Replies: 2 comments
-
@woshigui you can do it manually by organizing your dataset folders by fold or by hacking the code w/ extra metadata and args to filter the folds, scripts are hackable to allow that, not attempting to cover all possible use cases |
Beta Was this translation helpful? Give feedback.
0 replies
-
you can see some (very very old) code to do fold validation in https://github.com/rwightman/pytorch-planet-amazon validation.py script (which could be hacked into timm validation with a similar Dataset class...) |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Thank you for you amazing work。
I want to use K-fold cross validation for image classification,but I don't know how to use this way in timm. I can't find this in the train.py.
If there is API for K-fold cross validation, such as 5-flod or 10-fold? The train might be like 'python train.py dataset/dog_cat --model resnet50 --fold_cross_validation 5' 'python train.py dataset/dog_cat --model resnet50 --fold_cross_validation 10'?
The train dataset include dogs and cats, like this:
dog_cat
├── dog
│ ├── dog_1
│ ├── dog_2
└── cat
├── cat_1
├── cat_2
Looking forward to your reply! Thanks in advance!
Beta Was this translation helpful? Give feedback.
All reactions