mirror of
https://github.com/RYDE-WORK/Langchain-Chatchat.git
synced 2026-01-23 15:29:59 +08:00
253168a187
* 修复Azure 不设置Max token的bug * 重写agent 1. 修改Agent实现方式，支持多参数，仅剩 ChatGLM3-6b和 OpenAI GPT4 支持，剩余模型将在暂时缺席Agent功能 2. 删除agent_chat 集成到llm_chat中 3. 重写大部分工具，适应新Agent * 更新架构 * 删除web_chat,自动融合 * 移除所有聊天，都变成Agent控制 * 更新配置文件 * 更新配置模板和提示词 * 更改参数选择bug
70 lines
3.0 KiB
Python
70 lines
3.0 KiB
Python
from fastapi import Body
|
||
from fastapi.responses import StreamingResponse
|
||
from configs import LLM_MODEL_CONFIG
|
||
from server.utils import wrap_done, get_OpenAI
|
||
from langchain.chains import LLMChain
|
||
from langchain.callbacks import AsyncIteratorCallbackHandler
|
||
from typing import AsyncIterable, Optional
|
||
import asyncio
|
||
from langchain.prompts import PromptTemplate
|
||
|
||
from server.utils import get_prompt_template
|
||
|
||
|
||
async def completion(query: str = Body(..., description="用户输入", examples=["恼羞成怒"]),
|
||
stream: bool = Body(False, description="流式输出"),
|
||
echo: bool = Body(False, description="除了输出之外，还回显输入"),
|
||
model_name: str = Body(None, description="LLM 模型名称。"),
|
||
temperature: float = Body(0.01, description="LLM 采样温度", ge=0.0, le=1.0),
|
||
max_tokens: Optional[int] = Body(1024, 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)"),
|
||
):
|
||
|
||
#todo 因ApiModelWorker 默认是按chat处理的，会对params["prompt"] 解析为messages，因此ApiModelWorker 使用时需要有相应处理
|
||
async def completion_iterator(query: str,
|
||
model_name: str = None,
|
||
prompt_name: str = prompt_name,
|
||
echo: bool = echo,
|
||
) -> AsyncIterable[str]:
|
||
nonlocal max_tokens
|
||
callback = AsyncIteratorCallbackHandler()
|
||
if isinstance(max_tokens, int) and max_tokens <= 0:
|
||
max_tokens = None
|
||
|
||
model = get_OpenAI(
|
||
model_name=model_name,
|
||
temperature=temperature,
|
||
max_tokens=max_tokens,
|
||
callbacks=[callback],
|
||
echo=echo
|
||
)
|
||
|
||
prompt_template = get_prompt_template("completion", prompt_name)
|
||
prompt = PromptTemplate.from_template(prompt_template)
|
||
chain = LLMChain(prompt=prompt, llm=model)
|
||
|
||
# Begin a task that runs in the background.
|
||
task = asyncio.create_task(wrap_done(
|
||
chain.acall({"input": query}),
|
||
callback.done),
|
||
)
|
||
|
||
if stream:
|
||
async for token in callback.aiter():
|
||
# Use server-sent-events to stream the response
|
||
yield token
|
||
else:
|
||
answer = ""
|
||
async for token in callback.aiter():
|
||
answer += token
|
||
yield answer
|
||
|
||
await task
|
||
|
||
return EventSourceResponse(completion_iterator(query=query,
|
||
model_name=model_name,
|
||
prompt_name=prompt_name),
|
||
)
|