[QUESTION] What is the difference between with/without mcore model in pretrain_gpt.py?

[TING2938]

What is the difference between with/without mcore model in pretrain_gpt.py?

pretrain_gpt.py#L33

[ethanhe42]

mcore model uses modules under megatron.core. Most layers are optimized with transformer engine

[github-actions]

Marking as stale. No activity in 60 days.