|
|
@@ -56,7 +56,6 @@ struct server_params {
|
|
|
std::string hostname = "127.0.0.1";
|
|
|
std::vector<std::string> api_keys;
|
|
|
std::string public_path = "examples/server/public";
|
|
|
- std::string chat_template = "";
|
|
|
int32_t port = 8080;
|
|
|
int32_t read_timeout = 600;
|
|
|
int32_t write_timeout = 600;
|
|
|
@@ -427,16 +426,6 @@ struct llama_server_context
|
|
|
return true;
|
|
|
}
|
|
|
|
|
|
- void validate_model_chat_template(server_params & sparams) {
|
|
|
- llama_chat_message chat[] = {{"user", "test"}};
|
|
|
- std::vector<char> buf(1);
|
|
|
- int res = llama_chat_apply_template(model, nullptr, chat, 1, true, buf.data(), buf.size());
|
|
|
- if (res < 0) {
|
|
|
- LOG_ERROR("The chat template comes with this model is not yet supported, falling back to chatml. This may cause the model to output suboptimal responses", {});
|
|
|
- sparams.chat_template = "chatml";
|
|
|
- }
|
|
|
- }
|
|
|
-
|
|
|
void initialize() {
|
|
|
// create slots
|
|
|
all_slots_are_idle = true;
|
|
|
@@ -2535,7 +2524,6 @@ static void server_params_parse(int argc, char **argv, server_params &sparams, g
|
|
|
invalid_param = true;
|
|
|
break;
|
|
|
}
|
|
|
- sparams.chat_template = argv[i];
|
|
|
}
|
|
|
else if (arg == "--override-kv")
|
|
|
{
|
|
|
@@ -3008,11 +2996,6 @@ int main(int argc, char **argv) {
|
|
|
}
|
|
|
const auto model_meta = llama.model_meta();
|
|
|
|
|
|
- if (sparams.chat_template.empty()) { // custom chat template is not supplied
|
|
|
- // check if the template comes with the model is supported by us
|
|
|
- llama.validate_model_chat_template(sparams);
|
|
|
- }
|
|
|
-
|
|
|
// Middleware for API key validation
|
|
|
auto validate_api_key = [&sparams](const httplib::Request &req, httplib::Response &res) -> bool {
|
|
|
// If API key is not set, skip validation
|
Michael Yang