vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-09 19:46:11 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

chat : avoid including json in chat.h (#21306)

Browse source
This commit is contained in:
Georgi Gerganov 2026-04-03 09:07:59 +03:00 • committed by GitHub
parent 39b27f0da0
commit 57ace0d612
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 33 additions and 61 deletions
Download patch file Download diff file Expand all files Collapse all files

36
common/chat.cpp
View file

@ -13,6 +13,8 @@
#include "jinja/caps.h"
#include "peg-parser.h"
#include "nlohmann/json.hpp"
#include <cstdio>
#include <cstdlib>
#include <ctime>
@ -762,12 +764,12 @@ static void foreach_parameter(const json &
}
}
std::string common_chat_template_direct_apply(
static std::string common_chat_template_direct_apply_impl(
const common_chat_template & tmpl,
const autoparser::generation_params & inputs,
const std::optional<json> & messages_override,
const std::optional<json> & tools_override,
const std::optional<json> & additional_context) {
const std::optional<json> & messages_override = std::nullopt,
const std::optional<json> & tools_override = std::nullopt,
const std::optional<json> & additional_context = std::nullopt) {
jinja::context ctx(tmpl.source());
nlohmann::ordered_json inp = nlohmann::ordered_json{
@ -814,6 +816,12 @@ std::string common_chat_template_direct_apply(
return result;
}
std::string common_chat_template_direct_apply(
const common_chat_template & tmpl,
const autoparser::generation_params & inputs) {
return common_chat_template_direct_apply_impl(tmpl, inputs, std::nullopt, std::nullopt, std::nullopt);
}
static common_chat_params common_chat_params_init_ministral_3(const common_chat_template & tmpl,
const autoparser::generation_params & inputs) {
common_chat_params data;
@ -864,7 +872,7 @@ static common_chat_params common_chat_params_init_ministral_3(const common_chat_
data.supports_thinking = true;
data.thinking_start_tag = "[THINK]";
data.thinking_end_tag = "[/THINK]";
data.prompt = common_chat_template_direct_apply(tmpl, inputs, /* messages_override = */ adjusted_messages);
data.prompt = common_chat_template_direct_apply_impl(tmpl, inputs, /* messages_override = */ adjusted_messages);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.preserved_tokens = {
"[THINK]",
@ -947,7 +955,7 @@ static common_chat_params common_chat_params_init_gpt_oss(const common_chat_temp
adjusted_messages.push_back(msg);
}
auto prompt = common_chat_template_direct_apply(tmpl, inputs, /* messages_override= */ adjusted_messages);
auto prompt = common_chat_template_direct_apply_impl(tmpl, inputs, /* messages_override= */ adjusted_messages);
// Check if we need to replace the return token with end token during
// inference and without generation prompt. For more details see:
@ -1074,7 +1082,7 @@ static common_chat_params common_chat_params_init_functionary_v3_2(const common_
const autoparser::generation_params & inputs) {
common_chat_params data;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.prompt = common_chat_template_direct_apply_impl(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.preserved_tokens = {
">>>all",
@ -1168,7 +1176,7 @@ static common_chat_params common_chat_params_init_kimi_k2(const common_chat_temp
const autoparser::generation_params & inputs) {
common_chat_params data;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.prompt = common_chat_template_direct_apply_impl(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.supports_thinking = true;
data.preserved_tokens = {
@ -1291,7 +1299,7 @@ static common_chat_params common_chat_params_init_lfm2(const common_chat_templat
const autoparser::generation_params & inputs) {
common_chat_params data;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.prompt = common_chat_template_direct_apply_impl(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.supports_thinking = true;
data.preserved_tokens = {
@ -1370,7 +1378,7 @@ static common_chat_params common_chat_params_init_lfm2_5(const common_chat_templ
const autoparser::generation_params & inputs) {
common_chat_params data;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.prompt = common_chat_template_direct_apply_impl(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.supports_thinking = true;
data.preserved_tokens = {
@ -1441,7 +1449,7 @@ static common_chat_params common_chat_params_init_gigachat_v3(
common_chat_params data;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.prompt = common_chat_template_direct_apply_impl(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.supports_thinking = false;
data.preserved_tokens = {
@ -1724,9 +1732,9 @@ static common_chat_params common_chat_templates_apply_jinja(const struct common_
}
params.add_generation_prompt = false;
std::string no_gen_prompt = common_chat_template_direct_apply(tmpl, params);
std::string no_gen_prompt = common_chat_template_direct_apply_impl(tmpl, params);
params.add_generation_prompt = true;
std::string gen_prompt = common_chat_template_direct_apply(tmpl, params);
std::string gen_prompt = common_chat_template_direct_apply_impl(tmpl, params);
auto diff = calculate_diff_split(no_gen_prompt, gen_prompt);
params.generation_prompt = diff.right;
@ -1760,7 +1768,7 @@ static common_chat_params common_chat_templates_apply_jinja(const struct common_
common_chat_params data;
auto params_copy = params;
params_copy.reasoning_format = COMMON_REASONING_FORMAT_NONE;
data.prompt = common_chat_template_direct_apply(tmpl, params_copy);
data.prompt = common_chat_template_direct_apply_impl(tmpl, params_copy);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.generation_prompt = params.generation_prompt;
auto parser = build_chat_peg_parser([&params](common_chat_peg_builder &p) {