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

tweak delta logic

Browse source
This commit is contained in:
Olivier Chafik 2025-02-03 17:57:38 +00:00
parent df3474e2c2
commit c397bd1f5f
1 changed files with 3 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

3
common/chat-template.hpp
View file

@ -203,6 +203,9 @@ class chat_template {
const json tools; const json tools;
auto prefix = apply(json::array({user_msg}), tools, /* add_generation_prompt= */ true); auto prefix = apply(json::array({user_msg}), tools, /* add_generation_prompt= */ true);
auto full = apply(json::array({user_msg, tool_call_msg}), tools, /* add_generation_prompt= */ false); auto full = apply(json::array({user_msg, tool_call_msg}), tools, /* add_generation_prompt= */ false);
if (full.find(prefix) != 0 && prefix.length() > 0 && prefix[prefix.length() - 1] == '\n') {
prefix = prefix.substr(0, prefix.length() - 1);
}
if (full.find(prefix) != 0) { if (full.find(prefix) != 0) {
if (prefix.rfind(eos_token_) == prefix.size() - eos_token_.size()) { if (prefix.rfind(eos_token_) == prefix.size() - eos_token_.size()) {
prefix = prefix.substr(0, prefix.size() - eos_token_.size()); prefix = prefix.substr(0, prefix.size() - eos_token_.size());