A question about one "OPTION" in your shell scripts

[dlagul]

Set up model

case ${MODEL} in
"gan" ) ;;
"ambientgan" )
OPTION="${OPTION} --noise_measure" ;;
"sinrgan1" )
OPTION="${OPTION} --prior additive_gaussian" ;;
"sinrgan2" )
OPTION="${OPTION} --gn_train --rotation --channel_shuffle --color_inversion" ;;
"sdnrgan1" )
OPTION="${OPTION} --gn_train --prior ${PRIOR}" ;;
"sdnrgan2" )
OPTION="${OPTION} --gn_train --implicit --prior additive_gaussian" ;;
"sdnrgan3" )
OPTION="${OPTION} --gn_train --implicit --color_inversion" ;;
* )
echo "unknown MODEL: ${MODEL}"
exit 1 ;;
esac

I read your paper, an excellent work which inspires me a lot. However, when I view your source code, I have a question about one "OPTION" in your shell scripts.
As shown above, in the case of "sinrgan1", OPTION="${OPTION} --prior additive_gaussian". Refer to you paper, in the case of "SI-NR-GAN-I", do I need to add "--gn_train" to this option, i.e., OPTION="${OPTION} --gn_train --prior additive_gaussian", to train a "netGn" ? I am not sure here, maybe I missed some details. If there is no need to add anything here, could you please explain why.
Your kind reply will help me understand your work well. Thanks a lot.

[takuhirok]

Hi, @dlagul
Thank you for pointing it out. It seems that I have deleted "--gn_train" by mistake when I cleaned up the code. As you indicated, "--gn_train" should be used in the case of "sinrgan1." I have corrected the code now. So, please check out the latest version.

[dlagul]

Hi author, thanks for your kind reply.
Best regards.