stable plaza
Hey! I'm looking for guidance on this topic,
I'm creating a file_search assistant, that leverages few documents in a vector_store_id to perform a search, while using the OpenAI playground (agent and threads are created in my custom backend i'm just using the playground for testing purposes) i'm receiving really accurate responses, but in my Mobile app using the API i'm getting not-related / not accurate response, I've tryed to add the following parameters to the payload:
"ranking_options": [ "score_threshold": 0.7, "ranker": "default_2024_08_21" ]
But i'm receiving:
"error": { "message": "Unknown parameter: 'tool_resources.file_search.ranking_options'.", "type": "invalid_request_error", "param": "tool_resources.file_search.ranking_options", "code": "unknown_parameter" }
and before adding this I noticed that: "score_threshold": 0.0 is getting received which in my brain clicks on the VS is not giving me the most accurate result but the least likely result according to me.
"tools": [
{
"type": "file_search",
"file_search": {
"max_num_results": 10,
"ranking_options": {
"ranker": "default_2024_08_21",
"score_threshold": 0.0
}
}
}
],
"tool_resources": {
"file_search": {
"vector_store_ids": [
"vs_......"
]
}
},
"metadata": {},
"temperature": 0.21,
"top_p": 1.0,
"reasoning_effort": null,
"max_completion_tokens": null,
"max_prompt_tokens": null,
"truncation_strategy": {
"type": "auto",
"last_messages": null
},
"incomplete_details": null,
"usage": null,
"response_format": "auto",
"tool_choice": "auto",
"parallel_tool_calls": true
}
Anyone have experience this before?
Any guidance to navigate this issue?
Am I doing something wrong I have not realized yet? (likely)
grim flume