Speed checks using Tensorboard and a chrome trace file

[adbailey4]

nanotensor.py

• made it possible to save a trace file ("timeline.json") to look at the time of computations within tensorflow. All you need to do is type in chrome://tracing/ into chrome and load the json file and it will show the speed of one training computation.

nanotensor1.config.json

added new parameters

• "save_model" - Is an integer value in seconds which corresponds to how often you save a model Saving a model is very slow and should be done rarely when training for a long time
• "record_step" - Integer value corresponding to the number of training iterations which occur before saving summary data regarding training and testing accuracy and cost
• "trace_name" - Name of trace file (trace files are saved as json so it should have .json extension)
• "save_trace" - boolean for deciding to save a trace of a training step (Saves last training step)

network.py

• Added training and testing summary options when reporting summary statistics in tensorboard

[codecov-io]

Codecov Report

:exclamation: No coverage uploaded for pull request base (master@184405d). Click here to learn what that means. The diff coverage is 0%.

@@           Coverage Diff            @@
##             master     #29   +/-   ##
========================================
  Coverage          ?   5.51%           
========================================
  Files             ?      10           
  Lines             ?    1070           
  Branches          ?       0           
========================================
  Hits              ?      59           
  Misses            ?    1011           
  Partials          ?       0

Impacted Files	Coverage Δ
nanotensor/network.py	0% <0%> (ø)
nanotensor/nanotensor.py	0% <0%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 184405d...211ae5b. Read the comment docs.