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

consolidate gen_chatcmplid() callsite

Browse source
This commit is contained in:
Minsoo Cheong 2024-03-05 23:11:35 +09:00
parent aa95dc5568
commit 293378b20b
2 changed files with 5 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/server/oai.hpp
View file

@ -69,7 +69,7 @@ inline static json oaicompat_completion_params_parse(
return llama_params; return llama_params;
} }
inline static json format_final_response_oaicompat(const json &request, const task_result &response, bool streaming = false) inline static json format_final_response_oaicompat(const json &request, const task_result &response, std::string id, bool streaming = false)
{ {
json result = response.result_json; json result = response.result_json;
@ -105,7 +105,7 @@ inline static json format_final_response_oaicompat(const json &request, const ta
json{{"completion_tokens", num_tokens_predicted}, json{{"completion_tokens", num_tokens_predicted},
{"prompt_tokens", num_prompt_tokens}, {"prompt_tokens", num_prompt_tokens},
{"total_tokens", num_tokens_predicted + num_prompt_tokens}}}, {"total_tokens", num_tokens_predicted + num_prompt_tokens}}},
{"id", gen_chatcmplid()}}; {"id", id}};
if (server_verbose) { if (server_verbose) {
res["__verbose"] = result; res["__verbose"] = result;

6
examples/server/server.cpp
View file

@ -3221,13 +3221,14 @@ int main(int argc, char **argv)
const int task_id = llama.queue_tasks.get_new_id(); const int task_id = llama.queue_tasks.get_new_id();
llama.queue_results.add_waiting_task_id(task_id); llama.queue_results.add_waiting_task_id(task_id);
llama.request_completion(task_id, data, false, false, -1); llama.request_completion(task_id, data, false, false, -1);
const std::string completion_id = gen_chatcmplid();
if (!json_value(data, "stream", false)) { if (!json_value(data, "stream", false)) {
std::string completion_text; std::string completion_text;
task_result result = llama.queue_results.recv(task_id); task_result result = llama.queue_results.recv(task_id);
if (!result.error && result.stop) { if (!result.error && result.stop) {
json oaicompat_result = format_final_response_oaicompat(data, result); json oaicompat_result = format_final_response_oaicompat(data, result, completion_id);
res.set_content(oaicompat_result.dump(-1, ' ', false, res.set_content(oaicompat_result.dump(-1, ' ', false,
json::error_handler_t::replace), json::error_handler_t::replace),
@ -3238,8 +3239,7 @@ int main(int argc, char **argv)
} }
llama.queue_results.remove_waiting_task_id(task_id); llama.queue_results.remove_waiting_task_id(task_id);
} else { } else {
const auto chunked_content_provider = [task_id, &llama](size_t, httplib::DataSink &sink) { const auto chunked_content_provider = [task_id, &llama, completion_id](size_t, httplib::DataSink &sink) {
const std::string completion_id = gen_chatcmplid();
while (true) { while (true) {
task_result llama_result = llama.queue_results.recv(task_id); task_result llama_result = llama.queue_results.recv(task_id);
if (!llama_result.error) { if (!llama_result.error) {