The link of RCV1 dataset is invalid

[AppleXY]

The link of RCV1 dataset is invalid

[YipingNUS]

You can know the format of the data by looking at the load_data method.

In the line, you see the data is pickle files containing four attributes (the last two are never used and can thus ignore).

[train, test, vocab, catgy] = pickle.load(fin)

Then looking at the load_data_and_labels method, you see the train/test data are a list of document dicts with key 'text' for the plain text document and 'catgy' for the label.

There's another closed issue providing a link to some other datasets used in the paper.

[purviprajapati196]

Please provide .p file for eurlex, wiki10, amazonCat datasets