vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fixed refusing to quantize some models

Browse source
This commit is contained in:
Concedo 2023-07-05 17:29:35 +08:00
parent 69add28324
commit ea79e549f0
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
llama.cpp
View file

@ -2404,9 +2404,9 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
int ny = tensor.ne.at(1); int ny = tensor.ne.at(1);
if (nx % QK_K != 0 || ny % QK_K != 0) { if (nx % QK_K != 0 || ny % QK_K != 0) {
fprintf(stderr, "\n\n========================= Tensor sizes %d x %d are not divisible by %d\n",nx,ny,QK_K); fprintf(stderr, "\n\n========================= Tensor sizes %d x %d are not divisible by %d\n",nx,ny,QK_K);
fprintf(stderr, "This is required to be able to use k-quants for now!\n"); fprintf(stderr, "Verify before using\n");
fprintf(stderr, "========================================================================================\n\n"); fprintf(stderr, "========================================================================================\n\n");
throw std::runtime_error("Unsupported tensor size encountered\n"); // throw std::runtime_error("Unsupported tensor size encountered\n");
} }
} }
if (tensor.name == "output.weight") { if (tensor.name == "output.weight") {