Function do_eval

code_generation_benchmark/main.py:174–179

(logger, task)

        logger.error(str(e))

def do_eval(logger, task):
    data_path = Path(task['data_path'])
    model_type = task['model_type']
    setting_name = task['setting_name']
    with file_lock:
        single_exp(logger, data_path, model_type, setting_name=setting_name, eval_only=True, low_level=True, high_level=True, eval_model=EVAL_MODEL)

def judge_need_to_chat(model_dir: Path):
    sname = model_dir.name