如何在Tensorflow中使用多个摘要集合？

Question

我有两组不同的摘要.每批收集一个,每个时期收集一个.我如何分别用merge_all_summaries(key='???')这两组收集摘要？手动执行它总是一个选项,但似乎有更好的方法.

我认为它应该如何工作的插图:

      # once per batch 
      tf.scalar_summary("loss", graph.loss)
      tf.scalar_summary("batch_acc", batch_accuracy)
      # once per epoch
      gradients = tf.gradients(graph.loss, [W, D])
      tf.histogram_summary("embedding/W", W, collections='per_epoch')
      tf.histogram_summary("embedding/D", D, collections='per_epoch')

      tf.merge_all_summaries()                # -> (MergeSummary...) :)
      tf.merge_all_summaries(key='per_epoch') # -> NONE              :(

Answer 1

问题解决了.collections摘要的参数应该是一个列表.解:

  # once per batch 
  tf.scalar_summary("loss", graph.loss)
  tf.scalar_summary("batch_acc", batch_accuracy)
  # once per epoch
  tf.histogram_summary("embedding/W", W, collections=['per_epoch'])
  tf.histogram_summary("embedding/D", D, collections=['per_epoch'])

  tf.merge_all_summaries()                # -> (MergeSummary...) :)
  tf.merge_all_summaries(key='per_epoch') # -> (MergeSummary...) :)

编辑.TF的句法变化:

# once per batch 
  tf.summary.scalar("loss", graph.loss)
  tf.summary.scalar("batch_acc", batch_accuracy)
  # once per epoch
  tf.summary.histogram("embedding/W", W, collections=['per_epoch'])
  tf.summary.histogram("embedding/D", D, collections=['per_epoch'])

  tf.summary.merge_all()                # -> (MergeSummary...) :)
  tf.summary.merge_all(key='per_epoch') # -> (MergeSummary...) :)