Would you consider using KerasHub API's for this model?
#3
by Divyasreepat - opened
Sounds like a game changer for speech recognition!
Would you consider using KerasHub API's for this model?
- you could take advantage of the
save_to_preset()andfrom_preset()methods to load model presets from HuggingFace or Kaggle - you can easily upload to HuggingFace or Kaggle using
upload_preset()
Example of a KerasHub model on HF here - https://huggingface.co/google/paligemma-3b-pt-224-keras
Yes, absolutely. The inference code is already written in Keras 3. I haven't investigated keras-hub in detail, but will do so soon.
Great! Here are all of KerasHub models - https://huggingface.co/keras
Here is the GitHub repo for reference model implementations - https://github.com/keras-team/keras-hub/tree/master
This has been added to KerasHub - https://www.kaggle.com/models/keras/moonshine, https://github.com/keras-team/keras-hub/tree/master/keras_hub/src/models/moonshine