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

sync: minja https://github.com/google/minja/pull/46

Browse source
This commit is contained in:
Olivier Chafik 2025-02-03 23:31:49 +00:00
parent 1c302e18ba
commit 108da907f0
2 changed files with 162 additions and 78 deletions
Download patch file Download diff file Expand all files Collapse all files

232
common/chat-template.hpp
View file

@ -33,6 +33,29 @@ struct chat_template_caps {
bool requires_typed_content = false; bool requires_typed_content = false;
}; };
struct chat_template_inputs {
nlohmann::ordered_json messages;
nlohmann::ordered_json tools;
bool add_generation_prompt = true;
nlohmann::ordered_json extra_context;
std::chrono::system_clock::time_point now = std::chrono::system_clock::now();
};
struct chat_template_options {
bool apply_polyfills = true;
bool use_bos_token = true;
bool use_eos_token = true;
bool define_strftime_now = true;
bool polyfill_tools = true;
bool polyfill_tool_call_examples = true;
bool polyfill_tool_calls = true;
bool polyfill_tool_responses = true;
bool polyfill_system_role = true;
bool polyfill_object_arguments = true;
bool polyfill_typed_content = true;
};
class chat_template { class chat_template {
private: private:
@ -50,7 +73,18 @@ class chat_template {
const nlohmann::ordered_json & extra_context = nlohmann::ordered_json()) const const nlohmann::ordered_json & extra_context = nlohmann::ordered_json()) const
{ {
try { try {
auto prompt = apply(messages, tools, add_generation_prompt, extra_context, /* adjust_inputs= */ false); chat_template_inputs inputs;
inputs.messages = messages;
inputs.tools = tools;
inputs.add_generation_prompt = add_generation_prompt;
inputs.extra_context = extra_context;
// Use fixed date for tests
inputs.now = std::chrono::system_clock::from_time_t(0);
chat_template_options opts;
opts.apply_polyfills = false;
auto prompt = apply(inputs, opts);
// fprintf(stderr, "try_raw_render: %s\n", prompt.c_str()); // fprintf(stderr, "try_raw_render: %s\n", prompt.c_str());
return prompt; return prompt;
} catch (const std::exception & e) { } catch (const std::exception & e) {
@ -178,44 +212,56 @@ class chat_template {
caps_.supports_tool_call_id = contains(out, "call_911_"); caps_.supports_tool_call_id = contains(out, "call_911_");
} }
if (!caps_.supports_tools) { try {
const json user_msg { if (!caps_.supports_tools) {
{"role", "user"}, const json user_msg {
{"content", "Hey"}, {"role", "user"},
}; {"content", "Hey"},
const json tool_call_msg { };
{"role", "assistant"}, const json args {
{"content", nullptr}, {"arg1", "some_value"},
{"tool_calls", json::array({ };
{ const json tool_call_msg {
// TODO: detect if requires numerical id or fixed length == 6 like Nemo {"role", "assistant"},
{"id", "call_1___"}, {"content", nullptr},
{"type", "function"}, {"tool_calls", json::array({
{"function", { {
{"name", "tool_name"}, // TODO: detect if requires numerical id or fixed length == 6 like Nemo
{"arguments", (json { {"id", "call_1___"},
{"arg1", "some_value"}, {"type", "function"},
}).dump()}, {"function", {
}}, {"name", "tool_name"},
}, {"arguments", (caps_.requires_object_arguments ? args : json(minja::Value(args).dump(-1, /* to_json= */ true)))},
})}, }},
}; },
const json tools; })},
auto prefix = apply(json::array({user_msg}), tools, /* add_generation_prompt= */ true); };
auto full = apply(json::array({user_msg, tool_call_msg}), tools, /* add_generation_prompt= */ false); std::string prefix, full;
if (full.find(prefix) != 0 && prefix.length() > 0 && prefix[prefix.length() - 1] == '\n') { {
prefix = prefix.substr(0, prefix.length() - 1); chat_template_inputs inputs;
} inputs.messages = json::array({user_msg});
if (full.find(prefix) != 0) { inputs.add_generation_prompt = true;
if (prefix.rfind(eos_token_) == prefix.size() - eos_token_.size()) { prefix = apply(inputs);
