feat(turbo): add direct LLM chat mode with runtime toggle
Add 极速 mode that streams via /api/direct-chat without the lead agent stack, plus VITE_TURBO_MODE_VISIBLE in runtime-config.js (default off). Also disable global memory, tighten Qwen/vLLM thinking control for flash mode, and force thinking off on the direct-chat endpoint. Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
1 parent
4f7f07504a
commit
d765afe2ac
23 files changed
+1295
-394
No files matched your search
+15
-13
@@ -32,11 +32,11 @@ from langchain_core.messages import (
|
||||
)
|
||||
from langchain_core.messages.tool import tool_call_chunk
|
||||
from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult
|
||||
from langchain_openai import ChatOpenAI
|
||||
from langchain_openai.chat_models.base import _create_usage_metadata
|
||||
|
||||
from deerflow.models.message_ordering import normalize_chat_payload_messages
|
||||
|
||||
from langchain_openai import ChatOpenAI
|
||||
from langchain_openai.chat_models.base import _create_usage_metadata
|
||||
|
||||
from deerflow.models.message_ordering import normalize_chat_payload_messages
|
||||
|
||||
|
||||
def _normalize_vllm_chat_template_kwargs(payload: dict[str, Any]) -> None:
|
||||
"""Map DeerFlow's legacy ``thinking`` toggle to vLLM/Qwen's ``enable_thinking``.
|
||||
@@ -107,6 +107,8 @@ def _convert_delta_to_message_chunk_with_reasoning(_dict: Mapping[str, Any], def
|
||||
additional_kwargs["function_call"] = function_call
|
||||
|
||||
reasoning = _dict.get("reasoning")
|
||||
if reasoning is None:
|
||||
reasoning = _dict.get("reasoning_content")
|
||||
if reasoning is not None:
|
||||
additional_kwargs["reasoning"] = reasoning
|
||||
reasoning_text = _reasoning_to_text(reasoning)
|
||||
@@ -184,14 +186,14 @@ class VllmChatModel(ChatOpenAI):
|
||||
for payload_msg, orig_msg in zip(payload_messages, original_messages):
|
||||
if payload_msg.get("role") == "assistant" and isinstance(orig_msg, AIMessage):
|
||||
_restore_reasoning_field(payload_msg, orig_msg)
|
||||
else:
|
||||
ai_messages = [message for message in original_messages if isinstance(message, AIMessage)]
|
||||
assistant_payloads = [message for message in payload_messages if message.get("role") == "assistant"]
|
||||
for payload_msg, ai_msg in zip(assistant_payloads, ai_messages):
|
||||
_restore_reasoning_field(payload_msg, ai_msg)
|
||||
|
||||
normalize_chat_payload_messages(payload)
|
||||
return payload
|
||||
else:
|
||||
ai_messages = [message for message in original_messages if isinstance(message, AIMessage)]
|
||||
assistant_payloads = [message for message in payload_messages if message.get("role") == "assistant"]
|
||||
for payload_msg, ai_msg in zip(assistant_payloads, ai_messages):
|
||||
_restore_reasoning_field(payload_msg, ai_msg)
|
||||
|
||||
normalize_chat_payload_messages(payload)
|
||||
return payload
|
||||
|
||||
def _create_chat_result(self, response: dict | openai.BaseModel, generation_info: dict | None = None) -> ChatResult:
|
||||
"""Preserve vLLM reasoning on non-streaming responses."""
|
||||
|
||||
Reference in new issue
Block a user