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

Minor

Browse source
This commit is contained in:
Georgi Gerganov 2023-03-13 18:33:23 +02:00 • committed by GitHub
parent efaa30ef11
commit 3419f88f11
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

7
main.cpp
View file

@ -73,11 +73,10 @@ struct llama_model {
bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab & vocab, int n_ctx) { bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab & vocab, int n_ctx) {
printf("%s: loading model from '%s' - please wait ...\n", __func__, fname.c_str()); printf("%s: loading model from '%s' - please wait ...\n", __func__, fname.c_str());
const size_t f_buf_size = 1024*1024; std::vector<char> f_buf(1024*1024);
std::vector<char> f_buf(f_buf_size);
auto fin = std::ifstream(fname, std::ios::binary); auto fin = std::ifstream(fname, std::ios::binary);
fin.rdbuf()->pubsetbuf(f_buf.data(), f_buf_size); fin.rdbuf()->pubsetbuf(f_buf.data(), f_buf.size());
if (!fin) { if (!fin) {
fprintf(stderr, "%s: failed to open '%s'\n", __func__, fname.c_str()); fprintf(stderr, "%s: failed to open '%s'\n", __func__, fname.c_str());
return false; return false;
@ -315,7 +314,7 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
printf("%s: loading model part %d/%d from '%s'\n", __func__, i+1, n_parts, fname_part.c_str()); printf("%s: loading model part %d/%d from '%s'\n", __func__, i+1, n_parts, fname_part.c_str());
fin = std::ifstream(fname_part, std::ios::binary); fin = std::ifstream(fname_part, std::ios::binary);
fin.rdbuf()->pubsetbuf(f_buf.data(), f_buf_size); fin.rdbuf()->pubsetbuf(f_buf.data(), f_buf.size());
fin.seekg(file_offset); fin.seekg(file_offset);
// load weights // load weights