cd allennlp-multi-label-document-classification
pip install --editable .
For the time being, please install AllenNLPfrom source. You should also install PyTorch with CUDA support by following the instructions for your system here.
Datasets should be JSON lines files, where each line is valid JSON containing the fields "text" and "labels". You can specify different partitions in the configs under "train_data_path", "validation_data_path" and "test_data_path".