llama.cpp/common/chat-template.hpp

516 lines
22 KiB
C++
Raw Normal View History

Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
/*
Copyright 2024 Google LLC
Use of this source code is governed by an MIT-style
license that can be found in the LICENSE file or at
https://opensource.org/licenses/MIT.
*/
// SPDX-License-Identifier: MIT
#pragma once
#include "minja.hpp"
#include <json.hpp>
#include <string>
#include <vector>
using json = nlohmann::ordered_json;
namespace minja {
2025-01-30 11:30:27 +01:00
struct chat_template_caps {
bool supports_tools = false;
bool supports_tool_calls = false;
bool supports_tool_responses = false;
bool supports_system_role = false;
bool supports_parallel_tool_calls = false;
bool supports_tool_call_id = false;
// meta-llama/Llama-3.1-8B-Instruct expects arguments to be an object.
// Most other templates (and OpenAI's API) expect the arguments object to be stringified.
bool requires_object_arguments = false;
// CohereForAI/c4ai-command-r-plus simple variant
bool requires_non_null_content = false;
// MiniMaxAI/MiniMax-Text-01 special
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;
};
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
class chat_template {
private:
2025-01-30 11:30:27 +01:00
chat_template_caps caps_;
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
std::string source_;
std::string bos_token_;
std::string eos_token_;
std::shared_ptr<minja::TemplateNode> template_root_;
std::string tool_call_example_;
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
std::string try_raw_render(
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
const nlohmann::ordered_json & messages,
const nlohmann::ordered_json & tools,
bool add_generation_prompt,
const nlohmann::ordered_json & extra_context = nlohmann::ordered_json()) const
{
try {
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);
2025-01-30 11:30:27 +01:00
// fprintf(stderr, "try_raw_render: %s\n", prompt.c_str());
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
return prompt;
} catch (const std::exception & e) {
2025-01-30 11:30:27 +01:00
// fprintf(stderr, "try_raw_render error: %s\n", e.what());
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
return "";
}
}
public:
2025-01-30 11:30:27 +01:00
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
chat_template(const std::string & source, const std::string & bos_token, const std::string & eos_token)
: source_(source), bos_token_(bos_token), eos_token_(eos_token)
{
template_root_ = minja::Parser::parse(source_, {
/* .trim_blocks = */ true,
/* .lstrip_blocks = */ true,
/* .keep_trailing_newline = */ false,
});
2025-01-30 11:30:27 +01:00
auto contains = [](const std::string & haystack, const std::string & needle) {
return haystack.find(needle) != std::string::npos;
};
const std::string user_needle = "<User Needle>";
const std::string sys_needle = "<System Needle>";
const json dummy_str_user_msg = {{"role", "user"}, {"content", user_needle}};
const json dummy_typed_user_msg = {{"role", "user"}, {"content", json::array({{{"type", "text"}, {"text", user_needle}}})}};
caps_.requires_typed_content =
!contains(try_raw_render(json::array({dummy_str_user_msg}), {}, false), user_needle)
&& contains(try_raw_render(json::array({dummy_typed_user_msg}), {}, false), user_needle);
const auto dummy_user_msg = caps_.requires_typed_content
? dummy_typed_user_msg
: dummy_str_user_msg;
const json needle_system_msg = {
{"role", "system"},
{"content", caps_.requires_typed_content ? json::array({{{"type", "text"}, {"text", sys_needle}}}) : json(sys_needle)},
};
caps_.supports_system_role = contains(try_raw_render({needle_system_msg, dummy_user_msg,}, {}, false), sys_needle);
auto out = try_raw_render(json::array({
dummy_user_msg
}), json::array({
{
{"name", "some_tool"},
{"type", "function"},
{"function", {
{"name", "some_tool"},
{"description", "Some tool."},
{"parameters", {
{"type", "object"},
{"properties", {
{"arg", {
{"type", "string"},
{"description", "Some argument."},
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}},
2025-01-30 11:30:27 +01:00
}},
{"required", json::array({ "arg" })},
}},
}},
},
}), false);
caps_.supports_tools = contains(out, "some_tool");
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
2025-01-30 11:30:27 +01:00
auto make_tool_calls_msg = [&](const json & tool_calls) {
return json {
{"role", "assistant"},
{"content", nullptr},
{"tool_calls", tool_calls},
};
};
auto make_tool_call = [](const std::string & tool_name, const json & arguments) {
return json {
{"id", "call_1___"},
{"type", "function"},
{"function", {
{"arguments", arguments},
{"name", tool_name},
}},
};
};
const json dummy_args_obj {{"argument_needle", "print('Hello, World!')"}};
// Note: the arguments are rendered in both cases, but may be double-escaped, which we don't want.
