sushreebarsa

@javidcf tf.ragged.constant itself doesn't directly detect dense dimensions. Using the` inner_shape` argument approach could be helpful for defining the expected shape of inner values. ``` import tensorflow as tf #...

@sachinprasadhs I was able to replicate the issue reported [here](https://colab.research.google.com/gist/sushreebarsa/0782f8b41efbbf02e9b2a45c2b764fa5/61955.ipynb). Thank you!

@Praveer1981 If you saved the model please confirm that weights were included using model.save_weights() or a similar method. Please save the model with the same TensorFlow/Keras version you're using in...

@Praveer1981 Sorry for the late response! Please ensure the TensorFlow versions used for saving and loading the model are compatible. Mismatched versions can lead to loading errors. Could you please...

@MintyMH One workaround could be to rewrite the Gram-Schmidt function using vectorized operations and functional programming constructs like jax.map and jax.lax.cond. This avoids control flow and ensures compatibility with grad....

@singersbalm To use the existing graph saved with the older TensorFlow version in your current session with a newer version. However, some functionality might be limited or compatibility issues might...

@singersbalm In order to expedite the trouble-shooting process, please provide the code snippet to reproduce the issue reported here. Thank you!

@skwyddie ensorFlow 2.15 is not natively compatible with M1 so you need to switch to a version built for Apple Silicon. TensorFlow 2.10.0-rc0 is the latest release with official M1...

@skwyddie You're right, there isn't an official TensorFlow version compiled specifically for Apple M1 yet. However, there are workarounds and alternative options you can explore. Here's the guide to install...

@sachinprasadhs I was able to replicate the issue on colab. Please find the [gist](https://colab.research.google.com/gist/sushreebarsa/af609c8a065feeb7c2d4491ea0fac09c/62854.ipynb#scrollTo=hnC0ZaIEKabZ) here for reference. Thank you!