process_verifier_data.py

from datasets import load_dataset

data = load_dataset('TIGER-Lab/verifier-sft-data')

def process_data(data):
    response = data['response']
    response = response.split('Final Decision:')[-1].strip()
    data['response'] = 'Final Decision: ' + response
    return data

data = data.map(process_data)

data.push_to_hub('MrLight/verifier-sft-data-scale', private=True)