keras-hub-nightly 0.19.0.dev202503060350__py3-none-any.whl → 0.20.0.dev202503150350__py3-none-any.whl

keras_hub/src/utils/timm/convert_cspnet.py

@@ -0,0 +1,165 @@
+import numpy as np
+
+from keras_hub.src.models.cspnet.cspnet_backbone import CSPNetBackbone
+
+backbone_cls = CSPNetBackbone
+
+
+def convert_backbone_config(timm_config):
+    timm_architecture = timm_config["architecture"]
+
+    if timm_architecture == "cspdarknet53":
+        stem_filters = 32
+        stem_kernel_size = 3
+        stem_strides = 1
+        stackwise_depth = [1, 2, 8, 8, 4]
+        stackwise_num_filters = [64, 128, 256, 512, 1024]
+        bottle_ratio = (0.5,) + (1.0,)
+        block_ratio = (1.0,) + (0.5,)
+        expand_ratio = (2.0,) + (1.0,)
+        stage_type = "csp"
+        block_type = "dark_block"
+        down_growth = True
+        stackwise_strides = 2
+    else:
+        raise ValueError(
+            f"Currently, the architecture {timm_architecture} is not supported."
+        )
+    return dict(
+        stem_filters=stem_filters,
+        stem_kernel_size=stem_kernel_size,
+        stem_strides=stem_strides,
+        stackwise_depth=stackwise_depth,
+        stackwise_num_filters=stackwise_num_filters,
+        bottle_ratio=bottle_ratio,
+        block_ratio=block_ratio,
+        expand_ratio=expand_ratio,
+        stage_type=stage_type,
+        block_type=block_type,
+        stackwise_strides=stackwise_strides,
+        down_growth=down_growth,
+    )
+
+
+def convert_weights(backbone, loader, timm_config):
+    def port_conv2d(hf_weight_prefix, keras_layer_name):
+        loader.port_weight(
+            backbone.get_layer(keras_layer_name).kernel,
+            hf_weight_key=f"{hf_weight_prefix}.weight",
+            hook_fn=lambda x, _: np.transpose(x, (2, 3, 1, 0)),
+        )
+
+    def port_batch_normalization(hf_weight_prefix, keras_layer_name):
+        loader.port_weight(
+            backbone.get_layer(keras_layer_name).gamma,
+            hf_weight_key=f"{hf_weight_prefix}.weight",
+        )
+        loader.port_weight(
+            backbone.get_layer(keras_layer_name).beta,
+            hf_weight_key=f"{hf_weight_prefix}.bias",
+        )
+        loader.port_weight(
+            backbone.get_layer(keras_layer_name).moving_mean,
+            hf_weight_key=f"{hf_weight_prefix}.running_mean",
+        )
+        loader.port_weight(
+            backbone.get_layer(keras_layer_name).moving_variance,
+            hf_weight_key=f"{hf_weight_prefix}.running_var",
+        )
+
+    # Stem
+
+    stem_filter = backbone.stem_filters
+    if not isinstance(stem_filter, (tuple, list)):
+        stem_filter = [stem_filter]
+
+    for i in range(len(stem_filter)):
+        port_conv2d(f"stem.conv{i + 1}.conv", f"csp_stem_conv_{i}")
+        port_batch_normalization(f"stem.conv{i + 1}.bn", f"csp_stem_bn_{i}")
+
+    # Stages
+    stackwise_depth = backbone.stackwise_depth
+    stage_type = backbone.stage_type
+    block_type = backbone.block_type
+
+    for idx, block in enumerate(stackwise_depth):
+        port_conv2d(
+            f"stages.{idx}.conv_down.conv",
+            f"stage_{idx}_{stage_type}_conv_down_1",
+        )
+        port_batch_normalization(
+            f"stages.{idx}.conv_down.bn", f"stage_{idx}_{stage_type}_bn_1"
+        )
+        port_conv2d(
+            f"stages.{idx}.conv_exp.conv", f"stage_{idx}_{stage_type}_conv_exp"
+        )
+        port_batch_normalization(
+            f"stages.{idx}.conv_exp.bn", f"stage_{idx}_{stage_type}_bn_2"
+        )
+
+        for i in range(block):
+            port_conv2d(
+                f"stages.{idx}.blocks.{i}.conv1.conv",
+                f"stage_{idx}_block_{i}_{block_type}_conv_1",
+            )
+            port_batch_normalization(
+                f"stages.{idx}.blocks.{i}.conv1.bn",
+                f"stage_{idx}_block_{i}_{block_type}_bn_1",
+            )
+            port_conv2d(
+                f"stages.{idx}.blocks.{i}.conv2.conv",
+                f"stage_{idx}_block_{i}_{block_type}_conv_2",
+            )
+            port_batch_normalization(
+                f"stages.{idx}.blocks.{i}.conv2.bn",
+                f"stage_{idx}_block_{i}_{block_type}_bn_2",
+            )
+            if block_type == "bottleneck_block":
+                port_conv2d(
+                    f"stages.{idx}.blocks.{i}.conv3.conv",
+                    f"stage_{idx}_block_{i}_{block_type}_conv_3",
+                )
+                port_batch_normalization(
+                    f"stages.{idx}.blocks.{i}.conv3.bn",
+                    f"stage_{idx}_block_{i}_{block_type}_bn_3",
+                )
+
+        if stage_type == "csp":
+            port_conv2d(
+                f"stages.{idx}.conv_transition_b.conv",
+                f"stage_{idx}_{stage_type}_conv_transition_b",
+            )
+            port_batch_normalization(
+                f"stages.{idx}.conv_transition_b.bn",
+                f"stage_{idx}_{stage_type}_transition_b_bn",
+            )
+            port_conv2d(
+                f"stages.{idx}.conv_transition.conv",
+                f"stage_{idx}_{stage_type}_conv_transition",
+            )
+            port_batch_normalization(
+                f"stages.{idx}.conv_transition.bn",
+                f"stage_{idx}_{stage_type}_transition_bn",
+            )
+
+        else:
+            port_conv2d(
+                f"stages.{idx}.conv_transition.conv",
+                f"stage_{idx}_{stage_type}_conv_transition",
+            )
+            port_batch_normalization(
+                f"stages.{idx}.conv_transition.bn",
+                f"stage_{idx}_{stage_type}_transition_bn",
+            )
+
+
+def convert_head(task, loader, timm_config):
+    loader.port_weight(
+        task.output_dense.kernel,
+        hf_weight_key="head.fc.weight",
+        hook_fn=lambda x, _: np.transpose(np.squeeze(x)),
+    )
+    loader.port_weight(
+        task.output_dense.bias,
+        hf_weight_key="head.fc.bias",
+    )

keras_hub/src/utils/timm/convert_mobilenet.py

@@ -8,64 +8,135 @@ backbone_cls = MobileNetBackbone
 def convert_backbone_config(timm_config):
     timm_architecture = timm_config["architecture"]
 
-    if "mobilenetv3_" in timm_architecture:
-        input_activation = "hard_swish"
-        output_activation = "hard_swish"
-    else:
-        input_activation = "relu6"
-        output_activation = "relu6"
-
-    if timm_architecture == "mobilenetv3_small_050":
-        stackwise_num_blocks = [2, 3, 2, 3]
-        stackwise_expansion = [
+    kwargs = {
+        "stackwise_num_blocks": [2, 3, 2, 3],
+        "stackwise_expansion": [
             [40, 56],
             [64, 144, 144],
             [72, 72],
             [144, 288, 288],
-        ]
-        stackwise_num_filters = [[16, 16], [24, 24, 24], [24, 24], [48, 48, 48]]
-        stackwise_kernel_size = [[3, 3], [5, 5, 5], [5, 5], [5, 5, 5]]
-        stackwise_num_strides = [[2, 1], [2, 1, 1], [1, 1], [2, 1, 1]]
-        stackwise_se_ratio = [
+        ],
+        "stackwise_num_filters": [
+            [16, 16],
+            [24, 24, 24],
+            [24, 24],
+            [48, 48, 48],
+        ],
+        "stackwise_kernel_size": [[3, 3], [5, 5, 5], [5, 5], [5, 5, 5]],
+        "stackwise_num_strides": [[2, 1], [2, 1, 1], [1, 1], [2, 1, 1]],
+        "stackwise_se_ratio": [
             [None, None],
             [0.25, 0.25, 0.25],
             [0.25, 0.25],
             [0.25, 0.25, 0.25],
-        ]
-        stackwise_activation = [
+        ],
+        "stackwise_activation": [
             ["relu", "relu"],
             ["hard_swish", "hard_swish", "hard_swish"],
             ["hard_swish", "hard_swish"],
             ["hard_swish", "hard_swish", "hard_swish"],
-        ]
-        stackwise_padding = [[1, 1], [2, 2, 2], [2, 2], [2, 2, 2]]
-        output_num_filters = 1024
-        input_num_filters = 16
-        depthwise_filters = 8
-        squeeze_and_excite = 0.5
-        last_layer_filter = 288
+        ],
+        "stackwise_padding": [[1, 1], [2, 2, 2], [2, 2], [2, 2, 2]],
+        "output_num_filters": 1024,
+        "input_num_filters": 16,
+        "depthwise_filters": 8,
+        "depthwise_stride": 2,
+        "depthwise_residual": False,
+        "squeeze_and_excite": 0.5,
+        "last_layer_filter": 288,
+        "input_activation": "relu6",
+        "output_activation": "relu6",
+    }
+
+    if "mobilenetv3_" in timm_architecture:
+        kwargs["input_activation"] = "hard_swish"
+        kwargs["output_activation"] = "hard_swish"
+
+    if timm_architecture == "mobilenetv3_small_050":
+        pass
+    elif timm_architecture == "mobilenetv3_small_100":
+        modified_kwargs = {
+            "stackwise_expansion": [
+                [72, 88],
+                [96, 240, 240],
+                [120, 144],
+                [288, 576, 576],
+            ],
+            "stackwise_num_filters": [
+                [24, 24],
+                [40, 40, 40],
+                [48, 48],
+                [96, 96, 96],
+            ],
+            "depthwise_filters": 16,
+            "last_layer_filter": 576,
+        }
+        kwargs.update(modified_kwargs)
+    elif timm_architecture.startswith("mobilenetv3_large_100"):
+        modified_kwargs = {
+            "stackwise_num_blocks": [2, 3, 4, 2, 3],
+            "stackwise_expansion": [
+                [64, 72],
+                [72, 120, 120],
+                [240, 200, 184, 184],
+                [480, 672],
+                [672, 960, 960],
+            ],
+            "stackwise_num_filters": [
+                [24, 24],
+                [40, 40, 40],
+                [80, 80, 80, 80],
+                [112, 112],
+                [160, 160, 160],
+            ],
+            "stackwise_kernel_size": [
+                [3, 3],
+                [5, 5, 5],
+                [3, 3, 3, 3],
+                [3, 3],
+                [5, 5, 5],
+            ],
+            "stackwise_num_strides": [
+                [2, 1],
+                [2, 1, 1],
+                [2, 1, 1, 1],
+                [1, 1],
+                [2, 1, 1],
+            ],
+            "stackwise_se_ratio": [
+                [None, None],
+                [0.25, 0.25, 0.25],
+                [None, None, None, None],
+                [0.25, 0.25],
+                [0.25, 0.25, 0.25],
+            ],
+            "stackwise_activation": [
+                ["relu", "relu"],
+                ["relu", "relu", "relu"],
+                ["hard_swish", "hard_swish", "hard_swish", "hard_swish"],
+                ["hard_swish", "hard_swish"],
+                ["hard_swish", "hard_swish", "hard_swish"],
+            ],
+            "stackwise_padding": [
+                [1, 1],
+                [2, 2, 2],
+                [1, 1, 1, 1],
+                [1, 1],
+                [2, 2, 2],
+            ],
+            "depthwise_filters": 16,
+            "depthwise_stride": 1,
+            "depthwise_residual": True,
+            "squeeze_and_excite": None,
+            "last_layer_filter": 960,
+        }
+        kwargs.update(modified_kwargs)
     else:
         raise ValueError(
             f"Currently, the architecture {timm_architecture} is not supported."
         )
 
-    return dict(
-        input_num_filters=input_num_filters,
-        input_activation=input_activation,
-        depthwise_filters=depthwise_filters,
-        squeeze_and_excite=squeeze_and_excite,
-        stackwise_num_blocks=stackwise_num_blocks,
-        stackwise_expansion=stackwise_expansion,
-        stackwise_num_filters=stackwise_num_filters,
-        stackwise_kernel_size=stackwise_kernel_size,
-        stackwise_num_strides=stackwise_num_strides,
-        stackwise_se_ratio=stackwise_se_ratio,
-        stackwise_activation=stackwise_activation,
-        stackwise_padding=stackwise_padding,
-        output_num_filters=output_num_filters,
-        output_activation=output_activation,
-        last_layer_filter=last_layer_filter,
-    )
+    return kwargs
 
 
 def convert_weights(backbone, loader, timm_config):
@@ -120,9 +191,14 @@ def convert_weights(backbone, loader, timm_config):
     port_conv2d(stem_block.conv1, f"{hf_name}.conv_dw")
     port_batch_normalization(stem_block.batch_normalization1, f"{hf_name}.bn1")
 
-    stem_se_block = stem_block.se_layer
-    port_conv2d(stem_se_block.conv_reduce, f"{hf_name}.se.conv_reduce", True)
-    port_conv2d(stem_se_block.conv_expand, f"{hf_name}.se.conv_expand", True)
+    if stem_block.squeeze_excite_ratio:
+        stem_se_block = stem_block.se_layer
+        port_conv2d(
+            stem_se_block.conv_reduce, f"{hf_name}.se.conv_reduce", True
+        )
+        port_conv2d(
+            stem_se_block.conv_expand, f"{hf_name}.se.conv_expand", True
+        )
 
     port_conv2d(stem_block.conv2, f"{hf_name}.conv_pw")
     port_batch_normalization(stem_block.batch_normalization2, f"{hf_name}.bn2")

keras_hub/src/utils/timm/preset_loader.py

@@ -3,6 +3,7 @@
 from keras_hub.src.models.image_classifier import ImageClassifier
 from keras_hub.src.utils.preset_utils import PresetLoader
 from keras_hub.src.utils.preset_utils import jax_memory_cleanup
+from keras_hub.src.utils.timm import convert_cspnet
 from keras_hub.src.utils.timm import convert_densenet
 from keras_hub.src.utils.timm import convert_efficientnet
 from keras_hub.src.utils.timm import convert_mobilenet
@@ -17,6 +18,8 @@ class TimmPresetLoader(PresetLoader):
         architecture = self.config["architecture"]
         if "resnet" in architecture:
             self.converter = convert_resnet
+        elif "csp" in architecture:
+            self.converter = convert_cspnet
         elif "densenet" in architecture:
             self.converter = convert_densenet
         elif "mobilenet" in architecture:
@@ -51,6 +54,12 @@ class TimmPresetLoader(PresetLoader):
             )
         # Support loading the classification head for classifier models.
         kwargs["num_classes"] = self.config["num_classes"]
+        if (
+            "num_features" in self.config
+            and "mobilenet" in self.config["architecture"]
+        ):
+            kwargs["num_features"] = self.config["num_features"]
+
         task = super().load_task(cls, load_weights, load_task_weights, **kwargs)
         if load_task_weights:
             with SafetensorLoader(self.preset, prefix="") as loader:

keras_hub/src/version_utils.py

@@ -1,7 +1,7 @@
 from keras_hub.src.api_export import keras_hub_export
 
 # Unique source of truth for the version number.
-__version__ = "0.19.0.dev202503060350"
+__version__ = "0.20.0.dev202503150350"
 
 
 @keras_hub_export("keras_hub.version")

{keras_hub_nightly-0.19.0.dev202503060350.dist-info → keras_hub_nightly-0.20.0.dev202503150350.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: keras-hub-nightly
-Version: 0.19.0.dev202503060350
+Version: 0.20.0.dev202503150350
 Summary: Industry-strength Natural Language Processing extensions for Keras.
 Home-page: https://github.com/keras-team/keras-hub
 Author: Keras team

{keras_hub_nightly-0.19.0.dev202503060350.dist-info → keras_hub_nightly-0.20.0.dev202503150350.dist-info}/RECORD

@@ -1,14 +1,14 @@
 keras_hub/__init__.py,sha256=QGdXyHgYt6cMUAP1ebxwc6oR86dE0dkMxNy2eOCQtFo,855
 keras_hub/api/__init__.py,sha256=EzR6D-XWsm_gDrX5LDwKEmrah_gu3ffpj8GKBudE0yI,485
-keras_hub/api/layers/__init__.py,sha256=l0L6ma5QTpEkzWirUeUlaq1COZF_HdEC58kctTGFhow,3533
+keras_hub/api/layers/__init__.py,sha256=-yHyqsjWBhmFv9RSS2cMyPcieU1RkNzcNsq9IDXSVFE,3626
 keras_hub/api/metrics/__init__.py,sha256=So8Ec-lOcTzn_UUMmAdzDm8RKkPu2dbRUm2px8gpUEI,381
-keras_hub/api/models/__init__.py,sha256=bOF3APgZJe2o5-Mjw4qZFD4wWwhlt5jEEZvzfLQK6i8,17523
+keras_hub/api/models/__init__.py,sha256=vtOXBt8YNXvaDrfpKRpJ6MXjU5FzMv2a44Db_P8cGUg,17606
 keras_hub/api/samplers/__init__.py,sha256=n-_SEXxr2LNUzK2FqVFN7alsrkx1P_HOVTeLZKeGCdE,730
 keras_hub/api/tokenizers/__init__.py,sha256=lhvIqP8xqdkjmKSEBujHNxh5Tk5A3T0I7AUuMmKzx00,2597
 keras_hub/api/utils/__init__.py,sha256=Gp1E6gG-RtKQS3PBEQEOz9PQvXkXaJ0ySGMqZ7myN7A,215
 keras_hub/src/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 keras_hub/src/api_export.py,sha256=9pQZK27JObxWZ96QPLBp1OBsjWigh1iuV6RglPGMRk0,1499
-keras_hub/src/version_utils.py,sha256=ZdwY5pu1WrA5i8K4EtPVnq1bDFQqenXIIvcv_Crnq7s,222
+keras_hub/src/version_utils.py,sha256=YvL4MYuxP_E9IMScqF9Nm6jTSMDP_N6_Fs_H7TR6E54,222
 keras_hub/src/layers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 keras_hub/src/layers/modeling/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 keras_hub/src/layers/modeling/alibi_bias.py,sha256=1XBTHI52L_iJDhN_w5ydu_iMhCuTgQAxEPwcLA6BPuk,4411
@@ -111,9 +111,12 @@ keras_hub/src/models/clip/clip_text_encoder.py,sha256=BCIE24eKZJ3yc4T0sjD6-Msjr1
 keras_hub/src/models/clip/clip_tokenizer.py,sha256=6gIm_LWRbCeBQUI9M2gA8-OXb4tXGygixkbcL6joV1c,7444
 keras_hub/src/models/clip/clip_vision_embedding.py,sha256=6_qC7T1dqKd-39EreGmHZj-YfjOLEDDKjWnEKcKIyuY,3667
 keras_hub/src/models/clip/clip_vision_encoder.py,sha256=q62MXySZN38uCsjqq8cttfBxD7P5abaKQV2i8_u4N6E,6385
-keras_hub/src/models/csp_darknet/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-keras_hub/src/models/csp_darknet/csp_darknet_backbone.py,sha256=J3T9eFNIdryhtAzxAjmxDjAagUmbh6QuRn17s8vJ_po,13745
-keras_hub/src/models/csp_darknet/csp_darknet_image_classifier.py,sha256=2nMkmym36EF4v3BO-wwPIjO3OsRvGloDKW0RbHGB7ag,368
+keras_hub/src/models/cspnet/__init__.py,sha256=TOpvk2cfOVv1bPA1BOGZj0mhmhc6E98zZmW9e0PIvhk,257
+keras_hub/src/models/cspnet/cspnet_backbone.py,sha256=xCeu8BpQSpf-EgCrQehQDg4jNKRAWm0h8paWBfN2DGE,41381
+keras_hub/src/models/cspnet/cspnet_image_classifier.py,sha256=JqfBHIBTFxaLOyAWx6TdXs0aAOMbcCx1oo47RoQnytc,510
+keras_hub/src/models/cspnet/cspnet_image_classifier_preprocessor.py,sha256=ACRnOhjslk2ZZhpPfJioW4um4RLYa-Suk59z9wa5vfo,543
+keras_hub/src/models/cspnet/cspnet_image_converter.py,sha256=f-ICTY2T-RlCykU6qOHDxg0fY7ECfZ_xpSJzIVmbvpc,342
+keras_hub/src/models/cspnet/cspnet_presets.py,sha256=fWzPz3eZuhFNxxPn9MJHabcXiyJA2PRRVlzNmoFBwWg,533
 keras_hub/src/models/deberta_v3/__init__.py,sha256=6E-QtAD1uvTBobrn5bUoyB1qtaCJU-t73TtbAEH6i9g,288
 keras_hub/src/models/deberta_v3/deberta_v3_backbone.py,sha256=oXdV7naTiMowuU3GsXEUo5K0GXiKbPKxdo27o5fXWjc,7258
 keras_hub/src/models/deberta_v3/deberta_v3_masked_lm.py,sha256=ADBktf1DdiP9T6LCaMhdFiZ_mUbBRKMekY5mGwAeJIo,4186
@@ -183,7 +186,7 @@ keras_hub/src/models/flux/flux_presets.py,sha256=z7C_FbI1_F5YETXuWpc7Yh_0w-5N0eB
 keras_hub/src/models/flux/flux_text_to_image.py,sha256=Rf5dD2EhG0bE8Gyg9sqaA8YEexS1kdraofIkxiZDjvc,4166
 keras_hub/src/models/flux/flux_text_to_image_preprocessor.py,sha256=Fs9jr97QtmRUbRRz1kITpkuhDM2GoV3n0XSFC-qQA14,2252
 keras_hub/src/models/gemma/__init__.py,sha256=rVzOJMJ39bgVlT8UdC0t8PlN2c237GKTBmfHIsbPuOQ,251
-keras_hub/src/models/gemma/gemma_attention.py,sha256=wlpLfr74MOtJ18JFku6jpV2c_iRav5Q1bZiFm9e1nPU,9291
+keras_hub/src/models/gemma/gemma_attention.py,sha256=XShBTunOWQOOE4Aapy3HdV9uIWuMcdNdYS1k1P3ia60,9708
 keras_hub/src/models/gemma/gemma_backbone.py,sha256=GzAUSArw_pN9dtWQzTVhWDbW-XyWt4GyMcFLn9hwmh0,13391
 keras_hub/src/models/gemma/gemma_causal_lm.py,sha256=3OXaIXlrKqMIuUnBk-bUz-0SYFL-XkkQTWm8qRY2YII,16770
 keras_hub/src/models/gemma/gemma_causal_lm_preprocessor.py,sha256=bpKkEurWIfa6Kp9s4pz84-sBDSA6ZFNHP8nXG1fFQrg,2912
@@ -237,11 +240,11 @@ keras_hub/src/models/mit/mit_image_converter.py,sha256=Mw7nV-OzyBveGuZUNFsPPKyq9
 keras_hub/src/models/mit/mit_layers.py,sha256=HUJO5uhJ6jgwANpwbQdPlEVwLRVb3BZQ-Ftjg3B9XvY,9734
 keras_hub/src/models/mit/mit_presets.py,sha256=ooLrh2OoGZKxnCGnhB6BynYJtVCXH7nDDFhgQRWt36U,4528
 keras_hub/src/models/mobilenet/__init__.py,sha256=hxkNGGj_iAMu62iooUDEPA818sNOIgjG7pXMLEMOsAE,275
-keras_hub/src/models/mobilenet/mobilenet_backbone.py,sha256=kUEDzML7MzXjUVKDY0BFn-sjGFJbu8IB8DBvG8t2nLA,28880
-keras_hub/src/models/mobilenet/mobilenet_image_classifier.py,sha256=jgyEn_WO5zLqC9UIuRV564rIY_8bMsmuUBaJA9ag5N0,2310
+keras_hub/src/models/mobilenet/mobilenet_backbone.py,sha256=aZBSFeLUObYYoi3od9DI1KfgPCqh5GHTcAI8Y2ZHShA,29536
+keras_hub/src/models/mobilenet/mobilenet_image_classifier.py,sha256=rgPVJeSRqyp3-Fgf5ERbg_97c4cSawRmAtoJpdBN8WA,2437
 keras_hub/src/models/mobilenet/mobilenet_image_classifier_preprocessor.py,sha256=yqM4wQ3ae7wXTBO0aMuvJx6XqllA7Psqzjvpm2NABXM,573
 keras_hub/src/models/mobilenet/mobilenet_image_converter.py,sha256=a3Ka0UYYK5wHSOjf2oMHSgofRazTAeUfttklVefq14w,360
-keras_hub/src/models/mobilenet/mobilenet_presets.py,sha256=PVakQu7HFRqtQdDHUzuvx_tyTWPc7BwKmWTPiHzP9C8,474
+keras_hub/src/models/mobilenet/mobilenet_presets.py,sha256=--nhaM6LmaiCtQlZPDwoQTHW7ciU0igzS4f9ssdD9Lo,1903
 keras_hub/src/models/mobilenet/util.py,sha256=S7j4UacmVIJ3fU8cymyAoK49eHcpWIKTOyUQiEjcbzQ,721
 keras_hub/src/models/opt/__init__.py,sha256=6Ybj8etxNaPsVcuZvaeHnKB3As92Px--dbiFAqOCIT0,239
 keras_hub/src/models/opt/opt_backbone.py,sha256=mK5z_E5mSiIX5s0w4hr4IVQpT7K46W2ajZBmuMjxwaY,5873
@@ -313,8 +316,8 @@ keras_hub/src/models/siglip/siglip_image_converter.py,sha256=yjYc0XOyL37WLlr-X6V
 keras_hub/src/models/siglip/siglip_layers.py,sha256=c20n6v3cFsI-Im9GBVTknhj_IpX79I4a-fajBKRMzQA,19893
 keras_hub/src/models/siglip/siglip_loss.py,sha256=n6zmOeL0o7Nwb5iaoEZfrxiAsQoqZ9yLIlaCJsAfTg4,1442
 keras_hub/src/models/siglip/siglip_preprocessor.py,sha256=r1Ej7hVwr5BudFYTHkjW5yc3lk4OYZD1s3t32lKkuec,5660
-keras_hub/src/models/siglip/siglip_presets.py,sha256=YP4xxPcJtUyhbV3hFHKyC5OhL2nn4f-Bi9NWbyNjqM4,4880
-keras_hub/src/models/siglip/siglip_text_encoder.py,sha256=P7r2eyWuPiXxi4mkbc0fbzols_ReGIYraQPCgRjYDJE,5053
+keras_hub/src/models/siglip/siglip_presets.py,sha256=EOx72XhwD_kflqn1kgwosRc3a6QqDE2ku5Pkxg0kfMI,13213
+keras_hub/src/models/siglip/siglip_text_encoder.py,sha256=xOVvzyQHLX9ne30y4ussar99gNMXPXHYKlkbCX_On2Y,5380
 keras_hub/src/models/siglip/siglip_tokenizer.py,sha256=j_67JbIHJDRk-CbiemG2dgAO6lp3_0_JdnfroZ90G18,2579
 keras_hub/src/models/siglip/siglip_vision_encoder.py,sha256=CaNaFq5thBC3TUXXOf2qknk5vWsauM20ZoaDPYRnXcs,5927
 keras_hub/src/models/stable_diffusion_3/__init__.py,sha256=ZKYQuaRObyhKq8GVAHmoRvlXp6FpU8ChvutVCHyXKuc,343
@@ -396,20 +399,21 @@ keras_hub/src/tokenizers/unicode_codepoint_tokenizer.py,sha256=hRv_XxoPIPDpHfO0Z
 keras_hub/src/tokenizers/word_piece_tokenizer.py,sha256=vP6AZgbzsRiuPCt3W_n94nsF7XiERnagWcH_rqJHtVU,19943
 keras_hub/src/tokenizers/word_piece_tokenizer_trainer.py,sha256=cylrs02ZrYQ1TuZr9oyS3NrVbDwGctA3VXbIh1pFJMQ,6743
 keras_hub/src/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-keras_hub/src/utils/keras_utils.py,sha256=TNgp3ukTiCA0jrGUq2ZV_Xqtzc7CfiFQKyOH5t47z48,2313
+keras_hub/src/utils/keras_utils.py,sha256=IB_eIrln3N5sVyCapwv1jzLEmuBv8vBRwSVd3toSgyI,3097
 keras_hub/src/utils/pipeline_model.py,sha256=jgzB6NQPSl0KOu08N-TazfOnXnUJbZjH2EXXhx25Ftg,9084
-keras_hub/src/utils/preset_utils.py,sha256=ZbSEUSacKlr_mgHyB3ChUohgOQN7nMCkE6E2lGxt2HA,31927
+keras_hub/src/utils/preset_utils.py,sha256=5xEm6Uz1vfQkBqyENt97qaxWoq-P7mlPC0LIpXqDM70,31928
 keras_hub/src/utils/python_utils.py,sha256=N8nWeO3san4YnGkffRXG3Ix7VEIMTKSN21FX5TuL7G8,202
 keras_hub/src/utils/tensor_utils.py,sha256=lczQWgPVJU09cLtNbo8MErVFNV9ne4gNlrzbNVQazg4,15042
 keras_hub/src/utils/imagenet/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 keras_hub/src/utils/imagenet/imagenet_utils.py,sha256=MvIvv1WJo51ZXBxy4S7t_DsN3ZMtJWlC4cmRvKM2kIA,39304
 keras_hub/src/utils/timm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+keras_hub/src/utils/timm/convert_cspnet.py,sha256=O5HCdeKcSFWOoFr8_wIUQb4Noc0tBEo5Aogk2d6SEes,5676
 keras_hub/src/utils/timm/convert_densenet.py,sha256=fu8HBIQis5o3ib2tyI2qnmYScVrVIQySok8vTfa1qJ8,3393
 keras_hub/src/utils/timm/convert_efficientnet.py,sha256=SgEIlyyinS04qoQpEgh3WazHq544zNUCCpfmWh3EjSs,17100
-keras_hub/src/utils/timm/convert_mobilenet.py,sha256=0CHzc2kk36C1aaxt8x1UmfcxPtywQ8Jvfgt_6N8xICw,7215
+keras_hub/src/utils/timm/convert_mobilenet.py,sha256=XTqHOK4nJwigKefsw7ktWJtOgRpEVMO9MtRhuP5qP_k,9219
 keras_hub/src/utils/timm/convert_resnet.py,sha256=8JFkVtdpy5z9h83LJ97rD-a8FRejXPZvMNksNuStqjM,5834
 keras_hub/src/utils/timm/convert_vgg.py,sha256=MT5jGnLrzenPpe66Af_Lp1IdR9KGtsSrcmn6_UPqHvQ,2419
-keras_hub/src/utils/timm/preset_loader.py,sha256=yhEV8D99GszCCpOw4I9GdOzAWJB0_gBnlS6ecaaIRGk,3518
+keras_hub/src/utils/timm/preset_loader.py,sha256=j2HYi61Zbt0CGd33evFJ8j2fraXl0Zardf4qqAb82K0,3841
 keras_hub/src/utils/transformers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 keras_hub/src/utils/transformers/convert_albert.py,sha256=VdKclZpCxtDWq3UbUUQZf4fR9DJK_JYZ73B4O_G9skg,7695
 keras_hub/src/utils/transformers/convert_bart.py,sha256=Tk4h9Md9rwN5wjQbGIVrC7qzDpF8kI8qm-FKL8HlUok,14411
@@ -423,7 +427,7 @@ keras_hub/src/utils/transformers/convert_pali_gemma.py,sha256=B1leeDw96Yvu81hYum
 keras_hub/src/utils/transformers/convert_vit.py,sha256=9SUZ9utNJhW_5cj3acMn9cRy47u2eIcDsrhmzj77o9k,5187
 keras_hub/src/utils/transformers/preset_loader.py,sha256=DgGJXbTSB9Na8FIR-YWWVqQPOFxHwWrGm41EwcS_EFs,3797
 keras_hub/src/utils/transformers/safetensor_utils.py,sha256=CYUHyA4y-B61r7NDnCsFb4t_UmSwZ1k9L-8gzEd6KRg,3339
-keras_hub_nightly-0.19.0.dev202503060350.dist-info/METADATA,sha256=xr9oeYET8vNffBv3_FyHZEIbYRa8l7DQYTcdd-aumoo,7715
-keras_hub_nightly-0.19.0.dev202503060350.dist-info/WHEEL,sha256=jB7zZ3N9hIM9adW7qlTAyycLYW9npaWKLRzaoVcLKcM,91
-keras_hub_nightly-0.19.0.dev202503060350.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
-keras_hub_nightly-0.19.0.dev202503060350.dist-info/RECORD,,
+keras_hub_nightly-0.20.0.dev202503150350.dist-info/METADATA,sha256=0YkTiBXYORlmFuF6Hw7cPNf5SMx9UpToaYpls8o1Ddk,7715
+keras_hub_nightly-0.20.0.dev202503150350.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
+keras_hub_nightly-0.20.0.dev202503150350.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
+keras_hub_nightly-0.20.0.dev202503150350.dist-info/RECORD,,

{keras_hub_nightly-0.19.0.dev202503060350.dist-info → keras_hub_nightly-0.20.0.dev202503150350.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.8.2)
+Generator: setuptools (76.0.0)
 Root-Is-Purelib: true
 Tag: py3-none-any