Way too huge event files - Disabled feature for GPU in model_lib_v2.py

Prerequisites

Please answer the following question for yourself before submitting an issue.

• I checked to make sure that this feature has not been requested already.

1. The entire URL of the file you are using

https://github.com/tensorflow/models/blob/master/research/object_detection/model_lib_v2.py#L553

2. Describe the feature you request

I would like training on gpu to write less often to event files, since it's a waste of I/O and disk space

3. Additional context

I wonder what is the reason for this feature to be disabled? Only a comment about GPU issues, while it seems to be working just fine on my setup.

if use_tpu:
  num_steps_per_iteration = 100
else:
  # TODO(b/135933080) Explore setting to 100 when GPU performance issues
  # are fixed.
  num_steps_per_iteration = 1

4. Are you willing to contribute it? (Yes or No)

Yes