Will Rice
Will Rice
This issue is pretty old, but since it is still open does that mean I could work on an implementation? I work on ASR and would benefit from an implementation...
@iamjanvijay sounds good!
@iamjanvijay do you want me to just port your ops to the addons loss format and tag you in the PR?
> @will-rice Yes. That can be a good way to start then we can do iterations over it. What do you suggest? that works. I'll put it together and then...
What were you thinking about the "TensorFlow run-time for training models"? Like giving the user the option of using TensorFlow or PyTorch? I wouldn't mind taking a stab at the...
Something I want to add is that I'm not partial to `ada_norm` here. I just want to use DiT without needing class labels or the additional arguments for `PixArtAlphaCombinedTimestepSizeEmbeddings`.
@mthrok I came across [this](https://github.com/pytorch/audio/blob/2994ce2eb0dd3539683f8f30d23162bd138f88ee/test/torchaudio_unittest/functional/librosa_compatibility_test_impl.py#L99) while working on the tests and realized that the `power_to_db` functionality is already available. If it would be useful to separate the functions I could...
If you extract what you downloaded from the "Data" button, you can use the manifest to build text/speech pairs based on the `label` and `name` keys under the `training_data` key....
I think we are going to try and put this into tensorflow-addons which will include ensuring it stays up to date. https://github.com/tensorflow/addons/issues/641
Does that mean it is planned, or do I need to fork?