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

cleanup

Browse source
This commit is contained in:
Concedo 2023-06-22 10:38:27 +08:00
parent 1b71752a9f
commit e6ddb15c3a
4 changed files with 10 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

1
otherarch/gpt2_v3.cpp
View file

@ -347,6 +347,7 @@ ModelLoadResult gpt2_model_load(const std::string & fname, gpt2_model & model, g
//gpu offload //gpu offload
#if defined(GGML_USE_CLBLAST) #if defined(GGML_USE_CLBLAST)
if(gpulayers>0)
{ {
const auto & hparams = model.hparams; const auto & hparams = model.hparams;
size_t vram_total = 0; size_t vram_total = 0;

1
otherarch/gptj_v3.cpp
View file

@ -335,6 +335,7 @@ ModelLoadResult gptj_model_load(const std::string & fname, gptj_model & model, g
//gpu offload //gpu offload
#if defined(GGML_USE_CLBLAST) #if defined(GGML_USE_CLBLAST)
if(gpulayers>0)
{ {
const auto & hparams = model.hparams; const auto & hparams = model.hparams;
size_t vram_total = 0; size_t vram_total = 0;

5
otherarch/mpt_v3.cpp
View file

@ -15,7 +15,9 @@
#include "model_adapter.h" #include "model_adapter.h"
#if defined(GGML_USE_CLBLAST)
#include "ggml-opencl.h"
#endif
// load the model's weights from a file // load the model's weights from a file
bool mpt_model_load(const std::string & fname, mpt_model & model, gpt_vocab & vocab, int gpulayers) { bool mpt_model_load(const std::string & fname, mpt_model & model, gpt_vocab & vocab, int gpulayers) {
@ -280,6 +282,7 @@ bool mpt_model_load(const std::string & fname, mpt_model & model, gpt_vocab & vo
//gpu offload //gpu offload
#if defined(GGML_USE_CLBLAST) #if defined(GGML_USE_CLBLAST)
if(gpulayers>0)
{ {
const auto & hparams = model.hparams; const auto & hparams = model.hparams;
size_t vram_total = 0; size_t vram_total = 0;

5
otherarch/neox_v3.cpp
View file

@ -13,7 +13,9 @@
#include <vector> #include <vector>
#include <iostream> #include <iostream>
#if defined(GGML_USE_CLBLAST)
#include "ggml-opencl.h"
#endif
// load the model's weights from a file // load the model's weights from a file
ModelLoadResult gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_vocab & vocab, FileFormat file_format, int gpulayers) { ModelLoadResult gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_vocab & vocab, FileFormat file_format, int gpulayers) {
@ -320,6 +322,7 @@ ModelLoadResult gpt_neox_model_load(const std::string & fname, gpt_neox_model &
//gpu offload //gpu offload
#if defined(GGML_USE_CLBLAST) #if defined(GGML_USE_CLBLAST)
if(gpulayers>0)
{ {
const auto & hparams = model.hparams; const auto & hparams = model.hparams;
size_t vram_total = 0; size_t vram_total = 0;