Update app.py

app.py

@@ -1,97 +1,48 @@
 import gradio as gr
-import aiohttp
+from huggingface_hub import InferenceClient
 import os
-import json
-from collections import deque
-import asyncio
+import requests
 
-TOKEN = os.getenv("HUGGINGFACE_API_TOKEN")
+# 추론 API 클라이언트 설정
+hf_client = InferenceClient("mistralai/Mistral-Nemo-Instruct-2407", token=os.getenv("HF_TOKEN"))
 
-if not TOKEN:
-    raise ValueError("API token is not set. Please set the HUGGINGFACE_API_TOKEN environment variable.")
-
-print(f"API Token: {TOKEN[:5]}...{TOKEN[-5:]}")  # Check API token
-
-memory = deque(maxlen=10)
-
-async def test_api():
-    headers = {"Authorization": f"Bearer {TOKEN}"}
-    async with aiohttp.ClientSession() as session:
-        async with session.get("https://api-inference.huggingface.co/models/mistralai/Mistral-Nemo-Instruct-2407", headers=headers) as response:
-            print(f"Test API response: {await response.text()}")
-
-async def respond(
+def respond(
     message,
     history: list[tuple[str, str]],
-    system_message="AI Assistant Role",
-    max_tokens=512,
-    temperature=0.7,
-    top_p=0.95,
+    system_message,
+    max_tokens,
+    temperature,
+    top_p,
 ):
-    system_prefix = "System: Respond in the same language as the input (English, Korean, Chinese, Japanese, etc.)."
-    full_system_message = f"{system_prefix}{system_message}"
 
-    memory.append((message, None))
-    messages = [{"role": "system", "content": full_system_message}]
-    for val in memory:
+    system_prefix = """
+반드시 한글로 답변할것. 
+    """
+    
+    messages = [{"role": "system", "content": f"{system_prefix} {system_message}"}]  # prefix 추가
+
+    for val in history:
         if val[0]:
             messages.append({"role": "user", "content": val[0]})
         if val[1]:
             messages.append({"role": "assistant", "content": val[1]})
 
-    headers = {
-        "Authorization": f"Bearer {TOKEN}",
-        "Content-Type": "application/json"
-    }
-    payload = {
-        "model": "mistralai/Mistral-Nemo-Instruct-2407",
-        "max_tokens": max_tokens,
-        "temperature": temperature,
-        "top_p": top_p,
-        "messages": messages,
-        "stream": True
-    }
-
-    try:
-        async with aiohttp.ClientSession() as session:
-            async with session.post("https://api-inference.huggingface.co/v1/chat/completions", headers=headers, json=payload) as response:
-                print(f"Response status: {response.status}")
-                if response.status != 200:
-                    error_text = await response.text()
-                    print(f"Error response: {error_text}")
-                    yield "An API response error occurred. Please try again."
-                    return
+    messages.append({"role": "user", "content": message})
 
-                response_text = ""
-                async for chunk in response.content:
-                    if chunk:
-                        try:
-                            chunk_data = chunk.decode('utf-8')
-                            response_json = json.loads(chunk_data)
-                            if "choices" in response_json:
-                                content = response_json["choices"][0]["message"]["content"]
-                                response_text += content
-                                yield response_text
-                        except json.JSONDecodeError:
-                            continue
-                
-                if not response_text:
-                    yield "I apologize, but I couldn't generate a response. Please try again."
-    except Exception as e:
-        print(f"Exception occurred: {str(e)}")
-        yield f"An error occurred: {str(e)}"
+    response = ""
 
-    memory[-1] = (message, response_text)
+    for message in hf_client.chat_completion(
+        messages,
+        max_tokens=max_tokens,
+        stream=True,
+        temperature=temperature,
+        top_p=top_p,
+    ):
+        token = message.choices[0].delta.content
+        if token is not None:
+            response += token.strip("")  # 토큰 제거
+        yield response
 
-async def chat(message, history, system_message, max_tokens, temperature, top_p):
-    response = ""
-    try:
-        async for chunk in respond(message, history, system_message, max_tokens, temperature, top_p):
-            response = chunk
-            yield response
-    except Exception as e:
-        print(f"Chat function error: {str(e)}")
-        yield f"An error occurred in the chat function: {str(e)}"
 
 theme = "Nymbo/Nymbo_Theme"
 
@@ -102,17 +53,25 @@ footer {
 """
 
 demo = gr.ChatInterface(
-    css=css,
-    fn=chat,
-    theme=theme,
+    respond,
     additional_inputs=[
-        gr.Textbox(value="AI Assistant Role", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Textbox(value="""
+너는 AI 어시스턴트이다.
+""", label="시스템 프롬프트"),
+        gr.Slider(minimum=1, maximum=2000, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
-    ]
+        gr.Slider(
+            minimum=0.1,
+            maximum=1.0,
+            value=0.95,
+            step=0.05,
+            label="Top-p (nucleus sampling)",
+        ),
+    ],
 )
 
+
+
+
 if __name__ == "__main__":
-    asyncio.run(test_api())  # Run API test
-    demo.queue().launch(max_threads=20, show_error=True)
+    demo.launch()