Mark Llama distribution tests with multi_device

test_distribution, test_distribution_with_lora, and
test_layout_map_mesh_shapes were not marked, so the dedicated
multi_device CI job (added in keras-team#2809) was collecting zero real tests
for this model -- only test_layout_map_live_presets carried the
marker, and that test is deliberately excluded by the job's own
kaggle_key_required/extra_large filter.

For test_layout_map_mesh_shapes, which uses absl's
parameterized.named_parameters, the marker must be applied as the
innermost decorator (closest to the function) rather than outermost:
named_parameters generates fresh method objects per expansion that
don't inherit the pytestmark attribute pytest.mark sets on the
original function, so an outer marker never reaches the expanded
test cases. Verified empirically that only the inner-decorator
ordering makes the parameterized cases collectible under
-m multi_device.

Author: pctablet505

keras_hub/src/models/llama/llama_backbone_test.py

@@ -174,6 +174,7 @@ def test_all_presets(self):
                 input_data=self.input_data,
             )
 
+    @pytest.mark.multi_device
     def test_distribution(self):
         # Note (preserved from the pre-refactor manual test): mesh is
         # pinned to exactly 2 devices (not len(devices), see the shared
@@ -203,6 +204,7 @@ def test_distribution(self):
             allow_replicated=(),
         )
 
+    @pytest.mark.multi_device
     def test_distribution_with_lora(self):
         if keras.backend.backend() != "jax":
             self.skipTest("`ModelParallel` testing requires the Jax backend.")
@@ -248,6 +250,7 @@ def test_distribution_with_lora(self):
         for dims in (LLAMA2_7B_DIMS, LLAMA2_70B_DIMS)
         for shape in CAPPED_MESH_SHAPES
     )
+    @pytest.mark.multi_device
     def test_layout_map_mesh_shapes(self, dims, mesh_shape):
         if keras.backend.backend() != "jax":
             self.skipTest("`ModelParallel` testing requires the Jax backend.")