添加ChatGLM2教程 (#125)

* Create GLM2对接教程.md

* 添加GLM2接入教程

* Delete GLM2对接教程.md

* Delete image.png

* Delete openai_api.py

* Delete openai_api_int4.py

* Delete openai_api_int8.py

* Create GLM2对接教程.md

* 添加ChatGLM2接口

* Delete openai_api_int4.py

* Delete openai_api_int8.py

* Update openai_api.py

* Update GLM2对接教程.md

docs/zh/examples/ChatGLM2/GLM2对接教程.md

@@ -0,0 +1,36 @@
+# 3分钟在Fastgpt上用上GLM
+## 前言
+Fast GPT 允许你使用自己的 openai API KEY 来快速的调用 openai 接口，目前集成了 Gpt35, Gpt4 和 embedding. 可构建自己的知识库。但考虑到数据安全的问题，我们并不能将所有的数据都交付给云端大模型。那如何在fastgpt上接入私有化模型呢，本文就以清华的ChatGLM2为例，为各位讲解如何在fastgpt中接入私有化模型。
+## ChatGLM2简介
+ChatGLM2-6B 是开源中英双语对话模型 ChatGLM-6B 的第二代版本，具体介绍请看项目：https://github.com/THUDM/ChatGLM2-6B
+注意，ChatGLM2-6B 权重对学术研究完全开放，在获得官方的书面许可后，亦允许商业使用。本教程只是介绍了一种用法，并不会给予任何授权。
+## 推荐配置
+依据官方数据，同样是生成 8192 长度，量化等级为FP16要占用12.8GB  显存、INT8为8.1GB显存、INT4为5.1GB显存，量化后会稍微影响性能，但不多。
+因此推荐配置如下：
+fp16:内存>=16GB,显存>=16GB,硬盘空间>=25GB,启动时使用命令python openai_api.py 16
+int8:内存>=16GB,显存>=9GB,硬盘空间>=25GB,启动时选择python openai_api.py 8
+int4:内存>=16GB,显存>=6GB,硬盘空间>=25GB,启动时选择python openai_api.py 4
+## 环境配置
+Python 3.8.10
+CUDA 11.8
+科学上网环境
+## 简单的步骤
+1. 根据上面的环境配置配置好环境，具体教程自行GPT；
+1. 在命令行输入pip install -r requirments.txt
+2. 打开你需要启动的py文件，在代码的第76行配置token，这里的token只是加一层验证，防止接口被人盗用
+2. python openai_api.py 16//这里的数字根据上面的配置进行选择
+
+然后等待模型下载，直到模型加载完毕，出现报错先问GPT  
+上面两个文件在本文档的同目录
+
+启动成功后应该会显示如下地址：
+![Alt text](image.png)
+这里的http://0.0.0.0:6006就是连接地址
+
+然后现在回到.env.local文件，依照以下方式配置地址：
+
+OPENAI_BASE_URL=http://127.0.0.1:6006/v1
+OPENAIKEY=sk-aaabbbcccdddeeefffggghhhiiijjjkkk //这里是你在代码中配置的token
+这里的OPENAIKEY可以任意填写
+
+这样就成功接入ChatGLM2了

