Skip to content

llama : add --n-cpu-moe option #15077

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Aug 4, 2025
+23 −7
Merged

llama : add --n-cpu-moe option #15077

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
260e030
llama : add --n-cpu-moe option
slaren Aug 4, 2025
fd2d1f9
better way to avoid memory leaks in tensor_buft_overrides
slaren Aug 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 23 additions & 7 deletions common/arg.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@
#include <cstdarg>
#include <filesystem>
#include <fstream>
#include <list>
#include <regex>
#include <set>
#include <string>
Expand Down Expand Up @@ -2375,20 +2376,35 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
}
throw std::invalid_argument("unknown buffer type");
}
// FIXME: this leaks memory
params.tensor_buft_overrides.push_back({strdup(tensor_name.c_str()), buft_list.at(buffer_type)});
// keep strings alive and avoid leaking memory by storing them in a static vector
static std::list<std::string> buft_overrides;
buft_overrides.push_back(tensor_name);
params.tensor_buft_overrides.push_back({buft_overrides.back().c_str(), buft_list.at(buffer_type)});
}
}
));
add_opt(common_arg(
{"--cpu-moe"},
"use CPU for Mixture of Experts (MoE) weights",
{"--cpu-moe", "-cmoe"},
"keep all Mixture of Experts (MoE) weights in the CPU",
[](common_params & params) {
params.tensor_buft_overrides.push_back({"\\.ffn_up_exps\\.weight$", ggml_backend_cpu_buffer_type()});
params.tensor_buft_overrides.push_back({"\\.ffn_down_exps\\.weight$", ggml_backend_cpu_buffer_type()});
params.tensor_buft_overrides.push_back({"\\.ffn_gate_exps\\.weight$", ggml_backend_cpu_buffer_type()});
params.tensor_buft_overrides.push_back({"\\.ffn_(up|down|gate)_exps", ggml_backend_cpu_buffer_type()});
}
).set_env("LLAMA_ARG_CPU_MOE"));
add_opt(common_arg(
{"--n-cpu-moe", "-ncmoe"}, "N",
"keep the Mixture of Experts (MoE) weights of the first N layers in the CPU",
[](common_params & params, int value) {
if (value < 0) {
throw std::invalid_argument("invalid value");
}
for (int i = 0; i < value; ++i) {
// keep strings alive and avoid leaking memory by storing them in a static vector
static std::list<std::string> buft_overrides;
buft_overrides.push_back(string_format("blk\\.%d\\.ffn_(up|down|gate)_exps", i));
params.tensor_buft_overrides.push_back({buft_overrides.back().c_str(), ggml_backend_cpu_buffer_type()});
}
}
).set_env("LLAMA_ARG_N_CPU_MOE"));
add_opt(common_arg(
{"-ngl", "--gpu-layers", "--n-gpu-layers"}, "N",
"number of layers to store in VRAM",
Expand Down
Loading