mirror of
https://github.com/RYDE-WORK/Langchain-Chatchat.git
synced 2026-01-19 13:23:16 +08:00
b51ba11f45
新功能： - 将 LLM_MODEL 配置项改为 LLM_MODELS 列表，同时启动多个模型 - 将 wiki 纳入 samples 知识库 依赖变化： - 指定 streamlit~=1.27.0。1.26.0会报rerun错误，1.28.0会有无限刷新错误 修复优化： - 优化 get_default_llm_model 逻辑 - 适配 Qwen 在线 API 做 Embeddings 时最大 25 行的限制 - 列出知识库磁盘文件时跳过 . 开头的文件
84 lines
4.0 KiB
Python
84 lines
4.0 KiB
Python
from fastapi import Body
|
||
from fastapi.responses import StreamingResponse
|
||
from configs import LLM_MODELS, TEMPERATURE, SAVE_CHAT_HISTORY
|
||
from server.utils import wrap_done, get_ChatOpenAI
|
||
from langchain.chains import LLMChain
|
||
from langchain.callbacks import AsyncIteratorCallbackHandler
|
||
from typing import AsyncIterable
|
||
import asyncio
|
||
import json
|
||
from langchain.prompts.chat import ChatPromptTemplate
|
||
from typing import List, Optional
|
||
from server.chat.utils import History
|
||
from server.utils import get_prompt_template
|
||
from server.db.repository import add_chat_history_to_db, update_chat_history
|
||
|
||
|
||
async def chat(query: str = Body(..., description="用户输入", examples=["恼羞成怒"]),
|
||
history: List[History] = Body([],
|
||
description="历史对话",
|
||
examples=[[
|
||
{"role": "user", "content": "我们来玩成语接龙，我先来，生龙活虎"},
|
||
{"role": "assistant", "content": "虎头虎脑"}]]
|
||
),
|
||
stream: bool = Body(False, description="流式输出"),
|
||
model_name: str = Body(LLM_MODELS[0], description="LLM 模型名称。"),
|
||
temperature: float = Body(TEMPERATURE, description="LLM 采样温度", ge=0.0, le=1.0),
|
||
max_tokens: Optional[int] = Body(None, description="限制LLM生成Token数量，默认None代表模型最大值"),
|
||
# top_p: float = Body(TOP_P, description="LLM 核采样。勿与temperature同时设置", gt=0.0, lt=1.0),
|
||
prompt_name: str = Body("default", description="使用的prompt模板名称(在configs/prompt_config.py中配置)"),
|
||
):
|
||
history = [History.from_data(h) for h in history]
|
||
|
||
async def chat_iterator(query: str,
|
||
history: List[History] = [],
|
||
model_name: str = LLM_MODELS[0],
|
||
prompt_name: str = prompt_name,
|
||
) -> AsyncIterable[str]:
|
||
callback = AsyncIteratorCallbackHandler()
|
||
model = get_ChatOpenAI(
|
||
model_name=model_name,
|
||
temperature=temperature,
|
||
max_tokens=max_tokens,
|
||
callbacks=[callback],
|
||
)
|
||
|
||
prompt_template = get_prompt_template("llm_chat", prompt_name)
|
||
input_msg = History(role="user", content=prompt_template).to_msg_template(False)
|
||
chat_prompt = ChatPromptTemplate.from_messages(
|
||
[i.to_msg_template() for i in history] + [input_msg])
|
||
chain = LLMChain(prompt=chat_prompt, llm=model)
|
||
|
||
# Begin a task that runs in the background.
|
||
task = asyncio.create_task(wrap_done(
|
||
chain.acall({"input": query}),
|
||
callback.done),
|
||
)
|
||
|
||
answer = ""
|
||
chat_history_id = add_chat_history_to_db(chat_type="llm_chat", query=query)
|
||
|
||
if stream:
|
||
async for token in callback.aiter():
|
||
answer += token
|
||
# Use server-sent-events to stream the response
|
||
yield json.dumps(
|
||
{"text": token, "chat_history_id": chat_history_id},
|
||
ensure_ascii=False)
|
||
else:
|
||
async for token in callback.aiter():
|
||
answer += token
|
||
yield json.dumps(
|
||
{"text": answer, "chat_history_id": chat_history_id},
|
||
ensure_ascii=False)
|
||
|
||
if SAVE_CHAT_HISTORY and len(chat_history_id) > 0:
|
||
# 后续可以加入一些其他信息，比如真实的prompt等
|
||
update_chat_history(chat_history_id, response=answer)
|
||
await task
|
||
|
||
return StreamingResponse(chat_iterator(query=query,
|
||
history=history,
|
||
model_name=model_name,
|
||
prompt_name=prompt_name),
|
||
media_type="text/event-stream") |