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

Forgot to remove some testing code.

Browse source
This commit is contained in:
Randall Fitzgerald 2023-05-23 06:22:30 -07:00 • committed by Henri Vasserman
parent f93fe36c5b
commit df0e0d094c
No known key found for this signature in database
GPG key ID: 2995FC0F58B1A986
1 changed files with 0 additions and 18 deletions
Download patch file Download diff file Expand all files Collapse all files

18
examples/server/server.cpp
View file

@ -13,8 +13,6 @@ struct llama_server_context
{ {
bool as_loop = false; bool as_loop = false;
bool has_next_token = false; bool has_next_token = false;
std::string generated_text = ""; std::string generated_text = "";
size_t num_tokens_predicted = 0; size_t num_tokens_predicted = 0;
@ -65,21 +63,6 @@ struct llama_server_context
bool loadPrompt() { bool loadPrompt() {
params.prompt.insert(0, 1, ' '); // always add a first space params.prompt.insert(0, 1, ' '); // always add a first space
if(processed_tokens.size() != 0)
{
processed_tokens.erase(processed_tokens.begin() + 1, processed_tokens.end());
}
if(embd_inp.size() != 0)
{
embd_inp.erase(embd_inp.begin() + 1, embd_inp.end());
}
n_remain = 0;
n_past = 0;
n_consumed = 0;
std::vector<llama_token> prompt_tokens = ::llama_tokenize(ctx, params.prompt, true); std::vector<llama_token> prompt_tokens = ::llama_tokenize(ctx, params.prompt, true);
// compare the evaluated prompt with the new prompt // compare the evaluated prompt with the new prompt
for (n_past = 0; n_past < prompt_tokens.size() - 1 && n_past < processed_tokens.size(); n_past++) { for (n_past = 0; n_past < prompt_tokens.size() - 1 && n_past < processed_tokens.size(); n_past++) {
@ -117,7 +100,6 @@ struct llama_server_context
// Reset context // Reset context
const int n_left = n_past - params.n_keep; const int n_left = n_past - params.n_keep;
n_past = std::max(1, params.n_keep); n_past = std::max(1, params.n_keep);
last_n_tokens.erase(last_n_tokens.begin() + n_past, last_n_tokens.end());
processed_tokens.erase(processed_tokens.begin() + n_past, processed_tokens.end()); processed_tokens.erase(processed_tokens.begin() + n_past, processed_tokens.end());
embd.insert(embd.begin(), last_n_tokens.begin() + params.n_ctx - n_left / 2 - embd.size(), last_n_tokens.end() - embd.size()); embd.insert(embd.begin(), last_n_tokens.begin() + params.n_ctx - n_left / 2 - embd.size(), last_n_tokens.end() - embd.size());
} }