Datasets:

google

/

code_x_glue_cc_clone_detection_poj104

like 8

Follow

Google 4,337

Originally posted at https://github.com/microsoft/CodeXGLUE/issues/135, I think this is the more proper place to get help:

I get this error when I try to load your POJ-104 dataset from huggingface using load_dataset.

NonMatchingSplitsSizesError: [{'expected': SplitInfo(name='train', num_bytes=18878686, num_examples=32000, dataset_name='code_x_glue_cc_clone_detection_poj104'), 'recorded': SplitInfo(name='train', num_bytes=20179075, num_examples=32500, dataset_name='code_x_glue_cc_clone_detection_poj104')}, {'expected': SplitInfo(name='validation', num_bytes=5765303, num_examples=8000, dataset_name='code_x_glue_cc_clone_detection_poj104'), 'recorded': SplitInfo(name='validation', num_bytes=6382433, num_examples=8500, dataset_name='code_x_glue_cc_clone_detection_poj104')}]

As far as I can tell, the dataset expects to load 500 fewer examples than the downloaded files contain. I attached a notebook which reproduces the issue:

• Notebook (run with Python 3.8): test_poj104.zip
• Output: test_poj104.pdf

Could you fix the issue so that we can load the dataset correctly without ignore_verifications=True?