keras-hub-nightly 0.16.1.dev202410170342__tar.gz → 0.16.1.dev202410190340__tar.gz

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: keras-hub-nightly
-Version: 0.16.1.dev202410170342
+Version: 0.16.1.dev202410190340
 Summary: Industry-strength Natural Language Processing extensions for Keras.
 Home-page: https://github.com/keras-team/keras-hub
 Author: Keras team

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/api/layers/__init__.py

@@ -40,9 +40,7 @@ from keras_hub.src.models.deeplab_v3.deeplab_v3_image_converter import (
 from keras_hub.src.models.densenet.densenet_image_converter import (
     DenseNetImageConverter,
 )
-from keras_hub.src.models.mix_transformer.mix_transformer_image_converter import (
-    MiTImageConverter,
-)
+from keras_hub.src.models.mit.mit_image_converter import MiTImageConverter
 from keras_hub.src.models.pali_gemma.pali_gemma_image_converter import (
     PaliGemmaImageConverter,
 )
@@ -52,7 +50,7 @@ from keras_hub.src.models.resnet.resnet_image_converter import (
 from keras_hub.src.models.sam.sam_image_converter import SAMImageConverter
 from keras_hub.src.models.sam.sam_mask_decoder import SAMMaskDecoder
 from keras_hub.src.models.sam.sam_prompt_encoder import SAMPromptEncoder
-from keras_hub.src.models.vgg.vgg_image_classifier import VGGImageConverter
+from keras_hub.src.models.vgg.vgg_image_converter import VGGImageConverter
 from keras_hub.src.models.whisper.whisper_audio_converter import (
     WhisperAudioConverter,
 )

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/api/models/__init__.py

@@ -202,13 +202,9 @@ from keras_hub.src.models.mistral.mistral_causal_lm_preprocessor import (
     MistralCausalLMPreprocessor,
 )
 from keras_hub.src.models.mistral.mistral_tokenizer import MistralTokenizer
-from keras_hub.src.models.mix_transformer.mix_transformer_backbone import (
-    MiTBackbone,
-)
-from keras_hub.src.models.mix_transformer.mix_transformer_classifier import (
-    MiTImageClassifier,
-)
-from keras_hub.src.models.mix_transformer.mix_transformer_classifier_preprocessor import (
+from keras_hub.src.models.mit.mit_backbone import MiTBackbone
+from keras_hub.src.models.mit.mit_image_classifier import MiTImageClassifier
+from keras_hub.src.models.mit.mit_image_classifier_preprocessor import (
     MiTImageClassifierPreprocessor,
 )
 from keras_hub.src.models.mobilenet.mobilenet_backbone import MobileNetBackbone
@@ -299,7 +295,7 @@ from keras_hub.src.models.text_classifier_preprocessor import (
 from keras_hub.src.models.text_to_image import TextToImage
 from keras_hub.src.models.vgg.vgg_backbone import VGGBackbone
 from keras_hub.src.models.vgg.vgg_image_classifier import VGGImageClassifier
-from keras_hub.src.models.vgg.vgg_image_classifier import (
+from keras_hub.src.models.vgg.vgg_image_classifier_preprocessor import (
     VGGImageClassifierPreprocessor,
 )
 from keras_hub.src.models.vit_det.vit_det_backbone import ViTDetBackbone

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/deeplab_v3/deeplab_v3_presets.py

@@ -1,7 +1,7 @@
 """DeepLabV3 preset configurations."""
 
 backbone_presets = {
-    "deeplabv3_plus_resnet50_pascalvoc": {
+    "deeplab_v3_plus_resnet50_pascalvoc": {
         "metadata": {
             "description": (
                 "DeepLabV3+ model with ResNet50 as image encoder and trained on "
@@ -10,9 +10,9 @@ backbone_presets = {
             ),
             "params": 39190656,
             "official_name": "DeepLabV3",
-            "path": "deeplabv3",
+            "path": "deeplab_v3",
             "model_card": "https://arxiv.org/abs/1802.02611",
         },
-        "kaggle_handle": "kaggle://keras/deeplabv3/keras/deeplab_v3_plus_resnet50_pascalvoc/3",
+        "kaggle_handle": "kaggle://keras/deeplabv3plus/keras/deeplab_v3_plus_resnet50_pascalvoc/3",
     },
 }

