Thank you for your excellent work!
I have a question about reproducing the results shown in Figure 5. Specifically, regarding the "random" version: is the random transformation applied independently to each data sample during training, or is it applied consistently across the entire minibatch?
Currently, I apply random rotations within the __getitem__() function of the QM9 data loader. However, I haven't been able to reproduce the results reported in Figure 5.
Any guidance would be greatly appreciated!
Thank you for your excellent work!
I have a question about reproducing the results shown in Figure 5. Specifically, regarding the "random" version: is the random transformation applied independently to each data sample during training, or is it applied consistently across the entire minibatch?
Currently, I apply random rotations within the
__getitem__()function of the QM9 data loader. However, I haven't been able to reproduce the results reported in Figure 5.Any guidance would be greatly appreciated!