Fix data path issue which caused only a subset of data files to be used.
previously the logic is incorrect and resulted in only a few of the available files to be used for training. I think I've fixed it so it accomplished the intended logic. Please check it out.
Also tested by installing via pip from my fork. It is indeed doing much better to use all the file in the folder.
Fix data path issue which caused only a subset of data files to be used.
previously the logic is incorrect and resulted in only a few of the available files to be used for training. I think I've fixed it so it accomplished the intended logic. Please check it out.
Also tested by installing via pip from my fork. It is indeed doing much better to use all the file in the folder.