keras-hub-nightly 0.23.0.dev202510100415__py3-none-any.whl → 0.23.0.dev202510110411__py3-none-any.whl

keras_hub/src/utils/timm/convert_mobilenetv5.py

@@ -0,0 +1,321 @@
+import types
+
+import keras
+import numpy as np
+
+from keras_hub.src.models.mobilenetv5.mobilenetv5_attention import (
+    MobileAttention,
+)
+from keras_hub.src.models.mobilenetv5.mobilenetv5_backbone import (
+    MobileNetV5Backbone,
+)
+from keras_hub.src.models.mobilenetv5.mobilenetv5_blocks import EdgeResidual
+from keras_hub.src.models.mobilenetv5.mobilenetv5_blocks import (
+    UniversalInvertedResidual,
+)
+from keras_hub.src.models.mobilenetv5.mobilenetv5_builder import (
+    convert_arch_def_to_stackwise,
+)
+from keras_hub.src.models.mobilenetv5.mobilenetv5_layers import ConvNormAct
+from keras_hub.src.models.mobilenetv5.mobilenetv5_layers import RmsNorm2d
+
+backbone_cls = MobileNetV5Backbone
+
+MODEL_CONFIGS = {
+    "mobilenetv5_300m": {
+        "backbone": convert_arch_def_to_stackwise(
+            [
+                # Stage 0: 128x128 in
+                [
+                    "er_r1_k3_s2_e4_c128",
+                    "er_r1_k3_s1_e4_c128",
+                    "er_r1_k3_s1_e4_c128",
+                ],
+                # Stage 1: 256x256 in
+                [
+                    "uir_r1_a3_k5_s2_e6_c256",
+                    "uir_r1_a5_k0_s1_e4_c256",
+                    "uir_r1_a3_k0_s1_e4_c256",
+                    "uir_r1_a5_k0_s1_e4_c256",
+                    "uir_r1_a3_k0_s1_e4_c256",
+                ],
+                # Stage 2: 640x640 in
+                [
+                    "uir_r1_a5_k5_s2_e6_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a5_k0_s1_e4_c640",
+                    "uir_r1_a0_k0_s1_e1_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                    "mqa_r1_k3_h12_v2_s1_d64_c640",
+                    "uir_r1_a0_k0_s1_e2_c640",
+                ],
+                # Stage 3: 1280x1280 in
+                [
+                    "uir_r1_a5_k5_s2_e6_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                    "mqa_r1_k3_h16_s1_d96_c1280",
+                    "uir_r1_a0_k0_s1_e2_c1280",
+                ],
+            ]
+        ),
+        "stem_size": 64,
+        "num_features": 2048,
+        "norm_layer": "rms_norm",
+        "act_layer": "gelu",
+        "use_msfa": True,
+        "layer_scale_init_value": 1e-5,
+    },
+}
+
+
+def convert_head(task, loader, timm_config):
+    pass
+
+
+def convert_backbone_config(timm_config):
+    timm_architecture = timm_config["architecture"]
+    if timm_architecture not in MODEL_CONFIGS:
+        raise ValueError(f"Unsupported architecture: {timm_architecture}")
+    config = MODEL_CONFIGS[timm_architecture].copy()
+    backbone_config = config.pop("backbone")
+    backbone_config.update(config)
+    return backbone_config
+
+
+def convert_weights(backbone, loader, timm_config):
+    def key_exists(key):
+        try:
+            loader.get_tensor(key)
+            return True
+        except Exception:
+            return False
+
+    def _port_weights(layer, timm_key, transpose_dims=None):
+        hf_weight_key = f"{timm_key}.weight"
+        if not key_exists(hf_weight_key):
+            return
+        hook_fn = None
+        if transpose_dims:
+
+            def transpose_hook(x, _):
+                return np.transpose(x, transpose_dims)
+
+            hook_fn = transpose_hook
+        loader.port_weight(
+            layer.kernel, hf_weight_key=hf_weight_key, hook_fn=hook_fn
+        )
+        if layer.bias is not None:
+            hf_bias_key = f"{timm_key}.bias"
+            if key_exists(hf_bias_key):
+                loader.port_weight(
+                    layer.bias,
+                    hf_weight_key=hf_bias_key,
+                )
+
+    def _port_bn(layer, timm_prefix):
+        loader.port_weight(layer.gamma, f"{timm_prefix}.weight")
+        loader.port_weight(layer.beta, f"{timm_prefix}.bias")
+        loader.port_weight(layer.moving_mean, f"{timm_prefix}.running_mean")
+        loader.port_weight(layer.moving_variance, f"{timm_prefix}.running_var")
+
+    def _port_rms_norm(layer, timm_prefix):
+        loader.port_weight(layer.gamma, f"{timm_prefix}.weight")
+
+    def _port_cna(cna_layer: ConvNormAct, timm_conv_prefix, timm_norm_prefix):
+        if isinstance(cna_layer.conv, keras.layers.DepthwiseConv2D):
+            _port_weights(
+                cna_layer.conv,
+                timm_conv_prefix,
+                transpose_dims=(2, 3, 0, 1),
+            )
+        else:
+            _port_weights(
+                cna_layer.conv,
+                timm_conv_prefix,
+                transpose_dims=(2, 3, 1, 0),
+            )
+        if key_exists(f"{timm_norm_prefix}.running_mean"):
