ctga-v1_alcapa.py

import json

with open('ctga-v1.json', 'r', encoding='utf-8') as f:
    data = json.load(f)

# Convert to the simpler format
new_data = []
for item in data:
    new_item = {
        'instruction': item['task_input'].replace('{{context}}', item['context']),
        'input': item['task_type'],
        'output': item['task_output']
    }
    new_data.append(new_item)

new_data_json = json.dumps(new_data, ensure_ascii=False, indent=4)
with open('alcapa_format.json', 'w', encoding='utf-8') as f:
    f.write(new_data_json)