Skip to content

Instantly share code, notes, and snippets.

@yaroslavvb

yaroslavvb/scratch.py

Created December 8, 2016 22:51
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save yaroslavvb/b0b9e51d88bb97aa0f739f8130e1c3d5 to your computer and use it in GitHub Desktop.
Save yaroslavvb/b0b9e51d88bb97aa0f739f8130e1c3d5 to your computer and use it in GitHub Desktop.
Download ZIP
Example of using stats summarizer
Raw
scratch.py
import tensorflow as tf
a = tf.ones((10, 10))
b = tf.ones((10, 10))
c = tf.matmul(a, b)
ss = tf.contrib.stat_summarizer.NewStatSummarizer(tf.get_default_graph().as_graph_def().SerializeToString())
sess = tf.Session()
for i in range(10):
run_metadata = tf.RunMetadata()
run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
sess.run(c, options=run_options, run_metadata=run_metadata)
ss.ProcessStepStatsStr(run_metadata.step_stats.SerializeToString())
print(ss.GetOutputString())
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment