Update app.py

app.py

@@ -61,47 +61,67 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
 
     start_time = time.time()
 
+    # Debugging - Ensure that FAISS index and embeddings are correctly loaded
     if os.path.exists('faiss.index') and os.path.exists('embeddings.pkl'):
         log_output.write("Loading FAISS index from disk...\n")
         faiss_index, question_embeddings = load_faiss_index('faiss.index', 'embeddings.pkl')
+        print(f"FAISS index and embeddings loaded successfully. Number of embeddings: {len(question_embeddings)}")
     else:
         log_output.write("Creating and saving FAISS index...\n")
         
         embedding_model = OpenAIEmbeddings(openai_api_key=openai.api_key)
         faiss_index, question_embeddings = create_and_save_faiss_index(questions, embedding_model, 'faiss.index', 'embeddings.pkl')
+        print(f"Created new FAISS index. Number of questions: {len(questions)}")
 
+    # Debugging - Ensure questions and answers lists are valid
+    print(f"questions list length: {len(questions)}")  # Debugging print
+    print(f"answers list length: {len(answers)}")  # Debugging print
+
+    # Prepare message history
     messages = [{"role": "system", "content": system_message}]
     for user_message, bot_response in history:
         messages.append({"role": "user", "content": user_message})
         if bot_response:
             messages.append({"role": "assistant", "content": bot_response})
 
+    # Add the current user message
     user_message = message
     messages.append({"role": "user", "content": user_message})
 
-    response_text = retrieve_answer(user_message, faiss_index, OpenAIEmbeddings(openai_api_key=openai.api_key), answers=["..."], threshold=0.35)
+    # Retrieve answer from FAISS or fallback to GPT-4
+    response_text = retrieve_answer(user_message, faiss_index, OpenAIEmbeddings(openai_api_key=openai.api_key), answers, threshold=0.35)
 
-    if response_text == "No good match found in dataset. Using GPT-4o-mini to generate an answer.":
-        log_output.write("No good match found in dataset. Using GPT-4o-mini to generate an answer.\n")
+    if response_text == "No good match found in dataset. Using GPT-4 to generate an answer.":
+        log_output.write("No good match found in dataset. Using GPT-4 to generate an answer.\n")
         response_text = ask_openai_gpt4(user_message)
 
     # Stop the timer and calculate response time
     end_time = time.time()
     response_time = end_time - start_time  # Time in seconds
 
-    # Yield the response with the logs and response time
-    yield response_text, f"Response time: {response_time:.4f} seconds", log_output.getvalue()
+    # Return the response, response time, and logs
+    return response_text, f"Response time: {response_time:.4f} seconds", log_output.getvalue()
 
 
 # Gradio ChatInterface with additional inputs for model settings and response time
 demo = gr.ChatInterface(
     fn=respond,
+    additional_inputs=[
+        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Slider(minimum=1, maximum=2048, 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)")
+    ],
     title="Medical Chatbot with Customizable Parameters and Response Time",
-    description="A chatbot with customizable parameters using FAISS for quick responses or fallback to GPT-4 when no relevant answer is found. Response time is also tracked."
+    description="A chatbot with customizable parameters using FAISS for quick responses or fallback to GPT-4 when no relevant answer is found. Response time is also tracked.",
+    type='messages'  # Set type to 'messages' instead of 'tuples'
 )
 
 if __name__ == "__main__":
+    # Load dataset
     df = pd.read_csv("medquad.csv")
     questions = df['question'].tolist()
     answers = df['answer'].tolist()
-    demo.launch(cache_examples=False)
+
+    print(f"Loaded questions and answers. Number of questions: {len(questions)}, Number of answers: {len(answers)}")
+    demo.launch()