Update app.py

app.py

@@ -1,6 +1,6 @@
 import streamlit as st
 import json
-from huggingface_hub import InferenceClient
+import openai
 import time
 
 # Initialize Streamlit page configuration
@@ -11,18 +11,15 @@ st.set_page_config(
 )
 
 # Secret management
-def get_hf_api_token():
+def get_openai_api_token():
     # In production, use Streamlit secrets
-    return st.secrets["HUGGINGFACE_API_TOKEN"]
+    return st.secrets["OPENAI_API_KEY"]
 
-# Initialize HF client
+# Initialize OpenAI client
 @st.cache_resource
-def get_hf_client():
-    client = InferenceClient(
-        model="meta-llama/Llama-2-7b-chat-hf",
-        token=get_hf_api_token()
-    )
-    return client
+def get_openai_client():
+    openai.api_key = get_openai_api_token()
+    return openai
 
 # Load and cache knowledge base
 @st.cache_data
@@ -74,9 +71,9 @@ def get_context(query: str, knowledge_base: dict) -> str:
     
     return "\n".join(contexts)
 
-def generate_llama_prompt(query: str, context: str) -> str:
-    """Generate prompt for Llama model"""
-    return f"""[INST] You are Manyue's AI assistant. Use this context to answer questions about Manyue:
+def generate_openai_prompt(query: str, context: str) -> str:
+    """Generate prompt for OpenAI model"""
+    return f"""You are Manyue's AI assistant. Use this context to answer questions about Manyue:
 
 Context:
 {context}
@@ -89,36 +86,35 @@ Instructions:
 - Be specific about technical details and achievements
 - Keep responses concise but informative
 - Focus on relevant experience and skills
-- Maintain a professional tone
-
-Your response: [/INST]"""
+- Maintain a professional tone"""
 
 def get_chat_response(query: str, knowledge_base: dict) -> str:
-    """Get response from Llama model via HF API"""
+    """Get response from OpenAI API"""
     try:
         # Get context
         context = get_context(query, knowledge_base)
         
         # Generate prompt
-        prompt = generate_llama_prompt(query, context)
+        prompt = generate_openai_prompt(query, context)
         
-        # Get client
-        client = get_hf_client()
+        # Get OpenAI client
+        client = get_openai_client()
         
         # Generate response
-        response = client.text_generation(
-            prompt,
-            max_new_tokens=200,
-            temperature=0.7,
-            top_p=0.95,
-            repetition_penalty=1.1
+        response = client.ChatCompletion.create(
+            model="gpt-3.5-turbo",  # You can change to gpt-4 if preferred
+            messages=[
+                {"role": "system", "content": prompt},
+                {"role": "user", "content": query}
+            ],
+            max_tokens=200,
+            temperature=0.7
         )
         
-        # Clean response
-        response = response.strip()
-        response = response.split("[/INST]")[-1].strip()
+        # Extract and clean response
+        response_text = response.choices[0].message.content.strip()
         
-        return response
+        return response_text
         
     except Exception as e:
         st.error(f"Error generating response: {str(e)}")