TypeError: string indices must be integers

[Qzy568]

您好，我想运行的代码infer_gpt.py，输入python infer/infer_gpt.py gpt-4o cot vision
但是报如下错误，报错函数在vision_mmmu_doc_to_visual(doc)
Generating train split: 1730 examples [00:03, 455.85 examples/s]
Begin processing vision
Processing vision: 0%| | 0/1 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/workspace/mm_eval/qzy/code/MMMU/mmmu-pro/infer/infer_gpt.py", line 225, in
main()
File "/workspace/mm_eval/qzy/code/MMMU/mmmu-pro/infer/infer_gpt.py", line 221, in main
run_and_save()
File "/workspace/mm_eval/qzy/code/MMMU/mmmu-pro/infer/infer_gpt.py", line 216, in run_and_save
temp_files.append(process_and_save_part(dataset, SETTING, model_components))
File "/workspace/mm_eval/qzy/code/MMMU/mmmu-pro/infer/infer_gpt.py", line 207, in process_and_save_part
result, data = future.result()
File "/root/miniconda3/envs/humanevalv/lib/python3.10/concurrent/futures/_base.py", line 451, in result
return self.__get_result()
File "/root/miniconda3/envs/humanevalv/lib/python3.10/concurrent/futures/_base.py", line 403, in __get_result
raise self._exception
File "/root/miniconda3/envs/humanevalv/lib/python3.10/concurrent/futures/thread.py", line 58, in run
result = self.fn(*self.args, **self.kwargs)
File "/workspace/mm_eval/qzy/code/MMMU/mmmu-pro/infer/infer_gpt.py", line 172, in process_prompt
images = vision_mmmu_doc_to_visual(data)
File "/workspace/mm_eval/qzy/code/MMMU/mmmu-pro/infer/infer_gpt.py", line 71, in vision_mmmu_doc_to_visual
return [doc['image']]
TypeError: string indices must be integers

[Zheng0428]

看起来doc在你这里是一个字符串，但实际上他是一个字典。
可以检查一下load_dataset的dataset是否正确，我看你这里显示的是Generating train split，但脚本里面是split='test'，可以check一下是否是这个原因导致。希望可以解决你的问题