Model Test and code mistyping
ZhihuaLiuEd opened this issue · comments
Hi Katie:
Thanks for sharing your code.
-
If I just change the "is_train" to False to proceed model testing. Feels like it still uses the training data to do the test? As the model want to load checkpoint with "self.model_dir", but the "self.water_dataset_name" is coming from the param in my python command line. Therefore in order to get the saved checkpoint, I have to input the training dataset during testing.
-
modelmhl.py line 506, "out_name3 = self.resuts_dir + out_file3" -> "out_name3 = self.results_dir + out_file3"
Cheers
Zhihua