How to edit a tensorflow dataset loaded via tfds.load()?

General Discussion
#1

So I imported cifar10 dataset via tfds.load(). However cifar comes with ‘id’. I want to remove this id because jax throws unfiltered trace error because of this “id”. python - Why JAX throws an unfiltered stack trace? - Stack Overflow
Thus I want to remove it.

Here someone tries to edit the dataset by converting in into a panda’s dataframe. Is there a better way?

#3

You can apply .map() to the dataset. You need to write a function that will transform one sample of the dataset and pass this function name to .map(). In this way you can apply any transformation to the input data. See explanations and examples here: tf.data.Dataset  |  TensorFlow v2.10.0