root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-27 20:04:35 +00:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

server : minor sync

Browse Source
This commit is contained in:
Georgi Gerganov 2023-10-19 15:03:24 +03:00
parent 9740824ba5
commit 325d1793f7
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 9 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
examples/server/server.cpp
View File

@ -748,6 +748,7 @@ struct llama_server_context
const stop_type type, llama_client_slot &slot) const stop_type type, llama_client_slot &slot)
{ {
size_t stop_pos = std::string::npos; size_t stop_pos = std::string::npos;
for (const std::string &word : slot.params.antiprompt) for (const std::string &word : slot.params.antiprompt)
{ {
size_t pos; size_t pos;
@ -774,6 +775,7 @@ struct llama_server_context
} }
} }
return stop_pos; return stop_pos;
} }
@ -798,8 +800,7 @@ struct llama_server_context
pos = std::min(slot.sent_count, slot.generated_text.size()); pos = std::min(slot.sent_count, slot.generated_text.size());
} else { } else {
is_stop_full = false; is_stop_full = false;
stop_pos = find_stopping_strings(str_test, token_str.size(), stop_pos = find_stopping_strings(str_test, token_str.size(), STOP_PARTIAL, slot);
STOP_PARTIAL, slot);
} }
// check if there is any token to predict // check if there is any token to predict
@ -2286,8 +2287,7 @@ int main(int argc, char **argv)
const json data = format_final_response(llama, slot, completion_text, probs); const json data = format_final_response(llama, slot, completion_text, probs);
slot_print_timings(slot); slot_print_timings(slot);
res.set_content(data.dump(-1, ' ', false, json::error_handler_t::replace), res.set_content(data.dump(-1, ' ', false, json::error_handler_t::replace), "application/json");
"application/json");
} }
else else
{ {
@ -2342,9 +2342,11 @@ int main(int argc, char **argv)
"data: " + "data: " +
data.dump(-1, ' ', false, json::error_handler_t::replace) + data.dump(-1, ' ', false, json::error_handler_t::replace) +
"\n\n"; "\n\n";
LOG_VERBOSE("data stream", { LOG_VERBOSE("data stream", {
{ "to_send", str } { "to_send", str }
}); });
if (!sink.write(str.data(), str.size())) if (!sink.write(str.data(), str.size()))
{ {
slot->release(); slot->release();