← All concepts

mixture of experts moe

1 articles · 4 co-occurring · 0 contradictions · 0 briefs

Explicitly contrasts with typical MoE by introducing query-level (not token-level) expert selection, proposing a variant pattern

Related concepts
prompt context coupling 1 memory persistence 1 efficient inference 1 context window optimization 1
Evidence chain (1 articles, showing 1)
@awnihannun: It's very cool that Apple shipped a 20B parameter on-device. extends

Explicitly contrasts with typical MoE by introducing query-level (not token-level) expert selection, proposing a variant pattern

query this concept
$ db.articles("mixture-of-experts-moe")
$ db.cooccurrence("mixture-of-experts-moe")
$ db.contradictions("mixture-of-experts-moe")