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

Bump model_template to 16384 bytes to support larger chat templates.

Browse source
This commit is contained in:
DAN™ 2024-12-17 11:02:26 -05:00
parent 5437d4aaf5
commit 919fe432c3
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
src/llama.cpp
View file

@ -22652,7 +22652,7 @@ int32_t llama_chat_apply_template(
if (tmpl == nullptr) { if (tmpl == nullptr) {
GGML_ASSERT(model != nullptr); GGML_ASSERT(model != nullptr);
// load template from model // load template from model
std::vector<char> model_template(2048, 0); // longest known template is about 1200 bytes std::vector<char> model_template(16384, 0); // longest known template is about 12792 bytes
std::string template_key = "tokenizer.chat_template"; std::string template_key = "tokenizer.chat_template";
int32_t res = llama_model_meta_val_str(model, template_key.c_str(), model_template.data(), model_template.size()); int32_t res = llama_model_meta_val_str(model, template_key.c_str(), model_template.data(), model_template.size());
if (res < 0) { if (res < 0) {