002f9879b2
Replaces cherry-picked upstream_ref with complete source trees. xllm/ — Iluvatar official C++ inference engine (15MB, 1470 files) Complete: kernels → layers → models → runtime → scheduler → api Excluded: .git, binary images, third_party submodule checkouts ds_vllm/ — Iluvatar official vllm fork (8MB, 703 files) Included: csrc/ (ALL CUDA kernels), fused_moe/, qwen3_5 model, _custom_ops Excluded: tests, benchmarks, docs, examples (not needed for reference) Critical call chains now fully traceable: MoE: moe_topk_softmax_kernels.cuh → ixformer.h → fused_moe.cpp → layer GDN: qwen3_gated_delta_net_base.cpp → qwen3_5_gated_delta_net.cpp Attention: ixformer.h → xllm_paged_attention → attention.cpp
65 lines
2.2 KiB
C++
65 lines
2.2 KiB
C++
/* Copyright 2025 The xLLM Authors. All Rights Reserved.
|
|
Copyright 2024 The ScaleLLM Authors. All Rights Reserved.
|
|
|
|
Licensed under the Apache License, Version 2.0 (the "License");
|
|
you may not use this file except in compliance with the License.
|
|
You may obtain a copy of the License at
|
|
|
|
https://github.com/jd-opensource/xllm/blob/main/LICENSE
|
|
|
|
Unless required by applicable law or agreed to in writing, software
|
|
distributed under the License is distributed on an "AS IS" BASIS,
|
|
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
See the License for the specific language governing permissions and
|
|
limitations under the License.
|
|
==============================================================================*/
|
|
|
|
#pragma once
|
|
|
|
#include <optional>
|
|
#include <string>
|
|
|
|
namespace xllm {
|
|
|
|
struct ReasoningResult {
|
|
std::optional<std::string> normal_text = std::nullopt;
|
|
std::optional<std::string> reasoning_text = std::nullopt;
|
|
|
|
ReasoningResult() = default;
|
|
|
|
ReasoningResult(std::optional<std::string> normal,
|
|
std::optional<std::string> reasoning)
|
|
: normal_text(normal), reasoning_text(reasoning) {}
|
|
};
|
|
|
|
class ReasoningDetector {
|
|
public:
|
|
ReasoningDetector(const std::string& think_start_token,
|
|
const std::string& think_end_token,
|
|
bool force_reasoning = false,
|
|
bool stream_reasoning = true);
|
|
|
|
~ReasoningDetector() = default;
|
|
|
|
// Detects and parses reasoning sections in the provided text. Returns both
|
|
// reasoning content and normal text separately.
|
|
ReasoningResult detect_and_parse(std::string& text);
|
|
|
|
// Streaming incremental parsing for reasoning content.
|
|
// Handles partial reasoning tags and content.
|
|
//
|
|
// If stream_reasoning is False:
|
|
// Accumulates reasoning content until the end tag is found
|
|
// If stream_reasoning is True:
|
|
// Streams reasoning content as it arrives
|
|
ReasoningResult parse_streaming_increment(std::string& new_text);
|
|
|
|
protected:
|
|
std::string think_start_token_;
|
|
std::string think_end_token_;
|
|
bool in_reasoning_;
|
|
bool stream_reasoning_;
|
|
std::string buffer_ = "";
|
|
bool stripped_think_start_ = false;
|
|
};
|
|
} // namespace xllm
|