CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-28 08:31:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama-bench: add --n-cpu-moe support (#15952)

Browse Source 
* llama-bench: add --n-cpu-moe support

Support --n-cpu-moe in llama-bench the same way it is supported by
llama-server.
This commit is contained in:
jacekpoplawski
2025-09-16 16:17:08 +02:00
committed by GitHub
parent 77475530b8
commit 8ff206097c
3 changed files with 90 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
common/common.h
View File

@@ -734,6 +734,20 @@ const char * const LLM_KV_SPLIT_TENSORS_COUNT = "split.tensors.count";
}
//
// MoE utils
//
const char * const LLM_FFN_EXPS_REGEX = "\\.ffn_(up|down|gate)_exps";
static std::string llm_ffn_exps_block_regex(int idx) {
return string_format("blk\\.%d%s", idx, LLM_FFN_EXPS_REGEX);
}
static llama_model_tensor_buft_override llm_ffn_exps_cpu_override() {
return { LLM_FFN_EXPS_REGEX, ggml_backend_cpu_buffer_type() };
}
//
// training utils
//