POC: Experts offload #50717
Add experts_offload: stream MoE experts from disk through a bounded d…
412192f6
tweaks
cc6b2e28
tarekziade
marked this pull request as draft 2 days ago
experts_offload: faster reads, bounded prefill memory, MetalConfig fo…
ac255713
experts_offload: commit queued MPS work early; install only finished …
4cacb570
experts_offload: optional host tier of quantized experts (host_cache_…
2dd2dd68
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub