如何为 TensorBoard 使用带有“tf.train.Supervisor”的多个摘要编写器

Question

我想做一些类似于 TensorBoard tutorial 中的 train_writer 和 test_writer 的事情。但是使用 tf.train.Supervisor。然而，我不确定如何最好地解决这个问题。

伪代码：

train_op = #...
train_summaries = # ...
test_summaries = # ...

config = tf.ConfigProto(allow_soft_placement=True)
sv = tf.train.Supervisor(
    logdir              = ????,
    summary_op          = ????,
    summary_writer      = ????,
)
with sv.managed_session(config=config) as sess:
    while not sv.should_stop():
        sess.run(train_op)

所以我的问题是：如何保存 train_summaries 和 test_summaries 做不同的目录？ 例如./logdir/train 和 ./logdir/test/.

Answer 1

您正在寻找 summary_computed。文档字符串显示了如何创建自定义摘要编写器。你不能让 Supervisor 自动管理它，但它很简单。来自 https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/training/supervisor.py

# Create a Supervisor with no automatic summaries.
sv = Supervisor(logdir='/tmp/mydir', is_chief=is_chief, summary_op=None)
# As summary_op was None, managed_session() does not start the
# summary thread.
with sv.managed_session(FLAGS.master) as sess:
  for step in xrange(1000000):
    if sv.should_stop():
      break
    if is_chief and step % 100 == 0:
      # Create the summary every 100 chief steps.
      sv.summary_computed(sess, sess.run(my_summary_op))
    else:
      # Train normally
      sess.run(my_train_op)

如何为 TensorBoard 使用带有“tf.train.Supervisor”的多个摘要编写器

How to use multiple summary writers with `tf.train.Supervisor` for TensorBoard

deep-learning

tensorflow

tensorboard