docs/zh/examples/ChatGLM2/openai_api.py

@@ -0,0 +1,174 @@
+# coding=utf-8
+import time
+import torch
+import uvicorn
+from pydantic import BaseModel, Field
+from fastapi import FastAPI, HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from contextlib import asynccontextmanager
+from typing import Any, Dict, List, Literal, Optional, Union
+from transformers import AutoTokenizer, AutoModel
+from sse_starlette.sse import ServerSentEvent, EventSourceResponse
+from fastapi import Depends, HTTPException, Request
+from starlette.status import HTTP_401_UNAUTHORIZED
+import argparse
+
+
+@asynccontextmanager
+async def lifespan(app: FastAPI): # collects GPU memory
+    yield
+    if torch.cuda.is_available():
+        torch.cuda.empty_cache()
+        torch.cuda.ipc_collect()
+
+
+app = FastAPI(lifespan=lifespan)
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+
+class ChatMessage(BaseModel):
+    role: Literal["user", "assistant", "system"]
+    content: str
+
+
+class DeltaMessage(BaseModel):
+    role: Optional[Literal["user", "assistant", "system"]] = None
+    content: Optional[str] = None
+
+
+class ChatCompletionRequest(BaseModel):
+    model: str
+    messages: List[ChatMessage]
+    temperature: Optional[float] = None
+    top_p: Optional[float] = None
+    max_length: Optional[int] = None
+    stream: Optional[bool] = False
+
+
+class ChatCompletionResponseChoice(BaseModel):
+    index: int
+    message: ChatMessage
+    finish_reason: Literal["stop", "length"]
+
+
+class ChatCompletionResponseStreamChoice(BaseModel):
+    index: int
+    delta: DeltaMessage
+    finish_reason: Optional[Literal["stop", "length"]]
+
+
+class ChatCompletionResponse(BaseModel):
+    model: str
+    object: Literal["chat.completion", "chat.completion.chunk"]
+    choices: List[Union[ChatCompletionResponseChoice, ChatCompletionResponseStreamChoice]]
+    created: Optional[int] = Field(default_factory=lambda: int(time.time()))
+
+async def verify_token(request: Request):
+    auth_header = request.headers.get('Authorization')
+    if auth_header:
+        token_type, _, token = auth_header.partition(' ')
+        if token_type.lower() == "bearer" and token == "sk-aaabbbcccdddeeefffggghhhiiijjjkkk": # 这里配置你的token
+            return True
+    raise HTTPException(
+        status_code=HTTP_401_UNAUTHORIZED,
+        detail="Invalid authorization credentials",
+    )
+
+
+@app.post("/v1/chat/completions", response_model=ChatCompletionResponse)
+async def create_chat_completion(request: ChatCompletionRequest, token: bool = Depends(verify_token)):
+    global model, tokenizer
+
+    if request.messages[-1].role != "user":
+        raise HTTPException(status_code=400, detail="Invalid request")
+    query = request.messages[-1].content
+
+    prev_messages = request.messages[:-1]
+    if len(prev_messages) > 0 and prev_messages[0].role == "system":
+        query = prev_messages.pop(0).content + query
+
+    history = []
+    if len(prev_messages) % 2 == 0:
+        for i in range(0, len(prev_messages), 2):
+            if prev_messages[i].role == "user" and prev_messages[i+1].role == "assistant":
+                history.append([prev_messages[i].content, prev_messages[i+1].content])
+
+    if request.stream:
+        generate = predict(query, history, request.model)
+        return EventSourceResponse(generate, media_type="text/event-stream")
+
+    response, _ = model.chat(tokenizer, query, history=history)
+    choice_data = ChatCompletionResponseChoice(
+        index=0,
+        message=ChatMessage(role="assistant", content=response),
+        finish_reason="stop"
+    )
+
+    return ChatCompletionResponse(model=request.model, choices=[choice_data], object="chat.completion")
+
+
+async def predict(query: str, history: List[List[str]], model_id: str):
+    global model, tokenizer
+
+    choice_data = ChatCompletionResponseStreamChoice(
+        index=0,
+        delta=DeltaMessage(role="assistant"),
+        finish_reason=None
+    )
+    chunk = ChatCompletionResponse(model=model_id, choices=[choice_data], object="chat.completion.chunk")
+    yield "{}".format(chunk.json(exclude_unset=True, ensure_ascii=False))
+
+    current_length = 0
+
+    for new_response, _ in model.stream_chat(tokenizer, query, history):
+        if len(new_response) == current_length:
+            continue
+
+        new_text = new_response[current_length:]
+        current_length = len(new_response)
+
+        choice_data = ChatCompletionResponseStreamChoice(
+            index=0,
+            delta=DeltaMessage(content=new_text),
+            finish_reason=None
+        )
+        chunk = ChatCompletionResponse(model=model_id, choices=[choice_data], object="chat.completion.chunk")
+        yield "{}".format(chunk.json(exclude_unset=True, ensure_ascii=False))
+
+
+    choice_data = ChatCompletionResponseStreamChoice(
+        index=0,
+        delta=DeltaMessage(),
+        finish_reason="stop"
+    )
+    chunk = ChatCompletionResponse(model=model_id, choices=[choice_data], object="chat.completion.chunk")
+    yield "{}".format(chunk.json(exclude_unset=True, ensure_ascii=False))
+    yield '[DONE]'
+
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--model_name", default="16", type=str, help="Model name")
+    args = parser.parse_args()
+
+    model_dict = {
+        "4": "THUDM/chatglm2-6b-int4",
+        "8": "THUDM/chatglm2-6b-int8",
+        "16": "THUDM/chatglm2-6b"
+    }
+
+    model_name = model_dict.get(args.model_name, "THUDM/chatglm2-6b")
+
+    tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+    model = AutoModel.from_pretrained(model_name, trust_remote_code=True).cuda()
+    model.eval()
+
+    uvicorn.run(app, host='0.0.0.0', port=6006, workers=1)