Added examples and replaced raising error with explicit information.

examples.py

@@ -0,0 +1,6 @@
+FedMinutesSearch = [['Summarize the monetary policy outlook discussed by the participants during the meeting of June 2023 in three paragraphs.'],
+                    ['Who were the voting members present in the meeting of March 2023?'],
+                    ['What was the economic outlook from the staff presented in the meeting '
+                     'of April 2009 with respect to labour market developments and industrial production?'],
+                    ['How important was the pandemic of Covid-19 in the discussions during 2020 for the effects of monetary policy?'],
+                    ['What was the impact of the oil crisis for the economic outlook during 1973?']]

public_app.py

@@ -11,6 +11,7 @@ from langchain.chat_models import ChatOpenAI
 
 from prompts import PROMPT_EXTRACT_DATE, PROMPT_FED_ANALYST
 from filterminutes import search_with_filter
+from examples import FedMinutesSearch
 
 # --------------------------Load the sentence transformer and the vector store--------------------------#
 model_name = 'sentence-transformers/all-mpnet-base-v2'
@@ -54,7 +55,8 @@ def get_chain(query, api_key=os.environ['OPENAI_API_KEY']):
     date_response = date_extractor.run(query)
     if date_response == 'False':
         logging.info('No date elements found. Running the qa without filtering can output incorrect results.')
-        raise gr.Error('No date elements found. Please include temporal references in in your query.')
+        return 'No date elements found. Please include temporal references in your query. ' \
+               'If you believe this is wrong, please flag below as appropriate.'
     else:
         filter_date = json.loads(date_response)
         logging.info(f'Date parameters retrieved: {filter_date}')
@@ -78,8 +80,8 @@ if __name__ == '__main__':
                                    'tuned to focus on economic, '
                                    'cultural, financial, and political developments occurring at a given time.'
                                    ' The model actively looks for the presence of date elements in the query '
-                                   'and will raise an error if it cannot find them to minimize the risk of model '
-                                   'hallucination. Nevertheless the usual caveats for application making use of generative AI apply.'
+                                   'and will stop the execution if cannot find them to minimize the risk of model '
+                                   'hallucination. Nevertheless, the usual caveats for application making use of generative AI apply.'
                                    ' Click the query examples below to see some possible outputs from the model.',
                        article='**Disclaimer**: This app is for demonstration purposes only, and no assurance of uninterrupted'
                        ' functionality can be given at this time. Answers may take some'
@@ -90,12 +92,7 @@ if __name__ == '__main__':
                        flagging_options=["error", "ambiguous", "not useful"],
                        outputs=gr.Textbox(lines=1, label='Answer'),
                        title='Search the FED minutes',
-                       examples=[['Summarize the monetary policy outlook discussed by the participants during the meeting of June 2023 in three paragraphs.'],
-                           ['Who were the voting members present in the meeting of March 2023?'],
-                                 ['What was the economic outlook from the staff presented in the meeting '
-                                  'of April 2009 with respect to labour market developments and industrial production?'],
-                                 ['How important was the pandemic of Covid-19 in the discussions during 2020 for the effects of monetary policy?'],
-                                 ['What was the impact of the oil crisis for the economic outlook during 1973?']],
+                       examples=FedMinutesSearch,
                        cache_examples=True
                        )
     app.queue(concurrency_count=2)