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/densenet/densenet_presets.py

@@ -12,7 +12,7 @@ backbone_presets = {
             "path": "densenet",
             "model_card": "https://arxiv.org/abs/1608.06993",
         },
-        "kaggle_handle": "kaggle://keras/densenet/keras/densenet_121_imagenet/1",
+        "kaggle_handle": "kaggle://keras/densenet/keras/densenet_121_imagenet/2",
     },
     "densenet_169_imagenet": {
         "metadata": {
@@ -25,7 +25,7 @@ backbone_presets = {
             "path": "densenet",
             "model_card": "https://arxiv.org/abs/1608.06993",
         },
-        "kaggle_handle": "kaggle://keras/densenet/keras/densenet_169_imagenet/1",
+        "kaggle_handle": "kaggle://keras/densenet/keras/densenet_169_imagenet/2",
     },
     "densenet_201_imagenet": {
         "metadata": {
@@ -38,6 +38,6 @@ backbone_presets = {
             "path": "densenet",
             "model_card": "https://arxiv.org/abs/1608.06993",
         },
-        "kaggle_handle": "kaggle://keras/densenet/keras/densenet_201_imagenet/1",
+        "kaggle_handle": "kaggle://keras/densenet/keras/densenet_201_imagenet/2",
     },
 }

keras_hub_nightly-0.16.1.dev202410190340/keras_hub/src/models/mit/__init__.py

@@ -0,0 +1,6 @@
+from keras_hub.src.models.mit.mit_backbone import MiTBackbone
+from keras_hub.src.models.mit.mit_image_classifier import MiTImageClassifier
+from keras_hub.src.models.mit.mit_presets import backbone_presets
+from keras_hub.src.utils.preset_utils import register_presets
+
+register_presets(backbone_presets, MiTBackbone)

keras_hub_nightly-0.16.1.dev202410170342/keras_hub/src/models/mix_transformer/mix_transformer_backbone.py → keras_hub_nightly-0.16.1.dev202410190340/keras_hub/src/models/mit/mit_backbone.py

@@ -4,12 +4,8 @@ from keras import ops
 
 from keras_hub.src.api_export import keras_hub_export
 from keras_hub.src.models.feature_pyramid_backbone import FeaturePyramidBackbone
-from keras_hub.src.models.mix_transformer.mix_transformer_layers import (
-    HierarchicalTransformerEncoder,
-)
-from keras_hub.src.models.mix_transformer.mix_transformer_layers import (
-    OverlappingPatchingAndEmbedding,
-)
+from keras_hub.src.models.mit.mit_layers import HierarchicalTransformerEncoder
+from keras_hub.src.models.mit.mit_layers import OverlappingPatchingAndEmbedding
 
 
 @keras_hub_export("keras_hub.models.MiTBackbone")
@@ -61,7 +57,7 @@ class MiTBackbone(FeaturePyramidBackbone):
         ```python
         images = np.ones(shape=(1, 96, 96, 3))
         labels = np.zeros(shape=(1, 96, 96, 1))
-        backbone = keras_hub.models.MiTBackbone.from_preset("mit_b0_imagenet")
+        backbone = keras_hub.models.MiTBackbone.from_preset("mit_b0_ade20k_512")
 
         # Evaluate model
         model(images)

keras_hub_nightly-0.16.1.dev202410170342/keras_hub/src/models/mix_transformer/mix_transformer_classifier.py → keras_hub_nightly-0.16.1.dev202410190340/keras_hub/src/models/mit/mit_image_classifier.py

@@ -1,9 +1,7 @@
 from keras_hub.src.api_export import keras_hub_export
 from keras_hub.src.models.image_classifier import ImageClassifier
-from keras_hub.src.models.mix_transformer.mix_transformer_backbone import (
-    MiTBackbone,
-)
-from keras_hub.src.models.mix_transformer.mix_transformer_classifier_preprocessor import (
+from keras_hub.src.models.mit.mit_backbone import MiTBackbone
+from keras_hub.src.models.mit.mit_image_classifier_preprocessor import (
     MiTImageClassifierPreprocessor,
 )
 

keras_hub_nightly-0.16.1.dev202410170342/keras_hub/src/models/mix_transformer/mix_transformer_classifier_preprocessor.py → keras_hub_nightly-0.16.1.dev202410190340/keras_hub/src/models/mit/mit_image_classifier_preprocessor.py

