tf_dataset.map(sequential_model(x)) results in Cannot convert a symbolic tf.Tensor (args_0:0) to a numpy array.

[divyashreepathihalli]

Error with JAX and Torch backend

when dataset = tf_data.Dataset.from_tensor_slices((images, labels)) is used with map function to call a sequential model of augmentation layers results in Cannot convert a symbolic tf.Tensor (args_0:0) to a numpy array. error
repro here

[fchollet]

Yes, that's expected. Sequential cannot be used with tf.data unless you're running on the TF backend.