mirrors/llama.cpp
0
0 Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-10-28 09:45:55 +08:00
Code Activity

Enable tied weights with shared scales

Browse Source
This commit is contained in:
Oliver Simons committed 2026-10-27 18:47:46 +01:00
1 parent 3cb743d38c
commit f8999caf78
1 file changed
+2 -2
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-model.cpp
+2 -2
View File
@@ -1871,11 +1871,11 @@ bool llama_model_base::load_tensors(llama_model_loader & ml) {
// output scales
if (output) {
// weight scale
if (!output_s) {
if (!output_s && gguf_find_tensor(ml.metadata, tn(LLM_TENSOR_OUTPUT, "scale").str().c_str()) >= 0) {
output_s = create_tensor(tn(LLM_TENSOR_OUTPUT, "scale"), {1}, TENSOR_NOT_REQUIRED);
}
// input scale
if (!output_in_s) {
if (!output_in_s && gguf_find_tensor(ml.metadata, tn(LLM_TENSOR_OUTPUT, "input_scale").str().c_str()) >= 0) {
output_in_s = create_tensor(tn(LLM_TENSOR_OUTPUT, "input_scale"), {1}, TENSOR_NOT_REQUIRED);
}
}