Hi All,
I have an Azure open AI model reading on a csv file that has around 5000 rows.
I have implemented vector and similarity search. I use text-embedding-3-large and gpt-4 model.
Using the below code, I have restricted to show closest 5 records
results_large = search_client.search(
search_text=query,
vector_queries= [vector_query],
select=select_fields_from_index,
query_type=QueryType.SEMANTIC, semantic_configuration_name=‘my-semantic-config’, query_caption=QueryCaptionType.EXTRACTIVE, query_answer=QueryAnswerType.EXTRACTIVE,
top = 5,)
This works as expected. Business is now asking to show however many issues are available related to a question to be displayed.
For example: If the user questions asking “show me all issues related to a dryer”. If the repository csv has 40 records, they want to show it all.
Is this feasible? How do I approach this?
Any issues that might occur because of widening this search in terms of token or cost?
Any help would be much appreciated