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

rename n_ctx to kv_size

Browse source
This commit is contained in:
Pierrick HYMBERT 2024-02-18 20:59:26 +01:00 • committed by Georgi Gerganov
parent ef96e8b1f7
commit 606873401c
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
48 changed files with 403 additions and 393 deletions
Download patch file Download diff file Expand all files Collapse all files

6
examples/lookup/lookup.cpp
View file

@ -47,8 +47,8 @@ int main(int argc, char ** argv){
std::vector<llama_token> inp;
inp = ::llama_tokenize(ctx, params.prompt, add_bos, true);
const int max_context_size = llama_n_ctx(ctx);
const int max_tokens_list_size = max_context_size - 4;
const int max_kv_size = llama_kv_size(ctx);
const int max_tokens_list_size = max_kv_size - 4;
if ((int) inp.size() > max_tokens_list_size) {
fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) inp.size(), max_tokens_list_size);
@ -86,7 +86,7 @@ int main(int argc, char ** argv){
std::vector<llama_token> draft;
llama_batch batch_tgt = llama_batch_init(params.n_ctx, 0, 1);
llama_batch batch_tgt = llama_batch_init(params.kv_size, 0, 1);
// debug
struct llama_kv_cache_view kvc_view = llama_kv_cache_view_init(ctx, 1);