output shape inconsistent with comute_output_shape()

[AGFACBNNR]

Maybe I found a bug in Keras 3.0.5.
The minimal code snippet triggering the bug is as following.

from keras.layers import UpSampling2D
import numpy as np
import keras

x = np.ones((1,2,3,4))
layer = UpSampling2D(data_format='channels_first',interpolation='bilinear')

print(layer(x).shape)
print(layer.compute_output_shape(x.shape))
print(keras.__version__)

The output is:

(1, 2, 8, 4)
(1, 2, 6, 8)
3.0.5

I believe they should be the same shape. Could you please have a look?

[google-ml-butler]

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