jackz
jackz
[efficient-vit](https://github.com/davide-coccomini/Combining-EfficientNet-and-Vision-Transformers-for-Video-Deepfake-Detection/tree/main/efficient-vit) This is the model i'm using.
I have the same problem, have you resolved it?
You can first separate a part of the dfdc training set for verification and testing.
[email protected] This is my email, you can contact me through this.