How can I evaluate a trained model using landmark.py

[abdou31]

Hello Yin, Is it possible to evaluate a trained model without training and creating a new model with command that you give for example ( 500 steps ). I have a trained model that have 135000 steps, and I want to just evaluate it without training it again, How can I do that?

[yinguobing]

Use --eval_only.

[abdou31]

The argument --eval_only does not exist in landmark.py Please try to update your code