out = try_raw_render(json::array({
dummy_user_msg,
make_tool_calls_msg(json::array({make_tool_call("ipython", dummy_args_obj.dump())})),
}), {}, false);
auto tool_call_renders_str_arguments = contains(out, "\"argument_needle\":") || contains(out, "'argument_needle':");
out = try_raw_render(json::array({
dummy_user_msg,
make_tool_calls_msg(json::array({make_tool_call("ipython", dummy_args_obj)})),
}), {}, false);
auto tool_call_renders_obj_arguments = contains(out, "\"argument_needle\":") || contains(out, "'argument_needle':");
caps_.supports_tool_calls = tool_call_renders_str_arguments || tool_call_renders_obj_arguments;
caps_.requires_object_arguments = !tool_call_renders_str_arguments && tool_call_renders_obj_arguments;
auto out_empty = try_raw_render(json::array({dummy_user_msg, {{"role", "assistant"}, {"content", ""}}}), {}, false);
auto out_null = try_raw_render(json::array({dummy_user_msg, {{"role", "assistant"}, {"content", nullptr}}}), {}, false);
caps_.requires_non_null_content = contains(out_empty, user_needle) && !contains(out_null, user_needle);
if (caps_.supports_tool_calls) {
auto dummy_args = caps_.requires_object_arguments ? dummy_args_obj : json(dummy_args_obj.dump());
auto tc1 = make_tool_call("test_tool1", dummy_args);
auto tc2 = make_tool_call("test_tool2", dummy_args);
auto out = try_raw_render(json::array({
dummy_user_msg,
make_tool_calls_msg(json::array({tc1, tc2})),
}), {}, false);
caps_.supports_parallel_tool_calls = contains(out, "test_tool1") && contains(out, "test_tool2");
2025-01-30 11:30:27 +01:00
out = try_raw_render(json::array({
dummy_user_msg,
make_tool_calls_msg(json::array({tc1})),
{
{"role", "tool"},
{"name", "test_tool1"},
{"content", "Some response!"},
{"tool_call_id", "call_911_"},
}
}), {}, false);
caps_.supports_tool_responses = contains(out, "Some response!");
caps_.supports_tool_call_id = contains(out, "call_911_");
}
try {
if (!caps_.supports_tools) {
const json user_msg {
{"role", "user"},
{"content", "Hey"},
};
const json args {
{"arg1", "some_value"},
};
const json tool_call_msg {
{"role", "assistant"},
{"content", nullptr},
{"tool_calls", json::array({
{
// TODO: detect if requires numerical id or fixed length == 6 like Nemo
{"id", "call_1___"},
{"type", "function"},
{"function", {
{"name", "tool_name"},
{"arguments", (caps_.requires_object_arguments ? args : json(minja::Value(args).dump(-1, /* to_json= */ true)))},
}},
},
})},
};
std::string prefix, full;
{
chat_template_inputs inputs;
inputs.messages = json::array({user_msg});
inputs.add_generation_prompt = true;
prefix = apply(inputs);
}
{
chat_template_inputs inputs;
inputs.messages = json::array({user_msg, tool_call_msg});
inputs.add_generation_prompt = false;
full = apply(inputs);
}
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());
}
} catch (const std::exception & e) {
fprintf(stderr, "Failed to generate tool call example: %s\n", e.what());
}
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
const std::string & source() const { return source_; }
const std::string & bos_token() const { return bos_token_; }
const std::string & eos_token() const { return eos_token_; }
2025-01-30 11:30:27 +01:00
const chat_template_caps & original_caps() const { return caps_; }
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
// Deprecated, please use the form with chat_template_inputs and chat_template_options
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
std::string apply(
const nlohmann::ordered_json & messages,
const nlohmann::ordered_json & tools,
bool add_generation_prompt,
const nlohmann::ordered_json & extra_context = nlohmann::ordered_json(),
bool apply_polyfills = true)
{
fprintf(stderr, "[%s] Deprecated!\n", __func__);
chat_template_inputs inputs;
inputs.messages = messages;
inputs.tools = tools;
inputs.add_generation_prompt = add_generation_prompt;
inputs.extra_context = extra_context;
inputs.now = std::chrono::system_clock::now();
chat_template_options opts;
opts.apply_polyfills = apply_polyfills;
return apply(inputs, opts);
}
std::string apply(
const chat_template_inputs & inputs,
const chat_template_options & opts = chat_template_options()) const
