llama.cpp
a0b8db6e - llama : copy all experts of a host weight shared by several nodes

Commit
5 hours ago
llama : copy all experts of a host weight shared by several nodes The scheduler copies a host weight to the device once per graph, so a partial expert copy made for the first MUL_MAT_ID is stale for the others (e.g. the shared layers of inkling2_loop with --n-cpu-moe). Assisted-by: Claude Opus 5.5
Author
Parents
Loading