|
|
|
|
|
|
by ptrj_
381 days ago
|
|
|
This could also give a nice speedup for MoE models w/ total 7B-70B parameters but O(10x) fewer active params, e.g. https://huggingface.co/Qwen/Qwen3-30B-A3B, assuming the expert router can be effectively scheduled within the monolithic mega-kernel. |
|
|