新增流式读取数据集报错

[msn199959]

Traceback (most recent call last):
File "src/train_bash.py", line 23, in
main()
File "src/train_bash.py", line 8, in main
run_pt(model_args, data_args, training_args, finetuning_args)
File "/cpfs01/shared/public/stream-llama/LLaMA-Efficient-Tuning/src/llmtuner/tuner/pt/workflow.py", line 28, in run_pt
dataset = preprocess_dataset(dataset, tokenizer, data_args, training_args, stage="pt")
File "/cpfs01/shared/public/stream-llama/LLaMA-Efficient-Tuning/src/llmtuner/dsets/preprocess.py", line 176, in preprocess_dataset
print_unsupervised_dataset_example(next(iter(dataset)))
File "/cpfs01/user/wkspace/envs/vicuna/lib/python3.8/site-packages/datasets/iterable_dataset.py", line 1353, in iter
for key, example in ex_iterable:
File "/cpfs01/user/wkspace/envs/vicuna/lib/python3.8/site-packages/datasets/iterable_dataset.py", line 956, in iter
for x in self.ex_iterable:
File "/cpfs01/user/wkspace/envs/vicuna/lib/python3.8/site-packages/datasets/iterable_dataset.py", line 652, in iter
yield from self._iter()
File "/cpfs01/user/wkspace/envs/vicuna/lib/python3.8/site-packages/datasets/iterable_dataset.py", line 704, in _iter
raise ValueError(
ValueError: Column lengths mismatch: columns ['input_ids', 'attention_mask', 'labels'] have length [10, 10, 10] while prompt has length 50.

[hiyouga]

命令行参数发一下

[yuanzhiyong1999]

最后怎么解决的呢