David Roesler

David Roesler

Computational Linguistics Graduate Student, University of Washington

Deep Averaging Network (2021)

A Pytorch implementation of the Deep Averaging Network introduced in Iyyer et al (2015). Performs binary sentiment classification on the IMDB reviews dataset.

Github repository link to my code

About the code

Dependencies: See environment.yml for the Conda environment.

Command line parameters:

    # model arguments
    parser.add_argument('--embedding_dim', type=int, default=300)
    parser.add_argument('--hidden_dim', type=int, default=300)
    # training arguments
    parser.add_argument('--batch_size', type=int, default=64)
    parser.add_argument('--seed', type=int, default=572)
    parser.add_argument('--num_epochs', type=int, default=15)
    parser.add_argument('--patience', type=int, default=None)
    parser.add_argument('--L2', action="store_true")
    # data arguments
    parser.add_argument('--data_dir', type=str, default='/data')
    parser.add_argument('--vocab_size', type=int, default=20000)
    parser.add_argument('--padding_index', type=int, default=1)

Example output:

/images/DAN_output.png
Results for –num epochs 12 –patience 3 –L2.