wcneill

# Assume 26 unique characters
alphabet = ['a', 'b', ... , 'z']

# two sample sequences, inputs and targets
x = np.array(list('abc')) # inputs
y = np.array(list('xyz')) # targets

# define one-hot encoder and label encoder
onehot_encoder = OneHotEncoder(sparse=False).fit(alphabet)
label_encoder  = {ch: i for i, ch in enumerate(alphabet)}

wcneill

input_size  = 50  # representing the one-hot encoded vector size
hidden_size = 100 # number of hidden nodes in the LSTM layer
n_layers    = 2   # number of LSTM layers
output_size = 50  # output of 50 scores for the next character

lstm   = nn.LSTM(input_size, hidden_size, n_layers, batch_first=True)
linear = nn.Linear(hidden_size, output_size)

# Data Flow Protocol
# 1. network input shape: (batch_size, seq_length, num_features)

wcneill

input_size  = 50  # representing the one-hot encoded vector size
hidden_size = 100 # number of hidden nodes in the LSTM layer
n_layers    = 2   # number of LSTM layers
output_size = 50  # output of 50 scores for the next character

lstm   = nn.LSTM(input_size, n_hidden, n_layers, batch_first=True)
linear = nn.Linear(n_hidden, output_size)

wcneill

input_size = 1    # The number of variables in your sequence data. 
n_hidden   = 100  # The number of hidden nodes in the LSTM layer.
n_layers   = 2    # The total number of LSTM layers to stack.
out_size   = 1    # The size of the output you desire from your RNN.

lstm   = nn.LSTM(input_size, n_hidden, n_layers, batch_first=True)
linear = nn.Linear(n_hidden, 1)

# Data Flow Protocol:
# 1. network input shape: (batch_size, seq_length, num_features)

wcneill

input_size = 1    # The number of variables in your sequence data. 
n_hidden   = 100  # The number of hidden nodes in the LSTM layer.
n_layers   = 2    # The total number of LSTM models layers
out_size   = 1    # The size of the output you desire from your RNN

lstm   = nn.LSTM(input_size, n_hidden, n_layers)
linear = nn.Linear(n_hidden, 1)

wcneill

# load in content and style image, and create target image by copying content image
content = load_image('data/style/clouds-19.jpg').to(device)
style = load_image('data/style/abstract-art-freedom.jpg', shape=content.shape[-2:]).to(device)
target = content.clone().requires_grad_(True).to(device)

style_weights = {'conv1_1': .2,
                 'conv2_1': .2,
                 'conv3_1': .2,
                 'conv4_1': .2,
                 'conv5_1': .2}

wcneill

def style_loss(s_grams, t_features, weights):
    """
    Compute style loss, i.e. the weighted sum of MSE of all layers. 
    """
    # for each style feature, get target and style gramians, compare
    loss = 0
    for layer in weights:
        _, d, h, w = s_features[layer].shape
        t_gram = gramian(t_features[layer])    
        

wcneill

def content_loss(c_features, t_features):
    """
    Compute mean squared content loss of all feature maps. 
    """
    loss = 0.5 * (t_features['conv4_2'] - c_features['conv4_2']) ** 2
    return torch.mean(loss)

wcneill

def gramian(tensor):
    t = tensor.view(tensor.shape[1], -1)
    return t @ t.T

wcneill

def get_features(model, image):

    layers = {
        '0' : 'conv1_1',
        '5' : 'conv2_1',
        '10': 'conv3_1',
        '19': 'conv4_1',
        '21': 'conv4_2',
        '28': 'conv5_1'
    }