Can NNI be used in TF Estimator?

[fireindark707]

Environment:

• NNI version: latest
• NNI mode (local|remote|pai): local
• Client OS: Ubuntu
• Server OS (for remote mode only):
• Python version: 3.7
• PyTorch/TensorFlow version: TensorFlow-gpu 1.14.0
• Is conda/virtualenv/venv used?: yes
• Is running in Docker?: no

Log message:

• nnimanager.log:
• dispatcher.log:
• nnictl stdout and stderr:

I want to add the intermediate result to the Tensorflow training, but since the code is executed in estimator.train, I don't know how to get the intermediate result.

logging_hook = tf.train.LoggingTensorHook({"loss": total_loss}, every_n_iter=10)

I see that the code has a loss output like the above, but I don't know how to make nni get the loss

[ultmaster]

SessionRunHook can be inherited and you can implement some hook like LoggingTensorHook and put nni.report_intermediate_result into it. You can refer to the source code of LoggingTensorHook for reference.