prefix = prefix.substr(0, prefix.size() - eos_token_.size()); }
} else { {
throw std::runtime_error("prefix not found at start of full: " + prefix + " vs " + full); chat_template_inputs inputs;
inputs.messages = json::array({user_msg, tool_call_msg});
inputs.add_generation_prompt = false;
full = apply(inputs);
} }
} else {
if (full.find(prefix) != 0) {
if (prefix.rfind(eos_token_) == prefix.size() - eos_token_.size()) {
prefix = prefix.substr(0, prefix.size() - eos_token_.size());
}
}
if (full.find(prefix) != 0) {
fprintf(stderr, "Failed to infer a tool call example (possible template bug)\n");
}
tool_call_example_ = full.substr(prefix.size());
} }
tool_call_example_ = full.substr(prefix.size()); } catch (const std::exception & e) {
fprintf(stderr, "Failed to generate tool call example: %s\n", e.what());
} }
} }
@ -225,27 +271,49 @@ class chat_template {
const chat_template_caps & original_caps() const { return caps_; } const chat_template_caps & original_caps() const { return caps_; }
std::string apply( std::string apply(
const nlohmann::ordered_json & messages, const chat_template_inputs & inputs,
const nlohmann::ordered_json & tools, const chat_template_options & opts = chat_template_options()) const
bool add_generation_prompt,
const nlohmann::ordered_json & extra_context = nlohmann::ordered_json(),
bool adjust_inputs = true) const
{ {
json actual_messages; json actual_messages;
auto needs_adjustments = adjust_inputs && (false auto has_tools = inputs.tools.is_array() && !inputs.tools.empty();
|| !caps_.supports_system_role auto has_tool_calls = false;
|| !caps_.supports_tools auto has_tool_responses = false;
|| !caps_.supports_tool_responses auto has_string_content = false;
|| !caps_.supports_tool_calls for (const auto & message : inputs.messages) {
|| caps_.requires_object_arguments if (!message["tool_calls"].is_null()) {
|| caps_.requires_typed_content has_tool_calls = true;
}
if (message["role"] == "tool") {
has_tool_responses = true;
}
if (message["content"].is_string()) {
has_string_content = true;
}
}
auto polyfill_system_role = opts.polyfill_system_role && !caps_.supports_system_role;
auto polyfill_tools = opts.polyfill_tools && has_tools && !caps_.supports_tools;
auto polyfill_tool_call_example = polyfill_tools && opts.polyfill_tool_call_examples;
auto polyfill_tool_calls = opts.polyfill_tool_calls && has_tool_calls && !caps_.supports_tool_calls;
auto polyfill_tool_responses = opts.polyfill_tool_responses && has_tool_responses && !caps_.supports_tool_responses;
auto polyfill_object_arguments = opts.polyfill_object_arguments && has_tool_calls && caps_.requires_object_arguments;
auto polyfill_typed_content = opts.polyfill_typed_content && has_string_content && caps_.requires_typed_content;
auto needs_polyfills = opts.apply_polyfills && (false
|| polyfill_system_role
|| polyfill_tools
|| polyfill_tool_calls
|| polyfill_tool_responses
|| polyfill_object_arguments
|| polyfill_typed_content
); );
if (needs_adjustments) {
if (needs_polyfills) {
actual_messages = json::array(); actual_messages = json::array();
auto add_message = [&](const json & msg) { auto add_message = [&](const json & msg) {
if (caps_.requires_typed_content && msg.contains("content") && !msg.at("content").is_null() && msg.at("content").is_string()) { if (polyfill_typed_content && msg.contains("content") && !msg.at("content").is_null() && msg.at("content").is_string()) {
actual_messages.push_back({ actual_messages.push_back({
{"role", msg.at("role")}, {"role", msg.at("role")},
{"content", {{ {"content", {{
@ -268,16 +336,14 @@ class chat_template {
pending_system.clear(); pending_system.clear();
} }
}; };
auto needs_tools_in_system = !tools.is_null() && tools.size() > 0 && !caps_.supports_tools;
json adjusted_messages; json adjusted_messages;
if (needs_tools_in_system) { if (polyfill_tools) {
adjusted_messages = add_system(messages, adjusted_messages = add_system(inputs.messages,
"\n\n" "You can call any of the following tools to satisfy the user's requests: " + minja::Value(inputs.tools).dump(2, /* to_json= */ true) +
"You can call any of the following tools to satisfy the user's requests: " + tools.dump(2) + "\n\n" (!polyfill_tool_call_example || tool_call_example_.empty() ? "" : "\n\nExample tool call syntax:\n\n" + tool_call_example_));
"Example tool call syntax:\n\n" + tool_call_example_ + "\n\n");
} else { } else {
adjusted_messages = messages; adjusted_messages = inputs.messages;
} }
for (const auto & message_ : adjusted_messages) { for (const auto & message_ : adjusted_messages) {
@ -288,7 +354,7 @@ class chat_template {
std::string role = message.at("role"); std::string role = message.at("role");
if (message.contains("tool_calls")) { if (message.contains("tool_calls")) {
if (caps_.requires_object_arguments || !caps_.supports_tool_calls) { if (polyfill_object_arguments || polyfill_tool_calls) {
for (auto & tool_call : message.at("tool_calls")) { for (auto & tool_call : message.at("tool_calls")) {
if (tool_call["type"] == "function") { if (tool_call["type"] == "function") {
auto & function = tool_call.at("function"); auto & function = tool_call.at("function");
@ -303,7 +369,7 @@ class chat_template {
} }
} }
} }
if (!caps_.supports_tool_calls) { if (polyfill_tool_calls) {
auto content = message.at("content"); auto content = message.at("content");
auto tool_calls = json::array(); auto tool_calls = json::array();
for (const auto & tool_call : message.at("tool_calls")) { for (const auto & tool_call : message.at("tool_calls")) {
@ -330,7 +396,7 @@ class chat_template {
message.erase("tool_calls"); message.erase("tool_calls");
} }
} }
if (!caps_.supports_tool_responses && role == "tool") { if (polyfill_tool_responses && role == "tool") {
message["role"] = "user"; message["role"] = "user";
auto obj = json { auto obj = json {
{"tool_response", { {"tool_response", {
@ -347,7 +413,7 @@ class chat_template {
message.erase("name"); message.erase("name");
} }
if (!message["content"].is_null() && !caps_.supports_system_role) { if (!message["content"].is_null() && polyfill_system_role) {
std::string content = message.at("content"); std::string content = message.at("content");
if (role == "system") { if (role == "system") {
if (!pending_system.empty()) pending_system += "\n"; if (!pending_system.empty()) pending_system += "\n";
@ -366,28 +432,40 @@ class chat_template {
} }
add_message(message); add_message(message);
} }
if (!caps_.supports_system_role) { flush_sys();
flush_sys();
}
} else { } else {
actual_messages = messages; actual_messages = inputs.messages;
} }
auto context = minja::Context::make(json({ auto context = minja::Context::make(json({
{"messages", actual_messages}, {"messages", actual_messages},
{"add_generation_prompt", add_generation_prompt}, {"add_generation_prompt", inputs.add_generation_prompt},
{"bos_token", bos_token_},
{"eos_token", eos_token_},
})); }));
if (opts.use_bos_token) {
if (!tools.is_null()) { context->set("bos_token", bos_token_);
auto tools_val = minja::Value(tools);
context->set("tools", tools_val);
} }
if (!extra_context.is_null()) { if (opts.use_eos_token) {
for (auto & kv : extra_context.items()) { context->set("eos_token", eos_token_);
minja::Value val(kv.value()); }
context->set(kv.key(), val); if (opts.define_strftime_now) {
auto now = inputs.now;
context->set("strftime_now", Value::callable([now](const std::shared_ptr<minja::Context> &, minja::ArgumentsValue & args) {
args.expectArgs("strftime_now", {1, 1}, {0, 0});
auto format = args.args[0].get<std::string>();
auto time = std::chrono::system_clock::to_time_t(now);
auto local_time = *std::localtime(&time);
std::ostringstream ss;
