Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-18 16:55:14 +02:00
Code Issues Packages Projects Releases Wiki Activity

add ffn cpu flag

Browse Source
This commit is contained in:
Concedo
2026-09-06 18:08:03 +08:00
parent 4e4d88141c
commit 8e6ba2630b
3 changed files with 35 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
gpttype_adapter.cpp
+18
View File
@@ -3403,6 +3403,24 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
}
printf("Overriding %d MoE layers to CPU...\n",inputs.moecpu);
}
if(ggml_backend_dev_count()>1 && inputs.ffncpu>0)
{
std::string toadd = "";
for (int i = 0; i < inputs.ffncpu; ++i) {
std::string tmp = string_format("blk\\.%d\\.ffn_(up|down|gate)\\.=CPU", i);
if(i>0)
{
tmp = "," + tmp;
}
toadd += tmp;
}
if (tensoroverrides == "") {
tensoroverrides = toadd;
} else {
tensoroverrides += "," + toadd;
}
printf("Overriding %d dense FFN layers to CPU...\n",inputs.ffncpu);
}
if(tensoroverrides!="" && ggml_backend_dev_count()>1)
{
printf("Handling Override Tensors for backends: ");