Hi,
I’ve been having problems trying to load my own qa dataset, structured in the squad_2 format.
When the following is executed:
load_dataset(“Lisibonny/periodicos_dominicanos_question_answering_dataset”, name=‘v1.1.0’)
I get the error JSONDecodeError: Expecting value: line 1 column 1 (char 0). I’m using a loading script (Attached you will find a capture of the function that is giving me the error). The specific line is data = json.load (f).
When checking the value of f it points out to a local path, where apparently the datasets are cached:
<_io.TextIOWrapper name=‘/root/.cache/huggingface/datasets/downloads/9d3ecbece8440c15a4497a3df9a2309ba197728a28b956d749b471bbb2b2f914’ mode=‘r’ encoding=‘utf-8’>
The problem is that when I check the file the structure is not that of a json, but a HTML where none of my data appear.
Can somebody help me to see where the problem is?
Thanks.