@@ -2,12 +2,8 @@ from keras_hub.src.api_export import keras_hub_export
 from keras_hub.src.models.image_classifier_preprocessor import (
     ImageClassifierPreprocessor,
 )
-from keras_hub.src.models.mix_transformer.mix_transformer_backbone import (
-    MiTBackbone,
-)
-from keras_hub.src.models.mix_transformer.mix_transformer_image_converter import (
-    MiTImageConverter,
-)
+from keras_hub.src.models.mit.mit_backbone import MiTBackbone
+from keras_hub.src.models.mit.mit_image_converter import MiTImageConverter
 
 
 @keras_hub_export("keras_hub.models.MiTImageClassifierPreprocessor")

keras_hub_nightly-0.16.1.dev202410170342/keras_hub/src/models/mix_transformer/mix_transformer_image_converter.py → keras_hub_nightly-0.16.1.dev202410190340/keras_hub/src/models/mit/mit_image_converter.py

@@ -1,6 +1,6 @@
 from keras_hub.src.api_export import keras_hub_export
 from keras_hub.src.layers.preprocessing.image_converter import ImageConverter
-from keras_hub.src.models.mix_transformer import MiTBackbone
+from keras_hub.src.models.mit import MiTBackbone
 
 
 @keras_hub_export("keras_hub.layers.MiTImageConverter")

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/mobilenet/mobilenet_backbone.py

@@ -96,7 +96,7 @@ class MobileNetBackbone(Backbone):
         stackwise_activation,
         output_num_filters,
         inverted_res_block,
-        image_shape=(224, 224, 3),
+        image_shape=(None, None, 3),
         input_activation="hard_swish",
         output_activation="hard_swish",
         depth_multiplier=1.0,

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/pali_gemma/pali_gemma_backbone.py

@@ -61,8 +61,6 @@ class PaliGemmaBackbone(Backbone):
         vit_classifier_activation: activation function. The activation that
             is used for final output classification in the vision transformer.
         vit_name: string. The name used for vision transformer layers.
-        include_rescaling: bool. If true, the image input will be rescaled from
-            the range `[0, 255]`, to the range `[0, 1]`.
         layer_norm_epsilon: float. The epsilon value user for every layer norm
             in all transformer blocks.
         dropout: float. Dropout probability for the Transformer decoder blocks.
@@ -121,7 +119,6 @@ class PaliGemmaBackbone(Backbone):
         vit_pooling=None,
         vit_classifier_activation=None,
         vit_name=None,
-        include_rescaling=True,
         layer_norm_epsilon=1e-6,
         dropout=0,
         dtype=None,
@@ -145,7 +142,6 @@ class PaliGemmaBackbone(Backbone):
         vit_intermediate_dim = vit_intermediate_dim or 4304
         self.vit_encoder = PaliGemmaVit(
             image_size=image_size,
-            include_rescaling=include_rescaling,
             patch_size=vit_patch_size,
             num_heads=vit_num_heads,
             hidden_dim=vit_hidden_dim,
@@ -215,7 +211,6 @@ class PaliGemmaBackbone(Backbone):
         # === Config ===
         self.vocabulary_size = vocabulary_size
         self.image_size = image_size
-        self.include_rescaling = include_rescaling
         self.num_layers = num_layers
         self.num_query_heads = num_query_heads
         self.num_key_value_heads = num_key_value_heads
@@ -242,7 +237,6 @@ class PaliGemmaBackbone(Backbone):
             {
                 "vocabulary_size": self.vocabulary_size,
                 "image_size": self.image_size,
-                "include_rescaling": self.include_rescaling,
                 "num_layers": self.num_layers,
                 "num_query_heads": self.num_query_heads,
                 "num_key_value_heads": self.num_key_value_heads,

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/pali_gemma/pali_gemma_presets.py

