python - 如何从 Google Colab 的张量板上获取直方图

Question

我正在使用名为“tensorboardcolab”的包。但到目前为止，我只能根据以下代码从 tensorboard 获取图形：

!pip install -U tensorboardcolab
from tensorboardcolab import *

tbc=TensorBoardColab()

tf.reset_default_graph()

with tf.get_default_graph().as_default():
    x = tf.placeholder(tf.float32, [], name="x")
    y = tf.placeholder(tf.float32, [], name="y")
    k = tf.get_variable("k", shape=[])
    b = tf.get_variable("b", shape=[])
    loss = tf.square(k * x + b - y, name="square_error")

optimizer = tf.train.GradientDescentOptimizer(0.5)
train_op = optimizer.minimize(loss)

with tf.Session() as sess:
# Initialize `k` and `b`
sess.run(tf.global_variables_initializer())
for i in range(100):
    # y = 3x-1 + noise
    data_x = np.random.random()
    data_y = 3.0 * data_x - 1.0 + 0.0001*np.random.randn()

    k_val, b_val, loss_val, _ = sess.run([k, b, loss, train_op],
                                         feed_dict={x: data_x, y: data_y})
    if i % 20 == 0:
        print("k, b, loss: ", k_val, b_val, loss_val)

sess.close()

train_writer = tbc.get_writer();
train_writer.add_graph(sess.graph)

train_writer.flush();
tbc.close()

---

我可以添加一些代码来可视化直方图吗？我不知道要填写“write_grads = True”等参数。

标签： pythontensorflowdata-visualizationtensorboardgoogle-colaboratory