training setting

#5
by Znlunar5 - opened

Hello author, may I ask if the training model is designed for optical-optical and optical-SAR, respectively?

Yes. They were fine-tuned respectively.

Dear authors, thank you for your reply. I would like to further confirm whether “fine-tuned respectively” means that the Optical-Optical and Optical-SAR results were obtained from two separate fine-tuned checkpoints? Specifically, was the Optical-Optical model fine-tuned using only pre-optical + post-optical samples, and the Optical-SAR model fine-tuned using only pre-optical + post-SAR samples? Does this setting apply to both QA and referring segmentation models? Thank you!

Sign up or log in to comment