使用FOR循环通过TensorFlow初始化权重

import tensorflow as tf

input_size = 100
output_size = 10
hidden_size = 50

# 输入层
inputs = tf.placeholder(tf.float32, shape=[None, input_size])

# 隐藏层
weights_hidden = tf.Variable(tf.zeros([input_size, hidden_size]))
biases_hidden = tf.Variable(tf.zeros([hidden_size]))
hidden_layer = tf.nn.relu(tf.matmul(inputs, weights_hidden) + biases_hidden)

# 输出层
weights_output = tf.Variable(tf.zeros([hidden_size, output_size]))
biases_output = tf.Variable(tf.zeros([output_size]))
outputs = tf.matmul(hidden_layer, weights_output) + biases_output

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)

with tf.Session() as sess:
    sess.run(init)

    # 初始化隐藏层权重
    for i in range(input_size):
        for j in range(hidden_size):
            weights_hidden[i, j] = tf.random_normal([1], stddev=0.1)

    # 初始化输出层权重
    for i in range(hidden_size):
        for j in range(output_size):
            weights_output[i, j] = tf.random_normal([1], stddev=0.1)