ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-05-01 03:41:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix incomplete utf-8 characters in streaming text completions (#810)

Browse Source
This commit is contained in:
Viktor Ivakin
2025-10-13 16:25:29 +03:00
committed by GitHub
parent 9724ea9213
commit 41bdd86555
1 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
examples/server/server.cpp
View File

@@ -1968,7 +1968,9 @@ struct server_context {
slot.generated_text.erase( slot.generated_text.erase(
slot.generated_text.begin() + pos + stop_pos, slot.generated_text.begin() + pos + stop_pos,
slot.generated_text.end()); slot.generated_text.end());
pos = std::min(slot.n_sent_text, slot.generated_text.size()); // Update n_sent_text to not exceed the new generated_text size
slot.n_sent_text = std::min(slot.n_sent_text, slot.generated_text.size());
pos = slot.n_sent_text;
} else { } else {
is_stop_full = false; is_stop_full = false;
stop_pos = slot.find_stopping_strings(str_test, token_str.size(), false); stop_pos = slot.find_stopping_strings(str_test, token_str.size(), false);
@@ -1980,6 +1982,9 @@ struct server_context {
result.text_to_send = slot.generated_text.substr(pos, std::string::npos); result.text_to_send = slot.generated_text.substr(pos, std::string::npos);
slot.n_sent_text += result.text_to_send.size(); slot.n_sent_text += result.text_to_send.size();
// add the token to slot queue and cache // add the token to slot queue and cache
} else if (stop_pos != std::string::npos) {
// Handle partial stop - update n_sent_text to the end of the current text
slot.n_sent_text = slot.generated_text.size();
} }
slot.add_token_string(result); slot.add_token_string(result);