+            _port_bn(cna_layer.norm, timm_norm_prefix)
+        else:
+            _port_rms_norm(cna_layer.norm, timm_norm_prefix)
+
+    def _port_attn(attn_layer, attn_prefix):
+        _port_weights(
+            attn_layer.query_layers[-1],
+            f"{attn_prefix}.query.proj",
+            (2, 3, 1, 0),
+        )
+        if len(attn_layer.key_layers) > 1:
+            _port_weights(
+                attn_layer.key_layers[0],
+                f"{attn_prefix}.key.down_conv",
+                (2, 3, 0, 1),
+            )
+            key_norm_layer = attn_layer.key_layers[1]
+            if isinstance(key_norm_layer, RmsNorm2d):
+                _port_rms_norm(key_norm_layer, f"{attn_prefix}.key.norm")
+            else:
+                _port_bn(key_norm_layer, f"{attn_prefix}.key.norm")
+        _port_weights(
+            attn_layer.key_layers[-1], f"{attn_prefix}.key.proj", (2, 3, 1, 0)
+        )
+        if len(attn_layer.value_layers) > 1:
+            _port_weights(
+                attn_layer.value_layers[0],
+                f"{attn_prefix}.value.down_conv",
+                (2, 3, 0, 1),
+            )
+            value_norm_layer = attn_layer.value_layers[1]
+            if isinstance(value_norm_layer, RmsNorm2d):
+                _port_rms_norm(value_norm_layer, f"{attn_prefix}.value.norm")
+            else:
+                _port_bn(value_norm_layer, f"{attn_prefix}.value.norm")
+        _port_weights(
+            attn_layer.value_layers[-1],
+            f"{attn_prefix}.value.proj",
+            (2, 3, 1, 0),
+        )
+        _port_weights(
+            attn_layer.output_proj_layers[-2],
+            f"{attn_prefix}.output.proj",
+            (2, 3, 1, 0),
+        )
+
+    stem_layer = backbone.get_layer("conv_stem")
+    _port_cna(stem_layer, "conv_stem.conv", "conv_stem.bn")
+    block_layers = [
+        layer
+        for layer in backbone.layers
+        if isinstance(
+            layer, (EdgeResidual, UniversalInvertedResidual, MobileAttention)
+        )
+    ]
+    block_counter = 0
+    for stack_idx in range(len(backbone.stackwise_num_blocks)):
+        for block_idx_in_stage in range(
+            backbone.stackwise_num_blocks[stack_idx]
+        ):
+            block = block_layers[block_counter]
+            timm_prefix = f"blocks.{stack_idx}.{block_idx_in_stage}"
+            if isinstance(block, EdgeResidual):
+                _port_cna(
+                    block.conv_exp,
+                    f"{timm_prefix}.conv_exp",
+                    f"{timm_prefix}.bn1",
+                )
+                _port_cna(
+                    block.conv_pwl,
+                    f"{timm_prefix}.conv_pwl",
+                    f"{timm_prefix}.bn2",
+                )
+            elif isinstance(block, UniversalInvertedResidual):
+                if hasattr(block, "dw_start") and not isinstance(
+                    block.dw_start, types.FunctionType
+                ):
+                    _port_cna(
+                        block.dw_start,
+                        f"{timm_prefix}.dw_start.conv",
+                        f"{timm_prefix}.dw_start.bn",
+                    )
+                _port_cna(
+                    block.pw_exp,
+                    f"{timm_prefix}.pw_exp.conv",
+                    f"{timm_prefix}.pw_exp.bn",
+                )
+                if hasattr(block, "dw_mid") and not isinstance(
+                    block.dw_mid, types.FunctionType
+                ):
+                    _port_cna(
+                        block.dw_mid,
+                        f"{timm_prefix}.dw_mid.conv",
+                        f"{timm_prefix}.dw_mid.bn",
+                    )
+                _port_cna(
+                    block.pw_proj,
+                    f"{timm_prefix}.pw_proj.conv",
+                    f"{timm_prefix}.pw_proj.bn",
+                )
+                gamma_key = f"{timm_prefix}.layer_scale.gamma"
+                if key_exists(gamma_key):
+                    loader.port_weight(block.layer_scale.gamma, gamma_key)
+            elif isinstance(block, MobileAttention):
+                _port_rms_norm(block.norm, f"{timm_prefix}.norm")
+                gamma_key = f"{timm_prefix}.layer_scale.gamma"
+                if key_exists(gamma_key):
+                    loader.port_weight(block.layer_scale.gamma, gamma_key)
+                attn_prefix = f"{timm_prefix}.attn"
+                _port_attn(block.attn, attn_prefix)
+            block_counter += 1
+    try:
+        msfa_layer = backbone.get_layer("msfa")
+        ffn = msfa_layer.ffn
+        _port_cna(ffn.pw_exp, "msfa.ffn.pw_exp.conv", "msfa.ffn.pw_exp.bn")
+        _port_cna(ffn.pw_proj, "msfa.ffn.pw_proj.conv", "msfa.ffn.pw_proj.bn")
+        _port_rms_norm(msfa_layer.norm, "msfa.norm")
+    except ValueError:
+        pass

