description: For the given group, show the detailed breakdown of amounts different members owe to each other. If group_id is not

Evaluate using RAG-based retrieval on evaluation set with batch processing. Supports @n evaluation: generates n candidate answers and

Generate responses for multiple prompts using batch inference if available. Args: prompts: List of user prompts

Load data from a JSONL file. Args: file_path: Path to the JSONL file Returns: List of dictionaries, one per

Create a batch model function using async batch client or sequential requests. Args: clients: List of OpenAI client instances fo

Sends phone message to a user using phone number of the given app, e.g., amazon, venmo, flights, etc. This is used for app-specific n

Removes docstrings (multi-line triple-quote) and "#.." comments from python code.

Extract number from LaTeX formatted text. Args: text: Model response text with LaTeX formatting using_boxed: Whether to

Extract the final answer from model response. Args: text: Model response text using_boxed: Whether to extract from boxed

Retrieve relevant guidance for each subject in the batch. Args: subjects: List of subjects (required - should be

Default assert with additional features: 1. better messages out-of-the-box: Shows what left and right values are. Can be overridden by