Merge commit '3ca19b0e9f3f4f444d22c9f509805d037a611847' into concedo_experimental

# Conflicts:
#	.github/workflows/build.yml
#	common/CMakeLists.txt
#	common/chat-peg-parser.cpp
#	docs/backend/SYCL.md
#	docs/ops.md
#	docs/ops/SYCL.csv
#	ggml/src/ggml-sycl/common.hpp
#	ggml/src/ggml-sycl/convert.hpp
#	ggml/src/ggml-sycl/element_wise.cpp
#	ggml/src/ggml-sycl/ggml-sycl.cpp
#	ggml/src/ggml-sycl/norm.cpp
#	ggml/src/ggml-sycl/rope.cpp
#	ggml/src/ggml-sycl/rope.hpp
#	ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp
#	ggml/src/ggml-webgpu/ggml-webgpu.cpp
#	ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_decls.tmpl
#	ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_reg_tile.wgsl
#	ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_vec.wgsl
#	scripts/compare-llama-bench.py
#	scripts/sync_vendor.py
#	tests/CMakeLists.txt
#	tools/cli/cli.cpp
This commit is contained in:
Concedo
2026-03-15 11:11:31 +08:00
41 changed files with 836 additions and 212 deletions
+21 -6
View File
@@ -8,6 +8,7 @@
#include "log.h"
#include "json-partial.cpp"
#include "regex-partial.cpp"
#include "reasoning-budget.h"
#include "chat-auto-parser-generator.cpp"
#include "chat-auto-parser-helpers.cpp"
#include "chat-diff-analyzer.cpp"
@@ -871,7 +872,9 @@ static common_chat_params common_chat_params_init_ministral_3(const common_chat_
auto extract_reasoning = inputs.reasoning_format != COMMON_REASONING_FORMAT_NONE;
auto include_grammar = true;
data.supports_thinking = true;
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.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.preserved_tokens = {
@@ -1179,9 +1182,11 @@ static common_chat_params common_chat_params_init_kimi_k2(const common_chat_temp
const autoparser::templates_params & inputs) {
common_chat_params data;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.supports_thinking = true;
data.prompt = common_chat_template_direct_apply(tmpl, inputs);
data.format = COMMON_CHAT_FORMAT_PEG_NATIVE;
data.supports_thinking = true;
data.thinking_start_tag = "<think>";
data.thinking_end_tag = "</think>";
data.preserved_tokens = {
"<|tool_calls_section_begin|>",
"<|tool_calls_section_end|>",
@@ -1541,6 +1546,16 @@ static common_chat_params common_chat_templates_apply_jinja(const struct common_
autoparser.analyze_template(tmpl);
auto auto_params = autoparser::peg_generator::generate_parser(tmpl, params, autoparser);
auto_params.supports_thinking = autoparser.reasoning.mode != autoparser::reasoning_mode::NONE;
if (auto_params.supports_thinking) {
auto_params.thinking_start_tag = autoparser.reasoning.start;
auto_params.thinking_end_tag = autoparser.reasoning.end;
// FORCED_OPEN and FORCED_CLOSED both put <think> in the generation prompt
// (FORCED_CLOSED forces empty <think></think> when thinking is disabled,
// but forces <think> open when thinking is enabled)
auto_params.thinking_forced_open =
autoparser.reasoning.mode == autoparser::reasoning_mode::FORCED_OPEN ||
autoparser.reasoning.mode == autoparser::reasoning_mode::FORCED_CLOSED;
}
return auto_params;
} catch (const std::exception & e) {
throw std::invalid_argument(std::string("Unable to generate parser for this template. Automatic parser generation failed: ") + e.what());
@@ -1634,8 +1649,8 @@ common_chat_msg common_chat_peg_parse(const common_peg_arena & src_pars
build_chat_peg_parser([](common_chat_peg_builder & p) { return p.content(p.rest()) + p.end(); }) :
src_parser;
if (src_parser.empty()) {
LOG_WRN("No parser definition detected, assuming pure content parser.");
if (src_parser.empty()) {
LOG_DBG("No parser definition detected, assuming pure content parser.");
}
LOG_DBG("Parsing PEG input with format %s: %s\n", common_chat_format_name(params.format), input.c_str());