使用keras的backend對input進行conv1d

from keras import backend as K
a=K.variable(K.random_normal(shape=(5,10,40)))
w=K.variable(K.truncated_normal(shape=(3,40,128)))
conv_layer_3=K.conv1d(a,w,data_format='channels_last')
           

K.conv1d(x,kernel,strides=1,padding='valid',data_format=None,dilation_rate=1)#下面是以data_format='channels_last'的狀态來給x,kernel指派的
           

x: input tensor #形如(batch_size,input_length,embedding_dim)
           

kernel: kernel(filter) tensor #形如(output_channels,filter_rows,filter_columns)
           

stride: 表每次視窗滑動的步長
           

padding: 填充方式