I'm trying to write some code to do structured extraction as specified in
https://docs.llamaindex.ai/en/stable/examples/output_parsing/function_program/. I'm using Claude-3.5. On some invocations it will fail to do a tool call, and I get the message
ValueError: Expected at least one tool call, but got 0 tool calls.. The
FunctionCallingProgram class takes in a tool_choice parameter that in theory I could use to force it to use the one tool provided and give me my structured output, but as far as I can tell the value is never passed to the LLM and is not used in the __call__ method. Am I missing something or should I fill a pull request?
