vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 09:34:37 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	CMakeLists.txt
#	Makefile
#	README.md
#	common/CMakeLists.txt
#	docs/backend/SYCL.md
#	docs/build.md
#	docs/docker.md
#	examples/export-lora/export-lora.cpp
#	examples/main/README.md
#	examples/main/main.cpp
#	examples/run/README.md
#	examples/run/run.cpp
#	examples/server/README.md
#	examples/simple-chat/simple-chat.cpp
#	ggml/CMakeLists.txt
#	ggml/src/ggml-hip/CMakeLists.txt
#	src/CMakeLists.txt
#	tests/test-backend-ops.cpp
#	tests/test-chat-template.cpp
This commit is contained in:
Concedo 2025-01-25 14:16:50 +08:00
commit bec231422a
46 changed files with 4305 additions and 578 deletions
Download patch file Download diff file Expand all files Collapse all files

28
examples/main/main.cpp
View file

@ -4,6 +4,7 @@
#include "log.h"
#include "sampling.h"
#include "llama.h"
#include "chat-template.hpp"
#include "build-info.h"
#include <cstdio>
@ -85,14 +86,6 @@ static void sigint_handler(int signo) {
}
#endif
static std::string chat_add_and_format(struct llama_model * model, std::vector<common_chat_msg> & chat_msgs, const std::string & role, const std::string & content) {
common_chat_msg new_msg{role, content};
auto formatted = common_chat_format_single(model, g_params->chat_template, chat_msgs, new_msg, role == "user");
chat_msgs.push_back({role, content});
LOG_DBG("formatted: '%s'\n", formatted.c_str());
return formatted;
}
int main(int argc, char ** argv) {
common_params params;
g_params = &params;
@ -166,6 +159,7 @@ int main(int argc, char ** argv) {
}
const llama_vocab * vocab = llama_model_get_vocab(model);
auto chat_templates = common_chat_templates_from_model(model, params.chat_template);
LOG_INF("%s: llama threadpool init, n_threads = %d\n", __func__, (int) params.cpuparams.n_threads);
@ -208,7 +202,7 @@ int main(int argc, char ** argv) {
}
// auto enable conversation mode if chat template is available
const bool has_chat_template = !common_get_builtin_chat_template(model).empty() || !params.chat_template.empty();
const bool has_chat_template = chat_templates.has_explicit_template && chat_templates.template_default;
if (params.conversation_mode == COMMON_CONVERSATION_MODE_AUTO) {
if (has_chat_template) {
LOG_INF("%s: chat template is available, enabling conversation mode (disable it with -no-cnv)\n", __func__);
@ -226,7 +220,7 @@ int main(int argc, char ** argv) {
// print chat template example in conversation mode
if (params.conversation_mode) {
if (params.enable_chat_template) {
LOG_INF("%s: chat template example:\n%s\n", __func__, common_chat_format_example(model, params.chat_template).c_str());
LOG_INF("%s: chat template example:\n%s\n", __func__, common_chat_format_example(*chat_templates.template_default, params.use_jinja).c_str());
} else {
LOG_INF("%s: in-suffix/prefix is specified, chat template will be disabled\n", __func__);
}
@ -270,10 +264,18 @@ int main(int argc, char ** argv) {
std::vector<llama_token> embd_inp;
auto chat_add_and_format = [&chat_msgs, &chat_templates](const std::string & role, const std::string & content) {
common_chat_msg new_msg{role, content};
auto formatted = common_chat_format_single(*chat_templates.template_default, chat_msgs, new_msg, role == "user", g_params->use_jinja);
chat_msgs.push_back({role, content});
LOG_DBG("formatted: '%s'\n", formatted.c_str());
return formatted;
};
{
auto prompt = (params.conversation_mode && params.enable_chat_template)
// format the system prompt in conversation mode (fallback to default if empty)
? chat_add_and_format(model, chat_msgs, "system", params.prompt.empty() ? DEFAULT_SYSTEM_MESSAGE : params.prompt)
? chat_add_and_format("system", params.prompt.empty() ? DEFAULT_SYSTEM_MESSAGE : params.prompt)
// otherwise use the prompt as is
: params.prompt;
if (params.interactive_first || !params.prompt.empty() || session_tokens.empty()) {
@ -780,7 +782,7 @@ int main(int argc, char ** argv) {
}
if (params.enable_chat_template) {
chat_add_and_format(model, chat_msgs, "assistant", assistant_ss.str());
chat_add_and_format("assistant", assistant_ss.str());
}
is_interacting = true;
LOG("\n");
@ -845,7 +847,7 @@ int main(int argc, char ** argv) {
bool format_chat = params.conversation_mode && params.enable_chat_template;
std::string user_inp = format_chat
? chat_add_and_format(model, chat_msgs, "user", std::move(buffer))
? chat_add_and_format("user", std::move(buffer))
: std::move(buffer);
// TODO: one inconvenient of current chat template implementation is that we can't distinguish between user input and special tokens (prefix/postfix)
const auto line_pfx = common_tokenize(ctx, params.input_prefix, false, true);