This repositoy contains the PyTorch code for convolutional LSTM used in AnDi Challenge.
the model uses two convolutional layers followed by a 3 layer bidirectional LSTM
- The trained model which was submitted to the andi challenge "85_conv_LSTM_350K". This model can be used by using torch.load() in a jupyter notebook that contains the class ConejeroConvNet.
- Model Trainer This notebook contains all the code necessary to train our convolutional LSTM.