I want to create a databunch that can load training data from mulitple folders or can say from multiple paths ( for e.g. from 2 different folders). Can someone help me find an answer to this?
If you put multiple folders into the path with the data, say you path is ‘data’ and your different datasets are ‘data1’ and ‘data2’ inside ‘data’, that works without any changes.
No, I have data into two different paths like for e.g. one on harddisk connected to server and other on server itself . Then how can I create a databunch for them ??
Hello @vihari17 , if you are on a linux based system you should be able to implement @ptrampert mentioned. You will just need to create a single directory then within that create virtual directories. For local, you can use a symbolic link, for the external you will need to look up how to mount the remote directory. I am not familiar with your network set up but there should be quite a few options if you google it. This way the system will see it all as one directory.
Thank You Sam !