keras_hub/src/utils/timm/preset_loader.py

@@ -7,6 +7,7 @@ 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
+from keras_hub.src.utils.timm import convert_mobilenetv5
 from keras_hub.src.utils.timm import convert_resnet
 from keras_hub.src.utils.timm import convert_vgg
 from keras_hub.src.utils.transformers.safetensor_utils import SafetensorLoader
@@ -22,6 +23,8 @@ class TimmPresetLoader(PresetLoader):
             self.converter = convert_cspnet
         elif architecture.startswith("densenet"):
             self.converter = convert_densenet
+        elif architecture.startswith("mobilenetv5"):
+            self.converter = convert_mobilenetv5
         elif architecture.startswith("mobilenet"):
             self.converter = convert_mobilenet
         elif architecture.startswith("vgg"):
@@ -41,7 +44,8 @@ class TimmPresetLoader(PresetLoader):
         keras_config = self.converter.convert_backbone_config(self.config)
         backbone = cls(**{**keras_config, **kwargs})
         if load_weights:
-            jax_memory_cleanup(backbone)
+            if not self.config["architecture"].startswith("mobilenetv5"):
+                jax_memory_cleanup(backbone)
             # Use prefix="" to avoid using `get_prefixed_key`.
             with SafetensorLoader(self.preset, prefix="") as loader:
                 self.converter.convert_weights(backbone, loader, self.config)
@@ -54,9 +58,9 @@ 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"]
+        if "num_features" in self.config and (
+            "mobilenet" in self.config["architecture"]
+            or "mobilenetv5" in self.config["architecture"]
         ):
             kwargs["num_features"] = self.config["num_features"]
 

keras_hub/src/version.py

@@ -1,7 +1,7 @@
 from keras_hub.src.api_export import keras_hub_export
 
 # Unique source of truth for the version number.
-__version__ = "0.23.0.dev202510100415"
+__version__ = "0.23.0.dev202510110411"
 
 
 @keras_hub_export("keras_hub.version")

{keras_hub_nightly-0.23.0.dev202510100415.dist-info → keras_hub_nightly-0.23.0.dev202510110411.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: keras-hub-nightly
-Version: 0.23.0.dev202510100415
+Version: 0.23.0.dev202510110411
 Summary: Pretrained models for Keras.
 Author-email: Keras team <keras-users@googlegroups.com>
 License-Expression: Apache-2.0

