Are the docs correct for the RNN description of `batch_shape` ?

[newresu]

In the description the RNN it says it says:

If sequential model:
   `batch_input_shape=(...)` to the first layer in your model.

Yet seems that the parameter batch_input_shape will fail since it does not exist in most layers.

Is a user supposed to be adding an Input layer instead of that parameter, with the parameter batch_shape (note that the names don't match either.) ?

Should the description instead be the same for Functional or Sequential, or am I misunderstanding something ?

This is a follow up from this thread in case you need more context.

[sachinprasadhs]

batch_input_shape is part of **kwargs and it will be treated as batch_shape in the Input layer.
Below is the part where it is handled.

keras/keras/src/layers/core/input_layer.py

Lines 30 to 31 in 816d219

	if "batch_input_shape" in kwargs:
	batch_shape = kwargs.pop("batch_input_shape")

[newresu]

batch_input_shape is part of **kwargs and it will be treated as batch_shape in the Input layer.
Below is the part where it is handled.

keras/keras/src/layers/core/input_layer.py

Lines 30 to 31 in 816d219

	if "batch_input_shape" in kwargs:
	batch_shape = kwargs.pop("batch_input_shape")

But why not here then?

#20106 (comment)

[sachinprasadhs]

I have replied to the thread with the comment #20106 (comment) which accepts batch_input_shape for the InputLayer

[newresu]

I have replied to the thread with the comment #20106 (comment) which accepts batch_input_shape for the InputLayer

Thank you, i didnt know inputlayer and input were different.

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No