How do I set --input_dims option in caffe_translator.py

Hello,

I want to translate caffe model to caffe2 model using caffe_translator.py.
but I cannot figure out how to specify the --input_dims options.

usage: caffe_translator.py [-h] [--init_net INIT_NET]
                           [--predict_net PREDICT_NET] [--remove_legacy_pad]
                           [--input_dims INPUT_DIMS [INPUT_DIMS ...]]
                           prototext caffemodel

Is there anybody explain the usage of --input_dims with caffe’ mnist tutorial
at http://caffe.berkeleyvision.org/gathered/examples/mnist.html.

Thanks.

This question is worth a Billion Dollars.

python caffe_translator.py --remove_legacy_pad new.prototxt new.caffemodel --input_dims 1 3 210 280

putting it at the end
this worked for me