Hello OpenAI Community,
I’m working with the OpenAI Assistants API and have a question regarding the function_call_output
method.
When submitting the output of a tool call using function_call_output
, is it necessary to include the original instructions
and tools
definitions in the submission?
From my understanding, instructions
and tools
are typically defined during the initial setup of the assistant or when creating a run. Once set, they persist throughout the session. Therefore, I’m wondering if re-including them with each tool output submission is redundant or required.
inputList.Add(new
{
type = "function_call_output",
call_id = to.ToolCallId,
output = to.Output,
});
var payload = new
{
model = assistantInfo.Model,
instructions = assistantInfo.Instructions, // Is this needed here?
input,
tools = TranslateToolsToResponsesFormat(assistantInfo.Tools), // Is this needed here?
tool_choice = "auto",
temperature = assistantInfo.Temperature,
top_p = assistantInfo.TopP,
previous_response_id = threadState.LastResponseId,
stream = true,
store = true
};
Any insights or guidance on this matter would be greatly appreciated.
Thank you!