↓ 2 callersFunctionpermute Examples: x.shape == [128, 32, 1] x = permute(x, [0, 2, 1]) x.shape == [128, 1, 32] y.shape == [128, 64, 32]
_codes/my_tensorflow/src/utils/array_op.py:12
↓ 1 callersFunctionhighway_conv2d用于 conv2d 的 highway Input shape: [batch_size, in_h, in_w, in_channels] Output shape: [batch_size, in_h, in_w, in_channels] 公式 `o
_codes/my_tensorflow/src/layers/highway.py:51
↓ 1 callersFunctionhighway_dense用于全连接层的 highway Input shape: [batch_size, n_input] Output shape: [batch_size, n_input] 公式 `o = H(x, W)T(x, W) + x(1 - T(x, W))`
_codes/my_tensorflow/src/layers/highway.py:15
Functionattention_flowAttention Flow Match Layer Input shape: h: [N, T, d] # 原文中的 shape 为 [N, T, 2d], 因为经过了 bi-LSTM, 维度扩了一倍 u: [N, J, d] Output sha
_codes/my_tensorflow/src/layers/match/attention_flow.py:62
Functionattention_flow_selfAttention Flow Self Match Layer Input shape: h: [N, T, d] # 原文中的 shape 为 [N, T, 2d], 因为经过了 bi-LSTM, 维度扩了一倍 u: [N, J, d] Outpu
_codes/my_tensorflow/src/layers/match/attention_flow.py:17
Functionconv2d2-D 卷积层 Input shape: [batch_size, in_h, in_w, in_channels] Output shape: [batch_size, out_h, out_w, out_channels] Args: x(tf.Ten
_codes/my_tensorflow/src/layers/cnn.py:17