tango

tango copied to clipboard

Hi! Thanks so much for this work! When I tried to train the model on AudioCaps (didn't change the training script other than file paths), I got this issue: File "/tango/train.py", line 553, in main() File "/tango/train.py", line 459, in main mixed_mel, _, _, mixed_captions = torch_tools.augment_wav_to_fbank(audios, text, len(audios), File "/tango/tools/torch_tools.py", line 118, in augment_wav_to_fbank waveform, captions = augment(paths, texts) File "/tango/tools/torch_tools.py", line 108, in augment waveform = torch.tensor(np.concatenate(mixed_sounds, 0)) File "<array_function internals>", line 180, in concatenate ValueError: need at least one array to concatenate

It would be highly appreciated if you could kindly help me with this problem, thanks!

It seems like the length of texts is 1. The last batch of the training data may have only one instance, and then this error would come up. Did this happen at the end of the training loop?

I have made a few small changes in the train.py to handle this. Let me know if that fixes the issue.

I am assuming you used the per device train batch size of at least 2. Augmentation won't work with batch size of 1.

It seems like the length of texts is 1. The last batch of the training data may have only one instance, and then this error would come up. Did this happen at the end of the training loop?

I have made a few small changes in the train.py to handle this. Let me know if that fixes the issue.

I am assuming you used the per device train batch size of at least 2. Augmentation won't work with batch size of 1.

Thanks for your help! The problem occurred before training started, when I tried to run the new train.py, I got the typical "cuda out of memory" error. Have you ever tried to train without data augment, how's the result?