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

Fix for the regen issue.

Browse source
This commit is contained in:
digiwombat 2023-05-30 15:48:55 -04:00
parent 42cf4d8433
commit 03ea8f013a
1 changed files with 4 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/server/server.cpp
View file

@ -66,6 +66,10 @@ 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
std::vector<llama_token> prompt_tokens = ::llama_tokenize(ctx, params.prompt, true); std::vector<llama_token> prompt_tokens = ::llama_tokenize(ctx, params.prompt, true);
if (prompt_tokens == last_prompt_tokens)
{
embd.clear();
}
// 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++) {
if (prompt_tokens[n_past] != processed_tokens[n_past]) { if (prompt_tokens[n_past] != processed_tokens[n_past]) {