Update app.py

app.py

@@ -1,11 +1,9 @@
-
 import discord
 import logging
 import os
+from huggingface_hub import InferenceClient
 import asyncio
-import aiohttp
 import subprocess
-from huggingface_hub import InferenceClient
 
 # 로깅 설정
 logging.basicConfig(level=logging.DEBUG, format='%(asctime)s:%(levelname)s:%(name)s: %(message)s', handlers=[logging.StreamHandler()])
@@ -18,7 +16,7 @@ intents.guilds = True
 intents.guild_messages = True
 
 # 추론 API 클라이언트 설정
-hf_client = InferenceClient(token=os.getenv("HF_TOKEN"))
+hf_client = InferenceClient("CohereForAI/c4ai-command-r-plus", token=os.getenv("HF_TOKEN"))
 
 # 특정 채널 ID
 SPECIFIC_CHANNEL_ID = int(os.getenv("DISCORD_CHANNEL_ID"))
@@ -33,7 +31,7 @@ class MyClient(discord.Client):
 
     async def on_ready(self):
         logging.info(f'{self.user}로 로그인되었습니다!')
-        self.loop.create_task(start_web_server())
+        subprocess.Popen(["python", "web.py"])
         logging.info("Web.py server has been started.")
 
     async def on_message(self, message):
@@ -75,32 +73,22 @@ async def generate_response(message):
     messages = [{"role": "system", "content": f"{system_prefix} {system_message}"}] + conversation_history
     logging.debug(f'Messages to be sent to the model: {messages}')
 
-    async with aiohttp.ClientSession() as session:
-        async with session.post(
-            f'https://api-inference.huggingface.co/models/CohereForAI/c4ai-command-r-plus',
-            headers={"Authorization": f"Bearer {os.getenv('HF_TOKEN')}"},
-            json={"inputs": {"text": user_input}}
-        ) as resp:
-            response_data = await resp.json()
-            response_text = response_data.get('generated_text', '')
-
-    logging.debug(f'Full model response: {response_text}')
-
-    conversation_history.append({"role": "assistant", "content": response_text})
-    return f"{user_mention}, {response_text}"
-
-async def start_web_server():
-    # Gradio 서버를 실행할 포트를 찾기 위해 사용할 수 있는 포트를 동적으로 설정
-    import socket
-    port = 7860
-    while True:
-        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
-            if s.connect_ex(('localhost', port)) != 0:
-                break
-            port += 1
-    python_executable = 'python3' if os.name == 'posix' else 'python'
-    subprocess.Popen([python_executable, "web.py"], env={"GRADIO_SERVER_PORT": str(port)})
+    loop = asyncio.get_event_loop()
+    response = await loop.run_in_executor(None, lambda: hf_client.chat_completion(
+        messages, max_tokens=1000, stream=True, temperature=0.7, top_p=0.85))
+
+    full_response = []
+    for part in response:
+        logging.debug(f'Part received from stream: {part}')
+        if part.choices and part.choices[0].delta and part.choices[0].delta.content:
+            full_response.append(part.choices[0].delta.content)
+
+    full_response_text = ''.join(full_response)
+    logging.debug(f'Full model response: {full_response_text}')
+
+    conversation_history.append({"role": "assistant", "content": full_response_text})
+    return f"{user_mention}, {full_response_text}"
 
 if __name__ == "__main__":
     discord_client = MyClient(intents=intents)
-    discord_client.run(os.getenv('DISCORD_TOKEN'))
+    discord_client.run(os.getenv('DISCORD_TOKEN'))