Created
June 25, 2020 13:27
-
-
Save JanSchm/ca65833364e3bcf4314199a5a91ff01a to your computer and use it in GitHub Desktop.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| class TransformerEncoder(Layer): | |
| def __init__(self, d_k, d_v, n_heads, ff_dim, dropout=0.1, **kwargs): | |
| super(TransformerEncoder, self).__init__() | |
| self.d_k = d_k | |
| self.d_v = d_v | |
| self.n_heads = n_heads | |
| self.ff_dim = ff_dim | |
| self.attn_heads = list() | |
| self.dropout_rate = dropout | |
| def build(self, input_shape): | |
| self.attn_multi = MultiAttention(self.d_k, self.d_v, self.n_heads) | |
| self.attn_dropout = Dropout(self.dropout_rate) | |
| self.attn_normalize = LayerNormalization(input_shape=input_shape, epsilon=1e-6) | |
| self.ff_conv1D_1 = Conv1D(filters=self.ff_dim, kernel_size=1, activation='relu') | |
| self.ff_conv1D_2 = Conv1D(filters=7, kernel_size=1) # input_shape[0]=(batch, seq_len, 7), input_shape[0][-1]=7 | |
| self.ff_dropout = Dropout(self.dropout_rate) | |
| self.ff_normalize = LayerNormalization(input_shape=input_shape, epsilon=1e-6) | |
| def call(self, inputs): # inputs = (in_seq, in_seq, in_seq) | |
| attn_layer = self.attn_multi(inputs) | |
| attn_layer = self.attn_dropout(attn_layer) | |
| attn_layer = self.attn_normalize(inputs[0] + attn_layer) | |
| ff_layer = self.ff_conv1D_1(attn_layer) | |
| ff_layer = self.ff_conv1D_2(ff_layer) | |
| ff_layer = self.ff_dropout(ff_layer) | |
| ff_layer = self.ff_normalize(inputs[0] + ff_layer) | |
| return ff_layer |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment