CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-31 08:51:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update common_chat_format_example to use minja template wrapper

Browse Source
This commit is contained in:
ochafik
2025-01-13 22:57:35 +00:00
parent c04c50e40c
commit a6afb2735f
4 changed files with 17 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
common/common.cpp
View File

@@ -1811,15 +1811,23 @@ std::string common_chat_format_single(const struct llama_model * model,
return ss.str();
}
std::string common_chat_format_example(const struct llama_model * model,
const std::string & tmpl) {
std::string common_chat_format_example(const struct llama_model * model, const minja::chat_template & tmpl, bool use_jinja) {
std::vector<common_chat_msg> msgs = {
{"system", "You are a helpful assistant"},
{"user", "Hello"},
{"assistant", "Hi there"},
{"user", "How are you?"},
};
return common_chat_apply_template(model, tmpl, msgs, true);
const auto add_generation_prompt = true;
if (use_jinja) {
auto messages = json::array();
for (const auto & msg : msgs) {
messages.push_back({{"role", msg.role}, {"content", msg.content}});
}
return tmpl.apply(messages, /* tools= */ json(), add_generation_prompt);
} else {
return common_chat_apply_template(model, tmpl.source(), msgs, add_generation_prompt);
}
}
llama_chat_templates llama_chat_templates_from_model(const struct llama_model * model, const std::string & chat_template_override)

2
common/common.h
View File

@@ -619,7 +619,7 @@ std::string common_chat_format_single(const struct llama_model * model,
// Returns an example of formatted chat
std::string common_chat_format_example(const struct llama_model * model,
const std::string & tmpl);
const minja::chat_template & tmpl, bool use_jinja);
struct llama_chat_templates {

5
examples/main/main.cpp
View File

@@ -165,6 +165,7 @@ int main(int argc, char ** argv) {
}
const llama_vocab * vocab = llama_model_get_vocab(model);
auto chat_templates = llama_chat_templates_from_model(model, params.chat_template);
LOG_INF("%s: llama threadpool init, n_threads = %d\n", __func__, (int) params.cpuparams.n_threads);
@@ -207,7 +208,7 @@ 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.default_template.source().empty();
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__);
@@ -225,7 +226,7 @@ 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(model, chat_templates.default_template, params.use_jinja).c_str());
} else {
LOG_INF("%s: in-suffix/prefix is specified, chat template will be disabled\n", __func__);
}

4
examples/server/server.cpp
View File

@@ -4287,8 +4287,8 @@ int main(int argc, char ** argv) {
// print sample chat example to make it clear which template is used
LOG_INF("%s: chat template, chat_template: %s, example_format: '%s'\n", __func__,
params.chat_template.empty() ? "(built-in)" : params.chat_template.c_str(),
common_chat_format_example(ctx_server.model, params.chat_template).c_str());
get_chat_templates().default_template.source().c_str(),
common_chat_format_example(ctx_server.model, get_chat_templates().default_template, ctx_server.params_base.use_jinja).c_str());
ctx_server.queue_tasks.on_new_task(std::bind(
&server_context::process_single_task, &ctx_server, std::placeholders::_1));