How to directly write a resume that mimics scalar_summary?

Question

I would like to get something like scalar_summary , but not be used as Op , which accepts Variable . For example. If i call

 for i in 10000: value = 0.2 * i myfun(tag, value, i) 

In tensor events, I expect it to generate a graph called tag with a line in it.

How can i do this?

Answer 1

Alternatively, if you want to create a TensorBoard log in pure Python code, you can do the following:

 summary_writer = tf.train.SummaryWriter(log_dir) for i in 10000: value = 0.2 * i summary = tf.Summary(value=[tf.Summary.Value(tag=tag, simple_value=value)]) summary_writer.add_summary(summary, global_step=i) summary_writer.close() 

Answer 2

There is no requirement that scalar_summary be Variable . It just has to be a Tensor . You can make a value placeholder and pass many different values ​​inside the for loop.