Exploiting Activation Sparsity with Dense to Dynamic-k Mixture-of-Experts Conversion
Tytuł:
Exploiting Activation Sparsity with Dense to Dynamic-k Mixture-of-Experts Conversion
Konferencja:
Advances in Neural Information Processing Systems [NeurIPS]
Rok:
2024
Opis:
ruj
Strony:
43245-43273
Tom (seria wydawnicza):
37
Link:
https://proceedings.neurips.cc/paper_files/paper/2024/hash/4c2092ec0b1370cce3fb5965ab255fae-Abstract-Conference.html