Searched refs:num_heads (Results 1 – 5 of 5) sorted by relevance
/external/tensorflow/tensorflow/python/keras/layers/ |
D | multi_head_attention_test.py | 43 num_heads=12, 56 num_heads=12, key_dim=64) 65 num_heads=12, key_dim=64) 75 num_heads=12, key_dim=64) 87 num_heads=2, key_dim=2, use_bias=use_bias) 137 num_heads=12, 148 num_heads=2, key_dim=2) 201 num_heads=2, key_dim=2, attention_axes=attention_axes) 230 num_heads=2, key_dim=2, dropout=0.5) 264 test_layer = SubclassAttention(num_heads=12, key_dim=64) [all …]
|
D | multi_head_attention.py | 218 num_heads, argument 234 self._num_heads = num_heads
|
/external/tensorflow/tensorflow/python/keras/benchmarks/keras_examples_benchmarks/ |
D | text_classification_transformer_benchmark_test.py | 42 num_heads = 2 48 transformer_block = TransformerBlock(embed_dim, num_heads, ff_dim) 145 def __init__(self, embed_dim, num_heads=8): argument 148 self.num_heads = num_heads 149 if embed_dim % num_heads != 0: 152 self.projection_dim = embed_dim // num_heads 167 x = tf.reshape(x, (batch_size, -1, self.num_heads, self.projection_dim)) 197 def __init__(self, embed_dim, num_heads, ff_dim, rate=0.1): argument 199 self.att = MultiHeadSelfAttention(embed_dim, num_heads)
|
/external/tensorflow/tensorflow/tools/api/golden/v2/ |
D | tensorflow.keras.layers.-multi-head-attention.pbtxt | 132 …argspec: "args=[\'self\', \'num_heads\', \'key_dim\', \'value_dim\', \'dropout\', \'use_bias\', \'…
|
/external/tensorflow/tensorflow/tools/api/golden/v1/ |
D | tensorflow.keras.layers.-multi-head-attention.pbtxt | 132 …argspec: "args=[\'self\', \'num_heads\', \'key_dim\', \'value_dim\', \'dropout\', \'use_bias\', \'…
|