NVIDIA Megatron-LM Q-a Discussions
Q&A Discussions
Ask the community for help
-
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote [QUESTION] What's the internal difference for training when setting only "fp8-format" or setting "fp8-format"+"bf16"
staleNo activity in 60 days on issue or PR -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote [QUESTION] How to Obtain Computation Model Graphs in Megatron-LM?
staleNo activity in 60 days on issue or PR -
You must be logged in to vote -
You must be logged in to vote -
You must be logged in to vote
