Whisper with Spec_Augment

[arvindmn01]

Hi
I am finetuning whisper model. I want to add augmentation layer. is it possible to add Spec_augment layer to the whisper model?. I am using this code to add Spec_augment layer.

import spec_augment
class SpecAugmentModule(torch.nn.Module):
     def __init__(self, freq_mask_param, time_mask_param, n_freq_mask, n_time_mask, mask_value):
         super().__init__()
         self.spec_augment = spec_augment.SpecAugment(
             freq_mask_param=freq_mask_param,
             time_mask_param=time_mask_param,
             n_freq_mask=n_freq_mask,
             n_time_mask=n_time_mask,
             mask_value=mask_value)

     def forward(self, spectrogram):
         return self.spec_augment(spectrogram)



spec_augment_module = SpecAugmentModule(
     freq_mask_param=6,
     time_mask_param=12,
     n_freq_mask=6,
     n_time_mask=4,
     mask_value=-100)
model.add_module('spec_augment', spec_augment_module)

but I am getting this error when i start training.
RuntimeError: GET was unable to find an engine to execute this computation

[itaipee]

hi, did you find answer ? can you update ?