Update files from the datasets library (from 1.9.0)

Release notes: https://github.com/huggingface/datasets/releases/tag/1.9.0

README.md

@@ -419,7 +419,7 @@ The "types" are are:
 - `en-valid` and `en-valid-10k`
    - are the same as `en` and `en10k` except the train sets have been conveniently split into train and valid portions (90% and 10% split).
 
-To get a particular dataset, use `load_datasets('babi_qa',type=f'{type}',task_no=f'{task_no}')` where `type` is one of the types, and `task_no` is one of the task numbers. For example, `load_dataset('babi_qa', type='en', task_no='qa1')`.
+To get a particular dataset, use `load_dataset('babi_qa',type=f'{type}',task_no=f'{task_no}')` where `type` is one of the types, and `task_no` is one of the task numbers. For example, `load_dataset('babi_qa', type='en', task_no='qa1')`.
 ### Languages