Set trainable layers while doing transfer learning

[behzad1380]

Description

I am using t2t-trainer to train an image classifier. I have set the followings:
--hparams_set=resnet_50 --model=resnet --hparams='batch_size=4'
and no hidden layers on top of resnet_50. My impression is that, it trains the entire resnet from scratch. I couldn't find the parameters to set trainable layers or freeze layers?
...

Environment information

OS: <your answer here>

$ pip freeze | grep tensor
# your output here

$ python -V
# your output here

For bugs: reproduction and error logs

# Steps to reproduce:
...

# Error logs:
...