Update app.py

app.py

@@ -61,8 +61,8 @@ def load_pdf(file_path):
     return VectorStore
 
 
-def load_chatbot():
-    return load_qa_chain(llm=OpenAI(temperature=0.1, max_tokens=120), chain_type="stuff") 
+def load_chatbot(max_tokens=120):
+    return load_qa_chain(llm=OpenAI(temperature=0.5, max_tokens=max_tokens), chain_type="stuff") 
 
 
 def display_chat_history(chat_history):
@@ -97,10 +97,18 @@ def main():
             loading_message.text('Bot is thinking...')
             
             VectorStore = load_pdf(pdf)
-            chain = load_chatbot()
+            max_tokens = 100  # Initial max tokens
+            chain = load_chatbot(max_tokens=max_tokens)
             docs = VectorStore.similarity_search(query=query, k=3)
             with get_openai_callback() as cb:
                 response = chain.run(input_documents=docs, question=query)
+            
+            # Check if the response ends with a sentence-ending punctuation
+            while not response.strip().endswith(('.', '!', '?')) and max_tokens < MAX_TOKEN_LIMIT:
+                max_tokens += 50  # Increase the max_tokens limit
+                chain = load_chatbot(max_tokens=max_tokens)
+                additional_response = chain.run(input_documents=docs, question=query)
+                response += additional_response  # Append the additional response to the original response
 
             st.session_state['chat_history'].append(("Bot", response, "new"))
 
@@ -121,5 +129,8 @@ def main():
         # Mark all messages as old after displaying
         st.session_state['chat_history'] = [(sender, msg, "old") for sender, msg, _ in st.session_state['chat_history']]
 
+# Define a maximum token limit to avoid infinite loops
+MAX_TOKEN_LIMIT = 400
+
 if __name__ == "__main__":
     main()