convnext_inverse: kernel_size 4→2

This commit is contained in:
Starbeamrainbowlabs 2022-11-11 19:29:37 +00:00
parent 0662d0854b
commit 527b34942d
Signed by: sbrl
GPG key ID: 1BE5172E637709C2

View file

@ -41,7 +41,7 @@ def block_upscale(layer_in, block_number, depth, dim):
layer_next = tf.keras.layers.Conv2DTranspose(
name=f"cns.stage{block_number}.end.convtp",
filters=dim,
kernel_size=4,
kernel_size=2, # double each time, was 4
strides=2
)(layer_next)
layer_next = tf.keras.layers.LayerNormalization(name=f"cns.stage{block_number}.end.norm", epsilon=1e-6)(layer_next)