如何使用tensorboard在直方图中可视化每层多个内核

import tensorflow as tf
from tensorflow.python.framework import ops

# 定义输入和标签的占位符
x = tf.placeholder(tf.float32, shape=[None, input_size], name='input')
y = tf.placeholder(tf.float32, shape=[None, num_classes], name='labels')

# 定义模型的权重和偏置变量
weights = {
    'w1': tf.Variable(tf.random_normal([input_size, hidden_size1]), name='weight1'),
    'w2': tf.Variable(tf.random_normal([hidden_size1, hidden_size2]), name='weight2'),
    'w3': tf.Variable(tf.random_normal([hidden_size2, num_classes]), name='weight3')
}
biases = {
    'b1': tf.Variable(tf.random_normal([hidden_size1]), name='bias1'),
    'b2': tf.Variable(tf.random_normal([hidden_size2]), name='bias2'),
    'b3': tf.Variable(tf.random_normal([num_classes]), name='bias3')
}

# 定义模型的前向传播过程
hidden_layer1 = tf.add(tf.matmul(x, weights['w1']), biases['b1'])
hidden_layer2 = tf.add(tf.matmul(hidden_layer1, weights['w2']), biases['b2'])
output_layer = tf.add(tf.matmul(hidden_layer2, weights['w3']), biases['b3'])

# 定义损失函数和优化器
loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=output_layer, labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(loss)

# 创建TensorBoard的summary操作
tf.summary.histogram('weight1', weights['w1'])
tf.summary.histogram('weight2', weights['w2'])
tf.summary.histogram('weight3', weights['w3'])
tf.summary.histogram('bias1', biases['b1'])
tf.summary.histogram('bias2', biases['b2'])
tf.summary.histogram('bias3', biases['b3'])

# 合并所有summary操作
merged_summary = tf.summary.merge_all()

# 创建一个用于写入TensorBoard日志的写入器
summary_writer = tf.summary.FileWriter(logdir)

# 在训练过程中，将summary操作添加到训练步骤中
with tf.Session() as sess:
    # ...

    # 在每个训练步骤中运行merged_summary操作，并将结果写入日志文件
    _, summary = sess.run([optimizer, merged_summary], feed_dict={x: batch_x, y: batch_y})
    summary_writer.add_summary(summary, global_step=step)

# 启动TensorBoard服务器
tensorboard --logdir=logdir

# 在浏览器中打开TensorBoard的网址
http://localhost:6006