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
{
json actual_messages;
auto has_tools = inputs.tools.is_array() && !inputs.tools.empty();
auto has_tool_calls = false;
auto has_tool_responses = false;
auto has_string_content = false;
for (const auto & message : inputs.messages) {
if (message.contains("tool_calls") && !message["tool_calls"].is_null()) {
has_tool_calls = true;
}
if (message.contains("role") && message["role"] == "tool") {
has_tool_responses = true;
}
if (message.contains("content") && 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
2025-01-30 11:30:27 +01:00
);
if (needs_polyfills) {
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
actual_messages = json::array();
auto add_message = [&](const json & msg) {
if (polyfill_typed_content && msg.contains("content") && !msg.at("content").is_null() && msg.at("content").is_string()) {
actual_messages.push_back({
{"role", msg.at("role")},
{"content", {{
{"type", "text"},
{"text", msg.at("content")},
}}},
});
} else {
actual_messages.push_back(msg);
}
};
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
std::string pending_system;
auto flush_sys = [&]() {
if (!pending_system.empty()) {
add_message({
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
{"role", "user"},
{"content", pending_system},
});
pending_system.clear();
}
};
2025-01-30 11:30:27 +01:00
json adjusted_messages;
if (polyfill_tools) {
adjusted_messages = add_system(inputs.messages,
"You can call any of the following tools to satisfy the user's requests: " + minja::Value(inputs.tools).dump(2, /* to_json= */ true) +
(!polyfill_tool_call_example || tool_call_example_.empty() ? "" : "\n\nExample tool call syntax:\n\n" + tool_call_example_));
} else {
adjusted_messages = inputs.messages;
}
for (const auto & message_ : adjusted_messages) {
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
auto message = message_;
if (!message.contains("role") || !message.contains("content")) {
throw std::runtime_error("message must have 'role' and 'content' fields: " + message.dump());
}
std::string role = message.at("role");
if (message.contains("tool_calls")) {
if (polyfill_object_arguments || polyfill_tool_calls) {
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
for (auto & tool_call : message.at("tool_calls")) {
if (tool_call["type"] == "function") {
auto & function = tool_call.at("function");
2025-01-30 11:30:27 +01:00
auto & arguments = function.at("arguments");
if (arguments.is_string()) {
try {
arguments = json::parse(arguments.get<std::string>());
} catch (const std::exception & ecvt) {
fprintf(stderr, "Failed to parse arguments: %s\n", ecvt.what());
}
}
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
}
}
if (polyfill_tool_calls) {
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
auto content = message.at("content");
auto tool_calls = json::array();
for (const auto & tool_call : message.at("tool_calls")) {
if (tool_call.at("type") != "function") {
continue;
}
const auto & function = tool_call.at("function");
auto tc = json {
{"name", function.at("name")},
{"arguments", function.at("arguments")},
};
if (tool_call.contains("id")) {
tc["id"] = tool_call["id"];
}
tool_calls.push_back(tc);
}
auto obj = json {
{"tool_calls", tool_calls},
};
if (!content.is_null() && content != "") {
obj["content"] = content;
}
message["content"] = obj.dump(2);
message.erase("tool_calls");
}
}
if (polyfill_tool_responses && role == "tool") {
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
message["role"] = "user";
auto obj = json {
{"tool_response", {
{"content", message.at("content")},
}},
};
if (message.contains("name")) {
obj["tool_response"]["name"] = message.at("name");
}
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
if (message.contains("tool_call_id")) {
obj["tool_response"]["tool_call_id"] = message.at("tool_call_id");
}
message["content"] = obj.dump(2);
message.erase("name");
}
if (!message["content"].is_null() && polyfill_system_role) {
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
std::string content = message.at("content");
if (role == "system") {
if (!pending_system.empty()) pending_system += "\n";
