PyTorch Dilated Recurrent Neural Networks

PyTorch implementation of Dilated Recurrent Neural Networks (DilatedRNN).

Getting Started

Installation:

pip install -r requirements.txt

Run the tests:

python tests.py

Example

Define a dilated RNN based on GRU cells with 9 layers, dilations 1, 2, 4, 8, 16, ... Then pass the hidden state to a further update

import drnnimport torchn_input = 20n_hidden = 32n_layers = 9cell_type = 'GRU'model = drnn.DRNN(n_input, n_hidden, n_layers, cell_type)x1 = torch.autograd.Variable(torch.randn(23, 2, n_input))x2 = torch.autograd.Variable(torch.randn(23, 2, n_input))out, hidden = model(x1)out, hidden = model(x2, hidden)

Copy Task

python3 -m copy_memory.copymem_test --help

Character LeveL RNN

python3 -m char_rnn.char_rnn_test --help