@@ -12,7 +12,7 @@ backbone_presets = {
             "path": "pali_gemma",
             "model_card": "https://www.kaggle.com/models/google/paligemma",
         },
-        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_mix_224/2",
+        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_mix_224/3",
     },
     "pali_gemma_3b_mix_448": {
         "metadata": {
@@ -24,7 +24,7 @@ backbone_presets = {
             "path": "pali_gemma",
             "model_card": "https://www.kaggle.com/models/google/paligemma",
         },
-        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_mix_448/2",
+        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_mix_448/3",
     },
     "pali_gemma_3b_224": {
         "metadata": {
@@ -36,7 +36,7 @@ backbone_presets = {
             "path": "pali_gemma",
             "model_card": "https://www.kaggle.com/models/google/paligemma",
         },
-        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_224/2",
+        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_224/3",
     },
     "pali_gemma_3b_448": {
         "metadata": {
@@ -48,7 +48,7 @@ backbone_presets = {
             "path": "pali_gemma",
             "model_card": "https://www.kaggle.com/models/google/paligemma",
         },
-        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_448/2",
+        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_448/3",
     },
     "pali_gemma_3b_896": {
         "metadata": {
@@ -60,6 +60,6 @@ backbone_presets = {
             "path": "pali_gemma",
             "model_card": "https://www.kaggle.com/models/google/paligemma",
         },
-        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_896/2",
+        "kaggle_handle": "kaggle://keras/paligemma/keras/pali_gemma_3b_896/3",
     },
 }

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/pali_gemma/pali_gemma_vit.py

@@ -410,8 +410,6 @@ class PaliGemmaVit(keras.Model):
     Args:
         image_size: int. The height/width of the image. Both height and width is
             expected to be the same.
-        include_rescaling: bool. If true, the image input will be rescaled from
-            the range `[0, 255]`, to the range `[0, 1]`.
         patch_size: int. The size of each square patch in the input image.
         num_heads: int. The number of attention heads for the vision(image)
             transformer encoder.
@@ -452,7 +450,6 @@ class PaliGemmaVit(keras.Model):
         num_layers,
         intermediate_dim,
         num_classes,
-        include_rescaling=True,
         pooling=None,
         classifier_activation=None,
         dtype=None,
@@ -463,14 +460,6 @@ class PaliGemmaVit(keras.Model):
             shape=(image_size, image_size, 3), name="images"
         )
         x = image_input  # Intermediate result.
-        # TODO we have moved this rescaling to preprocessing layers for most
-        # models. We should consider removing it here, though it would break
-        # compatibility.
-        if include_rescaling:
-            rescaling = keras.layers.Rescaling(
-                scale=1.0 / 127.5, offset=-1.0, name="rescaling"
-            )
-            x = rescaling(image_input)
         x = PaliGemmaVitEncoder(
             hidden_dim=hidden_dim,
             num_layers=num_layers,
@@ -520,7 +509,6 @@ class PaliGemmaVit(keras.Model):
         self.pooling = pooling
         self.num_classes = num_classes
         self.image_size = image_size
-        self.include_rescaling = include_rescaling
         self.patch_size = patch_size
         self.classifier_activation = keras.activations.get(
             classifier_activation
@@ -549,7 +537,6 @@ class PaliGemmaVit(keras.Model):
                     self.classifier_activation
                 ),
                 "image_size": self.image_size,
-                "include_rescaling": self.include_rescaling,
                 "patch_size": self.patch_size,
             }
         )

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/preprocessor.py

@@ -71,6 +71,22 @@ class Preprocessor(PreprocessingLayer):
     def image_converter(self, value):
         self._image_converter = value
 
+    @property
+    def image_size(self):
+        """Shortcut to get/set the image size of the image converter."""
+        if self.image_converter is None:
+            return None
+        return self.image_converter.image_size
+
+    @image_size.setter
+    def image_size(self, value):
+        if self.image_converter is None:
+            raise ValueError(
+                "Cannot set `image_size` on preprocessor if `image_converter` "
+                " is `None`."
+            )
+        self.image_converter.image_size = value
+
     def get_config(self):
         config = super().get_config()
         if self.tokenizer:

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/resnet/resnet_backbone.py

@@ -68,7 +68,7 @@ class ResNetBackbone(FeaturePyramidBackbone):
     input_data = np.random.uniform(0, 1, size=(2, 224, 224, 3))
 
     # Pretrained ResNet backbone.
-    model = keras_hub.models.ResNetBackbone.from_preset("resnet50")
+    model = keras_hub.models.ResNetBackbone.from_preset("resnet_50_imagenet")
     model(input_data)
 
     # Randomly initialized ResNetV2 backbone with a custom config.

keras_hub_nightly-0.16.1.dev202410190340/keras_hub/src/models/resnet/resnet_presets.py

@@ -0,0 +1,223 @@
+"""ResNet preset configurations."""
+
+backbone_presets = {
+    "resnet_18_imagenet": {
+        "metadata": {
+            "description": (
+                "18-layer ResNet model pre-trained on the ImageNet 1k dataset "
+                "at a 224x224 resolution."
+            ),
+            "params": 11186112,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/2110.00476",
+        },
+        "kaggle_handle": "kaggle://keras/resnetv1/keras/resnet_18_imagenet/2",
+    },
+    "resnet_50_imagenet": {
+        "metadata": {
+            "description": (
+                "50-layer ResNet model pre-trained on the ImageNet 1k dataset "
+                "at a 224x224 resolution."
+            ),
+            "params": 23561152,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/2110.00476",
+        },
+        "kaggle_handle": "kaggle://keras/resnetv1/keras/resnet_50_imagenet/2",
+    },
+    "resnet_101_imagenet": {
+        "metadata": {
+            "description": (
+                "101-layer ResNet model pre-trained on the ImageNet 1k dataset "
+                "at a 224x224 resolution."
+            ),
+            "params": 42605504,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/2110.00476",
+        },
+        "kaggle_handle": "kaggle://keras/resnetv1/keras/resnet_101_imagenet/2",
+    },
+    "resnet_152_imagenet": {
+        "metadata": {
+            "description": (
+                "152-layer ResNet model pre-trained on the ImageNet 1k dataset "
+                "at a 224x224 resolution."
+            ),
+            "params": 58295232,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/2110.00476",
+        },
+        "kaggle_handle": "kaggle://keras/resnetv1/keras/resnet_152_imagenet/2",
+    },
+    "resnet_v2_50_imagenet": {
+        "metadata": {
+            "description": (
+                "50-layer ResNetV2 model pre-trained on the ImageNet 1k "
+                "dataset at a 224x224 resolution."
+            ),
+            "params": 23561152,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/2110.00476",
+        },
+        "kaggle_handle": "kaggle://keras/resnetv2/keras/resnet_v2_50_imagenet/2",
+    },
+    "resnet_v2_101_imagenet": {
+        "metadata": {
+            "description": (
+                "101-layer ResNetV2 model pre-trained on the ImageNet 1k "
+                "dataset at a 224x224 resolution."
+            ),
+            "params": 42605504,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/2110.00476",
+        },
+        "kaggle_handle": "kaggle://keras/resnetv2/keras/resnet_v2_101_imagenet/2",
+    },
+    "resnet_vd_18_imagenet": {
+        "metadata": {
+            "description": (
+                "18-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution."
+            ),
+            "params": 11722824,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_18_imagenet",
+    },
+    "resnet_vd_34_imagenet": {
+        "metadata": {
+            "description": (
+                "34-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution."
+            ),
+            "params": 21838408,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_34_imagenet",
+    },
+    "resnet_vd_50_imagenet": {
+        "metadata": {
+            "description": (
+                "50-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution."
+            ),
+            "params": 25629512,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_50_imagenet",
+    },
+    "resnet_vd_50_ssld_imagenet": {
+        "metadata": {
+            "description": (
+                "50-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution with knowledge distillation."
+            ),
+            "params": 25629512,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_50_ssld_imagenet",
+    },
+    "resnet_vd_50_ssld_v2_imagenet": {
+        "metadata": {
+            "description": (
+                "50-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution with knowledge distillation and AutoAugment."
+            ),
+            "params": 25629512,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_50_ssld_v2_imagenet",
+    },
+    "resnet_vd_50_ssld_v2_fix_imagenet": {
+        "metadata": {
+            "description": (
+                "50-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution with knowledge distillation, AutoAugment and "
+                "additional fine-tuning of the classification head."
+            ),
+            "params": 25629512,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_50_ssld_v2_fix_imagenet",
+    },
+    "resnet_vd_101_imagenet": {
+        "metadata": {
+            "description": (
+                "101-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution."
+            ),
+            "params": 44673864,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_101_imagenet",
+    },
+    "resnet_vd_101_ssld_imagenet": {
+        "metadata": {
+            "description": (
+                "101-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution with knowledge distillation."
+            ),
+            "params": 44673864,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_101_ssld_imagenet",
+    },
+    "resnet_vd_152_imagenet": {
+        "metadata": {
+            "description": (
+                "152-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution."
+            ),
+            "params": 60363592,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_152_imagenet",
+    },
+    "resnet_vd_200_imagenet": {
+        "metadata": {
+            "description": (
+                "200-layer ResNetVD (ResNet with bag of tricks) model "
+                "pre-trained on the ImageNet 1k dataset at a 224x224 "
+                "resolution."
+            ),
+            "params": 74933064,
+            "official_name": "ResNet",
+            "path": "resnet",
+            "model_card": "https://arxiv.org/abs/1812.01187",
+        },
+        "kaggle_handle": "kaggle://kerashub/resnetvd/keras/resnet_vd_200_imagenet",
+    },
+}

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/sam/sam_image_segmenter.py

@@ -31,7 +31,7 @@ class SAMImageSegmenter(ImageSegmenter):
 
 
     Args:
-      backbone: A `keras_hub.models.VGGBackbone` instance.
+      backbone: A `keras_hub.models.SAMBackbone` instance.
 
     Example:
     Load pretrained model using `from_preset`.

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/sam/sam_presets.py

@@ -9,7 +9,7 @@ backbone_presets = {
             "path": "sam",
             "model_card": "https://arxiv.org/abs/2304.02643",
         },
-        "kaggle_handle": "kaggle://keras/sam/keras/sam_base_sa1b/2",
+        "kaggle_handle": "kaggle://keras/sam/keras/sam_base_sa1b/4",
     },
     "sam_large_sa1b": {
         "metadata": {
@@ -19,7 +19,7 @@ backbone_presets = {
             "path": "sam",
             "model_card": "https://arxiv.org/abs/2304.02643",
         },
-        "kaggle_handle": "kaggle://keras/sam/keras/sam_large_sa1b/3",
+        "kaggle_handle": "kaggle://keras/sam/keras/sam_large_sa1b/4",
     },
     "sam_huge_sa1b": {
         "metadata": {
@@ -29,6 +29,6 @@ backbone_presets = {
             "path": "sam",
             "model_card": "https://arxiv.org/abs/2304.02643",
         },
-        "kaggle_handle": "kaggle://keras/sam/keras/sam_huge_sa1b/3",
+        "kaggle_handle": "kaggle://keras/sam/keras/sam_huge_sa1b/4",
     },
 }

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/stable_diffusion_3/stable_diffusion_3_presets.py

@@ -10,7 +10,7 @@ backbone_presets = {
             ),
             "params": 2987080931,
             "official_name": "StableDiffusion3",
-            "path": "stablediffusion3",
+            "path": "stable_diffusion_3",
             "model_card": "https://arxiv.org/abs/2110.00476",
         },
         "kaggle_handle": "kaggle://keras/stablediffusion3/keras/stable_diffusion_3_medium/1",

{keras_hub_nightly-0.16.1.dev202410170342 → keras_hub_nightly-0.16.1.dev202410190340}/keras_hub/src/models/task.py

@@ -280,7 +280,7 @@ class Task(PipelineModel):
 
         def highlight_number(x):
             if x is None:
-                f"[color(45)]{x}[/]"
+                return f"[color(45)]{x}[/]"
             return f"[color(34)]{x:,}[/]"  # Format number with commas.
 
         def highlight_symbol(x):
@@ -339,7 +339,10 @@ class Task(PipelineModel):
                         add_layer(layer, info)
                     elif isinstance(layer, ImageConverter):
                         info = "Image size: "
-                        info += highlight_shape(layer.image_size)
+                        image_size = layer.image_size
+                        if image_size is None:
+                            image_size = (None, None)
+                        info += highlight_shape(image_size)
                         add_layer(layer, info)
                     elif isinstance(layer, AudioConverter):
                         info = "Audio shape: "