pending_system += content;
continue;
} else {
if (role == "user") {
if (!pending_system.empty()) {
message["content"] = pending_system + (content.empty() ? "" : "\n" + content);
pending_system.clear();
}
} else {
flush_sys();
}
}
}
add_message(message);
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
flush_sys();
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
} else {
actual_messages = inputs.messages;
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
auto context = minja::Context::make(json({
{"messages", actual_messages},
{"add_generation_prompt", inputs.add_generation_prompt},
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}));
context->set("bos_token", opts.use_bos_token ? bos_token_ : "");
context->set("eos_token", opts.use_eos_token ? eos_token_ : "");
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));
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
if (!inputs.extra_context.is_null()) {
for (auto & kv : inputs.extra_context.items()) {
context->set(kv.key(), minja::Value(kv.value()));
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
}
2025-01-30 11:30:27 +01:00
auto ret = template_root_->render(context);
// fprintf(stderr, "actual_messages: %s\n", actual_messages.dump(2).c_str());
// fprintf(stderr, "apply: %s\n\n", ret.c_str());
return ret;
}
static nlohmann::ordered_json add_system(const nlohmann::ordered_json & messages, const std::string & system_prompt) {
json messages_with_system = messages;
if (messages_with_system.size() > 0 && messages_with_system[0].at("role") == "system") {
std::string existing_system = messages_with_system.at(0).at("content");
messages_with_system[0] = json {
{"role", "system"},
{"content", existing_system + "\n\n" + system_prompt},
2025-01-30 11:30:27 +01:00
};
} else {
messages_with_system.insert(messages_with_system.begin(), json {
{"role", "system"},
{"content", system_prompt},
});
}
return messages_with_system;
Add Jinja template support (#11016) * Copy minja from https://github.com/google/minja/commit/58f0ca6dd74bcbfbd4e71229736640322b31c7f9 * Add --jinja and --chat-template-file flags * Add missing <optional> include * Avoid print in get_hf_chat_template.py * No designated initializers yet * Try and work around msvc++ non-macro max resolution quirk * Update test_chat_completion.py * Wire LLM_KV_TOKENIZER_CHAT_TEMPLATE_N in llama_model_chat_template * Refactor test-chat-template * Test templates w/ minja * Fix deprecation * Add --jinja to llama-run * Update common_chat_format_example to use minja template wrapper * Test chat_template in e2e test * Update utils.py * Update test_chat_completion.py * Update run.cpp * Update arg.cpp * Refactor common_chat_* functions to accept minja template + use_jinja option * Attempt to fix linkage of LLAMA_CHATML_TEMPLATE * Revert LLAMA_CHATML_TEMPLATE refactor * Normalize newlines in test-chat-templates for windows tests * Forward decl minja::chat_template to avoid eager json dep * Flush stdout in chat template before potential crash * Fix copy elision warning * Rm unused optional include * Add missing optional include to server.cpp * Disable jinja test that has a cryptic windows failure * minja: fix vigogne (https://github.com/google/minja/pull/22) * Apply suggestions from code review Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Finish suggested renamings * Move chat_templates inside server_context + remove mutex * Update --chat-template-file w/ recent change to --chat-template * Refactor chat template validation * Guard against missing eos/bos tokens (null token otherwise throws in llama_vocab::impl::token_get_attr) * Warn against missing eos / bos tokens when jinja template references them * rename: common_chat_template[s] * reinstate assert on chat_templates.template_default * Update minja to https://github.com/google/minja/commit/b8437df626ac6cd0ce3b333b3c74ed1129c19f25 * Update minja to https://github.com/google/minja/pull/25 * Update minja from https://github.com/google/minja/pull/27 * rm unused optional header --------- Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-01-21 14:18:51 +01:00
}
};
} // namespace minja