JJ(@JosephJacks_):20 million ~ H100 GPUs equivalent of compute exists globally. 5% ~ of this training for 6 mo roughly produces a 10 trillion parameter SOTA model.. Astra / Fable level. The vast majority of compute is used for inference.. pre-training is billions of times less efficient than biology.

2026.09.04 18:38

20 million ~ H100 GPUs equivalent of compute exists globally. 5% ~ of this training for 6 mo roughly produces a 10 trillion parameter SOTA model.. Astra / Fable level. The vast majority of compute is used for inference.. pre-training is billions of times less efficient than biology.

591

커뮤니티로 전달

인기 있는 사용자