report.pdf is the report of our work on Extreme Classification using Autoencoders. In the report, we have provided results of our methods on 4 datasets : Bibtex, Delicious, Mediamill and Eurlex. Please find the links to download these datasets below:
Please go through the instructions in the main README about using these datasets. For Bibtex, Delicious and Mediamill, please remove the top line and run ./split_maker_small_dataset.sh to get the appropriate split files. Please note that the training scripts are expected to run only run on these splits and not the original splits, which are merely indices.