{keras_hub_nightly-0.23.0.dev202510100415.dist-info → keras_hub_nightly-0.23.0.dev202510110411.dist-info}/RECORD

@@ -1,11 +1,11 @@
 keras_hub/__init__.py,sha256=bJbUZkqwhZvTb1Tqx1fbkq6mzBYiEyq-Hin3oQIkhdE,558
-keras_hub/layers/__init__.py,sha256=8FTy8HwjgFdBvbl_QKTxXmOc13TXjUUBgLYrSTtkc0M,5807
+keras_hub/layers/__init__.py,sha256=ufJKHxMTFhwp--E3ixfGCZqq89pZOUOxCQYgI5pEUA8,5944
 keras_hub/metrics/__init__.py,sha256=KYalsMPBnfwim9BdGHFfJ5WxUKFXOQ1QoKIMT_0lwlM,439
-keras_hub/models/__init__.py,sha256=wy75CGuTVxRIEXSCqmXgMyf23vUbuRbByWrlAaPWXB0,30737
+keras_hub/models/__init__.py,sha256=E9_kQFlM75iKRiSUFEF7F8e9aqMo89-88XSSLy2sbtY,31172
 keras_hub/samplers/__init__.py,sha256=aFQIkiqbZpi8vjrPp2MVII4QUfE-eQjra5fMeHsoy7k,886
 keras_hub/src/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 keras_hub/src/api_export.py,sha256=9pQZK27JObxWZ96QPLBp1OBsjWigh1iuV6RglPGMRk0,1499
-keras_hub/src/version.py,sha256=XWjJrPwsNeGHKdXmdJIHeXWcFLk329AcGB-njcIbEik,222
+keras_hub/src/version.py,sha256=cgZOW8wAMr-nVcvJ7qZLsTzj0scV7mGTt7a3Nt5Uvl4,222
 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
@@ -321,6 +321,16 @@ keras_hub/src/models/mobilenet/mobilenet_image_classifier_preprocessor.py,sha256
 keras_hub/src/models/mobilenet/mobilenet_image_converter.py,sha256=a3Ka0UYYK5wHSOjf2oMHSgofRazTAeUfttklVefq14w,360
 keras_hub/src/models/mobilenet/mobilenet_presets.py,sha256=hR_3xxI_PigE8UprXW4lAuKRa3LFGdidBaN8LklxwRQ,1895
 keras_hub/src/models/mobilenet/util.py,sha256=S7j4UacmVIJ3fU8cymyAoK49eHcpWIKTOyUQiEjcbzQ,721
+keras_hub/src/models/mobilenetv5/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+keras_hub/src/models/mobilenetv5/mobilenetv5_attention.py,sha256=rH4cp1B5_r8g7gKvDdMvfEGfmMHUB2OMEbWQbX9yUMg,26499
+keras_hub/src/models/mobilenetv5/mobilenetv5_backbone.py,sha256=y19FpVh0M3w9jSmP34E-GixdjMsU2cEJKrtjLcFfGZU,17167
+keras_hub/src/models/mobilenetv5/mobilenetv5_blocks.py,sha256=aw2H-duaCkxGSHB-nKnG8nQhAPxNkmlPUn0FHDb_cTs,34026
+keras_hub/src/models/mobilenetv5/mobilenetv5_builder.py,sha256=_vgjuqJq9GXlccKaL783q77rgtzfa0Oc9aNOhGWeprc,17092
+keras_hub/src/models/mobilenetv5/mobilenetv5_image_classifier.py,sha256=BvL0yClapvG9I5hNMUFuYCXXfnBsBGmKR0ICL6MQqrc,5944
+keras_hub/src/models/mobilenetv5/mobilenetv5_image_classifier_preprocessor.py,sha256=4yhM71JqOzQWiCXTidWAMCNaaIO8QVq5vXl_129ylsI,602
+keras_hub/src/models/mobilenetv5/mobilenetv5_image_converter.py,sha256=HroX3OOwajIz9CIqlcGf9K9MYUEQ86wifABy9ZGRql4,381
+keras_hub/src/models/mobilenetv5/mobilenetv5_layers.py,sha256=wLyR_gTcqiNLUt86fhxhEbbhZH3YA9CbvMSPnA4vTvg,15889
+keras_hub/src/models/mobilenetv5/mobilenetv5_utils.py,sha256=ij2Dfguotb6RGSXuX-MsxF8JBXBWLdSi0LfDOKGp2rk,4868
 keras_hub/src/models/moonshine/__init__.py,sha256=WK_9Cy1dp5KplNAaTsaJbd-2DGLsiHQsIL5ZnXuCbDQ,275
