Function get_most_preferred

toolbench/tooleval/automatic_eval_sample.py:88–96

(d:list)

            [ref_ans,ans])
        return qid,ret
    def get_most_preferred(d:list)->np.ndarray:
        if np.iterable(d):
            d = np.asanyarray(d)
            bins = np.bincount(d)
            max_val = np.max(bins)
            argmax = np.where(max_val==bins)[0]
            return argmax
        else:
            return np.asarray([d])
    
    print('Evaluating...')
    prefer_dict = {}