Closed wbenoit26 closed 4 months ago
I'll point you to the frequency domain dataset. This is where q transform / spectrogram augmentations would go. Also, the [frequency domain arch] (https://github.com/ML4GW/aframev2/blob/main/projects/train/train/architectures/supervised.py#L91) is a wrapper around ResNet2d that plays nicely with lightning.
Changing the model arch class path and dataset class path in the train config should get you started.
I'll point you to the frequency domain dataset. This is where q transform / spectrogram augmentations would go. Also, the [frequency domain arch] (https://github.com/ML4GW/aframev2/blob/main/projects/train/train/architectures/supervised.py#L91) is a wrapper around ResNet2d that plays nicely with lightning.
Changing the model arch class path and dataset class path in the train config should get you started.