Fix shape inference bug for conv2dtranspose

keras_core/backend/common/backend_utils.py

@@ -187,6 +187,9 @@ def compute_conv_transpose_padding_args_for_torch(
 def _get_output_shape_given_tf_padding(
     input_size, kernel_size, strides, padding, output_padding, dilation_rate
 ):
+    if input_size is None:
+        return None
+
     assert padding.lower() in {"valid", "same"}
 
     kernel_size = (kernel_size - 1) * dilation_rate + 1

keras_core/layers/convolutional/conv_transpose_test.py

@@ -835,3 +835,23 @@ def test_conv1d_transpose_consistency(
         # Compare results
         kc_res = kc_layer(input)
         self.assertAllClose(expected_res, kc_res, atol=1e-5)
+
+    @parameterized.product(
+        kernel_size=list(range(1, 5)),
+        strides=list(range(1, 5)),
+        padding=["same", "valid"],
+        output_padding=[None] + list(range(1, 5)),
+    )
+    def test_shape_inference_static_unknown_shape(
+        self, kernel_size, strides, padding, output_padding
+    ):
+        x = layers.Input(shape=(None, None, 3))
+        x = layers.Conv2DTranspose(
+            filters=2,
+            kernel_size=kernel_size,
+            strides=strides,
+            padding=padding,
+            output_padding=output_padding,
+            dilation_rate=1,
+        )(x)
+        self.assertEqual(x.shape, (None, None, None, 2))