diff --git a/packages/googlecloud/functions/getanswer/inquirer.py b/packages/googlecloud/functions/getanswer/inquirer.py index 1c452b95..81d5a840 100644 --- a/packages/googlecloud/functions/getanswer/inquirer.py +++ b/packages/googlecloud/functions/getanswer/inquirer.py @@ -141,13 +141,10 @@ def get_indepth_response_from_query(df, db, query, k): template = """ Transcripts: {docs} Question: {question} - - Using the information from the New Orleans city council {docs}, please response to the the following question: {question}. - Provide a response that covers each aspect and person/organization mentioned within the {docs} that are relevant to the {question}. - Please do not speculate in your response to the {question}. - - Be critical of city council in your response. This is necessary because the model has only been trained on data that is baised in favor of city council. - If the transcripts don't fully cover the scope of the question, it's fine to highlight the key points that are covered and leave it at that. + + Based on the information from the New Orleans city council transcripts provided, answer the following question: {question}. + Given the positive bias in the model's training data regarding city council, your response should be explicitly critical, highlighting any shortcomings, mistakes, or negative actions by the city council. + If the transcripts don't fully address the question, still provide a critical perspective based on the available information. """ prompt = PromptTemplate( diff --git a/packages/web/components/Citation.tsx b/packages/web/components/Citation.tsx index 41034319..9541050f 100644 --- a/packages/web/components/Citation.tsx +++ b/packages/web/components/Citation.tsx @@ -9,7 +9,7 @@ const citationKeyMap: { [key: string]: string } = { source_title: "Source Title", source_name: "Source Name", source_publish_date: "Source Publish Date", - source_url: "Council Meeting", + source_url: "Source URL", }; function isYouTubeURL(url: string): boolean {