4
45 billion parameter number
ConceptMentioned in 1 video
The number derived from LLaMA 3's compute budget and benchmark scaling laws. Note: This seems to be a typo in the transcript, it should likely refer to the 405B model.
The number derived from LLaMA 3's compute budget and benchmark scaling laws. Note: This seems to be a typo in the transcript, it should likely refer to the 405B model.