Use DeepSpeed load myself " .csv " dataset.

### Describe the bug

When I use DeepSpeed train a model with my own " XXX.csv" dataset I got the follow question:
Traceback (most recent call last):
File "", line 1, in
File "/home/fm001/.conda/envs/hzl/lib/python3.8/site-packages/datasets/load.py", line 1767, in load_dataset
builder_instance = load_dataset_builder(
File "/home/fm001/.conda/envs/hzl/lib/python3.8/site-packages/datasets/load.py", line 1498, in load_dataset_builder
dataset_module = dataset_module_factory(
File "/home/fm001/.conda/envs/hzl/lib/python3.8/site-packages/datasets/load.py", line 1217, in dataset_module_factory
raise FileNotFoundError(
FileNotFoundError: Couldn't find a dataset script at /home/fm001/hzl/Data/qa.csv/qa.csv.py or any data file in the same directory.

### Steps to reproduce the bug

my code is :
from datasets import load_dataset
mydata = load_dataset("/home/fm001/hzl/Data/qa.csv")

### Expected behavior

。。。

### Environment info

。。。