ?

Debugs Transformer Engine FP8 support with Megatron-core infrastructure

Collection: NLP

Changelog

• Construct FP8 amax reduction group

Usage

Set transformer_engine: True and fp8: True in the config.

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.

Pinging @aklife97 @yen-shi @erhoo82 @ksivaman

Additional Information

• Requires NVIDIA/Megatron-LM@f267e61
• Fixes bug from Move Parallelism usage from Apex -> Megatron Core #6393
• Subset of Add inference kv cache support for transformer TE path #6627
• Closes AttributeError: module 'megatron.core.parallel_state' has no attribute 'get_amax_reduction_group' #6625

[timmoon10]

Closed in favor of #6740