Files
2024-09-25 18:04:29 +08:00

26 lines
823 B
Python

from data import get_bigcodebench, write_jsonl
subset = 'hard'
split = 'complete'
dataset = get_bigcodebench(subset=subset)
task = f'bigcodebench/bcb_{split}_{subset}'
outputs = []
for key, value in dataset.items():
outputs.append({
'task_id': value['task_id'],
'prompt': value['complete_prompt'] if split == 'complete' else value['instruct_prompt'],
'complete_prompt': value['complete_prompt'],
'instruct_prompt': value['instruct_prompt'],
'canonical_solution': value['canonical_solution'],
'code_prompt': value['code_prompt'],
'test': value['test'],
'entry_point': value['entry_point'],
'doc_struct': value['doc_struct'],
'libs': value['libs'],
'task': task,
})
save_path = f'{task}.jsonl'
write_jsonl(save_path, outputs)