gpt4o debug

app.py

@@ -116,6 +116,7 @@ if st.button("Change Selection", key="change_selection_btn"):
     st.session_state.pop("selected_models", None)
     st.session_state.pop("retriever", None)
     st.session_state.pop("reasoner", None)
+    st.session_state.messages = [{"role": "assistant", "content": "Let's start verifying the claims here! 👇"}]
     st.rerun()
 
 # Initialize chat history
@@ -226,9 +227,9 @@ if prompt := st.chat_input("Type here"):
         message_placeholder = st.empty()
         full_response = ""
         if decision.lower() == 'support':
-            assistant_response = f'The claim is correct because {reasoning}'
+            assistant_response = f'The claim is CORRECT because {reasoning}'
         elif decision.lower() == 'contradict':
-            assistant_response = f'The claim is incorrect because {reasoning}'
+            assistant_response = f'The claim is INCORRECT because {reasoning}'
 
         # Simulate stream of response with milliseconds delay
         for chunk in assistant_response.split():

llm_reasoner.py

@@ -34,7 +34,7 @@ class LLMReasoner():
                 model=self.model_name,
                 messages=prompt,
                 temperature=self.temp,
-                #max_completion_tokens=self.max_tokens,
+                max_completion_tokens=self.max_tokens,
                 top_p=self.top_p,
                 frequency_penalty=self.frequency_penalty,
                 presence_penalty=self.presence_penalty