CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-31 08:51:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : skip loading unused tensors (#12004)

Browse Source 
* llama : assign unknown/unused tensors to host buffer type

ggml-ci

* llama : skip unused tensors

ggml-ci
This commit is contained in:
Georgi Gerganov
2025-02-21 18:33:18 +02:00
committed by GitHub
parent 586d5fe6eb
commit 51f311e057
1 changed files with 8 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/llama-model.cpp
View File

@@ -1424,6 +1424,14 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
throw std::runtime_error(format("missing tensor info mapping for %s", tn.str().c_str())); throw std::runtime_error(format("missing tensor info mapping for %s", tn.str().c_str()));
} }
// skip unused tensors
if (info.op == GGML_OP_NONE) {
LLAMA_LOG_WARN("model has unused tensor %s -- ignoring\n", tn.str().c_str());
ml.n_created++;
return nullptr;
}
// tensors with "bias" suffix are always used with GGML_OP_ADD // tensors with "bias" suffix are always used with GGML_OP_ADD
ggml_op op; ggml_op op;
bool bias = tn.suffix != nullptr && strcmp(tn.suffix, "bias") == 0; bool bias = tn.suffix != nullptr && strcmp(tn.suffix, "bias") == 0;