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

quantize: options for output and token embedding tensors qtype (#6239)

Browse source 
* quantize: be able to specify the output tensor type

* quantize: be able to specify the token embedding tensor type

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow 2024-03-22 19:47:14 +01:00 • committed by GitHub
parent dba1af6129
commit 1d0331c12a
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 61 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

16
llama.h
View file

@ -275,13 +275,15 @@ extern "C" {
// model quantization parameters
typedef struct llama_model_quantize_params {
int32_t nthread; // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
enum llama_ftype ftype; // quantize to this llama_ftype
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; // quantize all tensors to the default type
void * imatrix; // pointer to importance matrix data
int32_t nthread; // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
enum llama_ftype ftype; // quantize to this llama_ftype
enum ggml_type output_tensor_type; // output tensor type
enum ggml_type token_embedding_type; // itoken embeddings tensor type
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; // quantize all tensors to the default type
void * imatrix; // pointer to importance matrix data
} llama_model_quantize_params;
// grammar types