register mobilenet presets

keras_hub/src/models/mobilenetv5/__init__.py

@@ -0,0 +1,9 @@
+from keras_hub.src.models.mobilenetv5.mobilenetv5_backbone import (
+    MobileNetV5Backbone,
+)
+from keras_hub.src.models.mobilenetv5.mobilenetv5_presets import (
+    backbone_presets,
+)
+from keras_hub.src.utils.preset_utils import register_presets
+
+register_presets(backbone_presets, MobileNetV5Backbone)

keras_hub/src/models/mobilenetv5/mobilenetv5_backbone_test.py

@@ -47,11 +47,10 @@ def test_backbone_basics(self):
         )
 
     @pytest.mark.large
-    @pytest.mark.skip(reason="TODO: Enable once presets have been uploaded.")
     def test_smallest_preset(self):
         self.run_preset_test(
             cls=MobileNetV5Backbone,
-            preset="mobilenetv5_300m_enc.gemma3n",
+            preset="mobilenetv5_300m_enc_gemma3n",
             input_data=keras.ops.ones((1, 224, 224, 3)),
             expected_output_shape=(1, 16, 16, 2048),
         )

keras_hub/src/models/mobilenetv5/mobilenetv5_presets.py

@@ -0,0 +1,15 @@
+"""MobileNetV5 preset configurations."""
+
+backbone_presets = {
+    "mobilenetv5_300m_enc_gemma3n": {
+        "metadata": {
+            "description": (
+                "Lightweight 300M-parameter convolutional vision encoder used "
+                "as the image backbone for Gemma 3n"
+            ),
+            "params": 294_284_096,
+            "path": "mobilenetv5",
+        },
+        "kaggle_handle": "kaggle://keras/mobilenetv5/keras/mobilenetv5_300m_enc_gemma3n/1",
+    }
+}

tools/checkpoint_conversion/convert_mobilenetv5_checkpoints.py

@@ -37,7 +37,7 @@
 from keras_hub.src.models.mobilenetv5.mobilenetv5_layers import RmsNorm2d
 
 PRESET_MAP = {
-    "mobilenetv5_300m_enc.gemma3n": {
+    "mobilenetv5_300m_enc_gemma3n": {
         "arch": "mobilenetv5_300m_enc",
         "hf_hub_id": "timm/mobilenetv5_300m.gemma3n",
     }