Michael Stepanovsky comments

Results 4 comments of


                                            Michael Stepanovsky

Error "Attempt to call single feature writer on packed feature writer" with embedding features on GPU

Having the same problem. Asked question on stackoverflow yesterday: https://stackoverflow.com/questions/77987029/getting-catboosterror-attempt-to-call-single-feature-writer-on-packed-feature Version: 1.2.2 GPU: A100/V100 Unfortunately, I don't have access to the older versions of catboost now, but I definitely remember...

Do we need lr scheduler?

It is true that there's single place where .step() is being called, иut the method .scheduler() in which .step() is called has never been used for any of the training...

In training Stage1 after 49th epoch getting RuntimeError: you can only change requires_grad flags of leaf variables, g_loss.requires_grad = True

I have the same issue, but for this code snippet: `d_loss = self._dl(wav.detach().unsqueeze(1).float(), y_rec.detach()).mean() ` Getting into details, it's the error in the forward method, in WavLMLoss class: ``` def...

In training Stage1 after 49th epoch getting RuntimeError: you can only change requires_grad flags of leaf variables, g_loss.requires_grad = True

Found the solution. you need to write `self.wavlm.eval()` in the start of forward method WavLMLoss class in losses module. Worked for me.