vllm接入

llm/GPT.py

@@ -0,0 +1,23 @@
+import openai
+
+
+class GPT():
+    def __init__(self, model_path = 'gpt-3.5-turbo', api_key = None, base_url = None):
+        openai.api_key = api_key
+        self.model_path = model_path
+        if base_url != None:
+            openai.base_url = base_url
+
+    def chat(self, message):
+        response = openai.ChatCompletion.create(
+            model=self.model_path,
+            messages=[
+                {"role": "user", "content": message}
+            ]
+        )
+        return response['choices'][0]['message']['content']
+
+if __name__ == '__main__':
+    llm = GPT('gpt-3.5-turbo', 'fk193752-RlcPi2mBQqPOU5u1F8SFkG2z0gtxD0HS','https://openai.api2d.net/v1')
+    # llm = GPT('gpt-3.5-turbo', 'fk193752-RlcPi2mBQqPOU5u1F8SFkG2z0gtxD0HS','https://openai.api2d.net/v1')
+    response = llm.chat("如何应对压力？")

llm/VllmGPT.py

@@ -0,0 +1,38 @@
+import json
+import requests
+# from core import content_db
+
+
+def question(cont):
+    chat_list = []
+    # contentdb = content_db.new_instance()
+    # list = contentdb.get_list('all','desc',11)
+    # answer_info = dict()
+    # chat_list = []
+    # i = len(list)-1
+    # while i >= 0:
+    #     answer_info = dict()
+    #     if list[i][0] == "member":
+    #         answer_info["role"] = "user"
+    #         answer_info["content"] = list[i][2]
+    #     elif list[i][0] == "fay":
+    #         answer_info["role"] = "bot"
+    #         answer_info["content"] = list[i][2]
+    #     chat_list.append(answer_info)
+    #     i -= 1
+    content = {
+        "model": "THUDM/chatglm3-6b",
+        "prompt":"请简单回复我。" +  cont,
+        "history":chat_list}
+    url = "http://192.168.1.3:8101/v1/completions"
+    req = json.dumps(content)
+    
+    headers = {'content-type': 'application/json'}
+    r = requests.post(url, headers=headers, data=req)
+    res = json.loads(r.text)
+    
+    return res['choices'][0]['text']
+
+if __name__ == "__main__":
+    req = question("你叫什么名字啊今年多大了")
+    print(req)