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

added it in more places

Browse source
This commit is contained in:
Justin 2023-03-15 02:33:56 +00:00
parent e3d780d833
commit 9eb4598fa3
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
main.cpp
View file

@ -218,7 +218,7 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
ctx_size += n_ctx*n_layer*n_embd*ggml_type_sizef(GGML_TYPE_F32); // memory_k
ctx_size += n_ctx*n_layer*n_embd*ggml_type_sizef(GGML_TYPE_F32); // memory_v
ctx_size += (5 + 10*n_layer)*256; // object overhead
ctx_size += (5 + 10*n_layer)*hparams.n_ctx; // object overhead
fprintf(stderr, "%s: ggml ctx size = %6.2f MB\n", __func__, ctx_size/(1024.0*1024.0));
}
@ -547,7 +547,7 @@ bool llama_eval(
const int d_key = n_embd/n_head;
static size_t buf_size = 512u*1024*1024;
static size_t buf_size = hparams.n_ctx*1024*1024;
static void * buf = malloc(buf_size);
if (mem_per_token > 0 && mem_per_token*N > buf_size) {