root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-25 10:54:36 +00:00
Code Issues Actions 9 Packages Projects Releases Wiki Activity

quantize : --pure option for disabling k-quant mixtures

Browse Source
This commit is contained in:
cebtenzzre 2023-10-28 16:32:49 -04:00
parent ee37e35dc5
commit 8a86b95e87
3 changed files with 10 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
examples/quantize/quantize.cpp
View File

@ -68,13 +68,14 @@ static bool try_parse_ftype(const std::string & ftype_str_in, llama_ftype & ftyp
}
// usage:
// ./quantize [--allow-requantize] [--leave-output-tensor] models/llama/ggml-model.gguf [models/llama/ggml-model-quant.gguf] type [nthreads]
// ./quantize [--allow-requantize] [--leave-output-tensor] [--pure] models/llama/ggml-model.gguf [models/llama/ggml-model-quant.gguf] type [nthreads]
//
[[noreturn]]
static void usage(const char * executable) {
printf("usage: %s [--help] [--allow-requantize] [--leave-output-tensor] model-f32.gguf [model-quant.gguf] type [nthreads]\n\n", executable);
printf("usage: %s [--help] [--allow-requantize] [--leave-output-tensor] [--pure] model-f32.gguf [model-quant.gguf] type [nthreads]\n\n", executable);
printf(" --allow-requantize: Allows requantizing tensors that have already been quantized. Warning: This can severely reduce quality compared to quantizing from 16bit or 32bit\n");
printf(" --leave-output-tensor: Will leave output.weight un(re)quantized. Increases model size but may also increase quality, especially when requantizing\n");
printf(" --pure: Disable k-quant mixtures and quantize all tensors to the same type\n");
printf("\nAllowed quantization types:\n");
for (auto & it : QUANT_OPTIONS) {
if (it.name != "COPY") {
@ -101,6 +102,8 @@ int main(int argc, char ** argv) {
params.quantize_output_tensor = false;
} else if (strcmp(argv[arg_idx], "--allow-requantize") == 0) {
params.allow_requantize = true;
} else if (strcmp(argv[arg_idx], "--pure") == 0) {
params.pure = true;
} else {
usage(argv[0]);
}

5
llama.cpp
View File

@ -8380,7 +8380,9 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
if (quantize) {
new_type = quantized_type;
new_type = get_k_quant_type(qs, new_type, tensor, ftype);
if (!params->pure) {
new_type = get_k_quant_type(qs, new_type, tensor, ftype);
}
// If we've decided to quantize to the same type the tensor is already
// in then there's nothing to do.
@ -8835,6 +8837,7 @@ struct llama_model_quantize_params llama_model_quantize_default_params() {
/*.allow_requantize =*/ false,
/*.quantize_output_tensor =*/ true,
/*.only_copy =*/ false,
/*.pure =*/ false,
};
return result;

1
llama.h
View File

@ -191,6 +191,7 @@ extern "C" {
bool allow_requantize; // allow quantizing non-f32/f16 tensors
bool quantize_output_tensor; // quantize output.weight
bool only_copy; // only copy tensors - ftype, allow_requantize and quantize_output_tensor are ignored
bool pure; // disable k-quant mixtures and quantize all tensors to the same type
} llama_model_quantize_params;
// grammar types