如何在keras模型架构中实现tensorflow2层，tf.nn.conv1d_transpose？

import tensorflow as tf
from tensorflow.keras.layers import Lambda

def conv1d_transpose_layer(input_shape, filters, kernel_size, strides):
    def conv1d_transpose(x):
        return tf.nn.conv1d_transpose(x, filters, output_shape=input_shape, strides=strides)
    
    return Lambda(conv1d_transpose)

# 定义输入形状
input_shape = (batch_size, input_length, input_channels)
# 定义卷积核参数
filters = tf.Variable(tf.random.normal(shape=(kernel_size, input_channels, output_channels)))
kernel_size = 3
strides = 2

# 创建模型
model = tf.keras.Sequential()
# 添加Lambda层来实现conv1d_transpose
model.add(conv1d_transpose_layer(input_shape, filters, kernel_size, strides))
# 添加其他层...

# 编译模型
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])