comparison transformer_network.py @ 5:9ec705bd11cb draft default tip

planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164

5:9ec705bd11cb

import tensorflow as tf
from tensorflow.keras.layers import (Dense, Dropout, Embedding, Layer,
                                     LayerNormalization, MultiHeadAttention)
from tensorflow.keras.models import Sequential


class TransformerBlock(Layer):
    def __init__(self, embed_dim, num_heads, ff_dim, rate=0.1):
        super(TransformerBlock, self).__init__()
        self.att = MultiHeadAttention(num_heads=num_heads, key_dim=embed_dim, dropout=rate)
        self.ffn = Sequential(
            [Dense(ff_dim, activation="relu"), Dense(embed_dim)]
        )
        self.layernorm1 = LayerNormalization(epsilon=1e-6)
        self.layernorm2 = LayerNormalization(epsilon=1e-6)
        self.dropout1 = Dropout(rate)
        self.dropout2 = Dropout(rate)

    def call(self, inputs, training):
        attn_output, attention_scores = self.att(inputs, inputs, inputs, return_attention_scores=True, training=training)
        attn_output = self.dropout1(attn_output, training=training)
        out1 = self.layernorm1(inputs + attn_output)
        ffn_output = self.ffn(out1)
        ffn_output = self.dropout2(ffn_output, training=training)
        return self.layernorm2(out1 + ffn_output), attention_scores


class TokenAndPositionEmbedding(Layer):
    def __init__(self, maxlen, vocab_size, embed_dim):
        super(TokenAndPositionEmbedding, self).__init__()
        self.token_emb = Embedding(input_dim=vocab_size, output_dim=embed_dim, mask_zero=True)
        self.pos_emb = Embedding(input_dim=maxlen, output_dim=embed_dim, mask_zero=True)

    def call(self, x):
        maxlen = tf.shape(x)[-1]
        positions = tf.range(start=0, limit=maxlen, delta=1)
        positions = self.pos_emb(positions)
        x = self.token_emb(x)
        return x + positions