 keras_hub/src/models/moonshine/moonshine_audio_converter.py,sha256=FnvR7SP44uVOsA3g9azUhQjsVg809eJ5nqoJZQ-DAq0,11854
 keras_hub/src/models/moonshine/moonshine_audio_to_text.py,sha256=dXFtjaxL1jpcIAiiZY1-kcNL-S4RiRJiAC2uR_a3Fyc,15865
@@ -558,7 +568,7 @@ keras_hub/src/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSu
 keras_hub/src/utils/keras_utils.py,sha256=IWsbg-p-XVLuOkba8PAYNf9zDo4G2RkINLr58p12MhA,5291
 keras_hub/src/utils/openvino_utils.py,sha256=P1ZvedLv91LZD-UAgAo2dy6WC5305elh1qvgmdYQIGc,4512
 keras_hub/src/utils/pipeline_model.py,sha256=jgzB6NQPSl0KOu08N-TazfOnXnUJbZjH2EXXhx25Ftg,9084
-keras_hub/src/utils/preset_utils.py,sha256=vSs7U9cy0p6UqOEyGvudzL-o3mxl3FX22r4XH6rOgMg,37309
+keras_hub/src/utils/preset_utils.py,sha256=lyCg_PRcYH1Jy8lGKaO8sgpIbMrP-Ik66EbjGD4gizc,37677
 keras_hub/src/utils/python_utils.py,sha256=N8nWeO3san4YnGkffRXG3Ix7VEIMTKSN21FX5TuL7G8,202
 keras_hub/src/utils/tensor_utils.py,sha256=bGM0pK-x0R4640emul49GfSZ3p4OSvOaVzZZPlm6eiM,16957
 keras_hub/src/utils/coco/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -570,9 +580,10 @@ keras_hub/src/utils/timm/convert_cspnet.py,sha256=9p1IF0B4UPbDTruQQXR6mJEUdhvQvH
 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=XTqHOK4nJwigKefsw7ktWJtOgRpEVMO9MtRhuP5qP_k,9219
+keras_hub/src/utils/timm/convert_mobilenetv5.py,sha256=B4qDcVH_v0dZCwcDmUnufbORbwpj-al8atnFMQX3bcg,12437
 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=4hULdq2K2hgPYTZR71PGV4YNDHLG1zcoxF9TXpg6fGE,3905
+keras_hub/src/utils/timm/preset_loader.py,sha256=3bNmKinKjwc5-ToPCrT2dC2MsvD9tpIMzEKRknrnB5A,4190
 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
@@ -599,7 +610,7 @@ keras_hub/src/utils/transformers/export/gemma.py,sha256=xX_vfQwvFZ_-lQX4kgMNOGKL
 keras_hub/src/utils/transformers/export/hf_exporter.py,sha256=Qk52c6LIA2eMHUNY9Vy4STJSpnhLMdJ_t-3ljqhSr4k,5081
 keras_hub/tokenizers/__init__.py,sha256=kyFWYm4mb--U4xYU-2Gb1COM8xEFWNK6LcKxr8h9Ivc,4561
 keras_hub/utils/__init__.py,sha256=jXPqVGBpJr_PpYmqD8aDG-fRMlxH-ulqCR2SZMn288Y,646
-keras_hub_nightly-0.23.0.dev202510100415.dist-info/METADATA,sha256=xpB7dOP6MkBQwzdaZL_NOKsxidg3ZUivQ2PWJjpEwhI,7395
-keras_hub_nightly-0.23.0.dev202510100415.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-keras_hub_nightly-0.23.0.dev202510100415.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
-keras_hub_nightly-0.23.0.dev202510100415.dist-info/RECORD,,
+keras_hub_nightly-0.23.0.dev202510110411.dist-info/METADATA,sha256=MxgUS9-n5WEsrwX-tpKOROxlSp55X-Ygd6XC9TaBx5c,7395
+keras_hub_nightly-0.23.0.dev202510110411.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+keras_hub_nightly-0.23.0.dev202510110411.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
+keras_hub_nightly-0.23.0.dev202510110411.dist-info/RECORD,,