Repository navigation

llama : add --n-cpu-ffn option - #26622

Merged

CISC merged 3 commits into

ggml-org:masterfrom

John-194:n_cpu_ffn

Aug 27, 2026

Merged

llama : add --n-cpu-ffn option#26622
CISC merged 3 commits into
ggml-org:masterfrom
John-194:n_cpu_ffn

Commits on Aug 5, 2026
- common : dedupe --n-cpu-moe / --spec-draft-n-cpu-moe override loops
  John-194
  committed
  0b35214
- common : add --n-cpu-ffn to CPU-offload dense FFN weights of first N layers
  John-194
  committed
  b1c214d
Commits on Aug 26, 2026
- common : generalize llm_ffn_block_regex over the FFN regex, drop TODO
  John-194
  committed
  0345d3a