ss << std::put_time(&local_time, format.c_str());
return ss.str();
}));
}
if (!inputs.tools.is_null()) {
context->set("tools", minja::Value(inputs.tools));
}
if (!inputs.extra_context.is_null()) {
for (auto & kv : inputs.extra_context.items()) {
context->set(kv.key(), minja::Value(kv.value()));
} }
} }
@ -404,7 +482,7 @@ class chat_template {
std::string existing_system = messages_with_system.at(0).at("content"); std::string existing_system = messages_with_system.at(0).at("content");
messages_with_system[0] = json { messages_with_system[0] = json {
{"role", "system"}, {"role", "system"},
{"content", existing_system + "\n" + system_prompt}, {"content", existing_system + "\n\n" + system_prompt},
}; };
} else { } else {
messages_with_system.insert(messages_with_system.begin(), json { messages_with_system.insert(messages_with_system.begin(), json {

8
common/minja.hpp
View file

@ -2194,7 +2194,7 @@ private:
} }
TemplateTokenVector tokenize() { TemplateTokenVector tokenize() {
static std::regex comment_tok(R"(\{#([-~]?)(.*?)([-~]?)#\})"); static std::regex comment_tok(R"(\{#([-~]?)([\s\S\r\n]*?)([-~]?)#\})");
static std::regex expr_open_regex(R"(\{\{([-~])?)"); static std::regex expr_open_regex(R"(\{\{([-~])?)");
static std::regex block_open_regex(R"(^\{%([-~])?[\s\n\r]*)"); static std::regex block_open_regex(R"(^\{%([-~])?[\s\n\r]*)");
static std::regex block_keyword_tok(R"((if|else|elif|endif|for|endfor|generation|endgeneration|set|endset|block|endblock|macro|endmacro|filter|endfilter|break|continue)\b)"); static std::regex block_keyword_tok(R"((if|else|elif|endif|for|endfor|generation|endgeneration|set|endset|block|endblock|macro|endmacro|filter|endfilter|break|continue)\b)");
@ -2615,6 +2615,7 @@ inline std::shared_ptr<Context> Context::builtins() {
})); }));
globals.set("join", simple_function("join", { "items", "d" }, [](const std::shared_ptr<Context> &, Value & args) { globals.set("join", simple_function("join", { "items", "d" }, [](const std::shared_ptr<Context> &, Value & args) {
auto do_join = [](Value & items, const std::string & sep) { auto do_join = [](Value & items, const std::string & sep) {
if (!items.is_array()) throw std::runtime_error("object is not iterable: " + items.dump());
std::ostringstream oss; std::ostringstream oss;
auto first = true; auto first = true;
for (size_t i = 0, n = items.size(); i < n; ++i) { for (size_t i = 0, n = items.size(); i < n; ++i) {
@ -2695,6 +2696,10 @@ inline std::shared_ptr<Context> Context::builtins() {
return Value::callable([=](const std::shared_ptr<Context> & context, ArgumentsValue & args) { return Value::callable([=](const std::shared_ptr<Context> & context, ArgumentsValue & args) {
args.expectArgs(is_select ? "select" : "reject", {2, (std::numeric_limits<size_t>::max)()}, {0, 0}); args.expectArgs(is_select ? "select" : "reject", {2, (std::numeric_limits<size_t>::max)()}, {0, 0});
auto & items = args.args[0]; auto & items = args.args[0];
if (items.is_null())
return Value::array();
if (!items.is_array()) throw std::runtime_error("object is not iterable: " + items.dump());
auto filter_fn = context->get(args.args[1]); auto filter_fn = context->get(args.args[1]);
if (filter_fn.is_null()) throw std::runtime_error("Undefined filter: " + args.args[1].dump()); if (filter_fn.is_null()) throw std::runtime_error("Undefined filter: " + args.args[1].dump());
@ -2772,6 +2777,7 @@ inline std::shared_ptr<Context> Context::builtins() {
auto & items = args.args[0]; auto & items = args.args[0];
if (items.is_null()) if (items.is_null())
return Value::array(); return Value::array();
if (!items.is_array()) throw std::runtime_error("object is not iterable: " + items.dump());
auto attr_name = args.args[1].get<std::string>(); auto attr_name = args.args[1].get<std::string>();
bool has_test = false; bool has_test = false;