keras-hub-nightly 0.23.0.dev202508250413__py3-none-any.whl → 0.23.0.dev202508260411__py3-none-any.whl

keras_hub/src/models/t5gemma/t5gemma_tokenizer.py

@@ -0,0 +1,84 @@
+from keras_hub.src.api_export import keras_hub_export
+from keras_hub.src.models.t5gemma.t5gemma_backbone import T5GemmaBackbone
+from keras_hub.src.tokenizers.sentence_piece_tokenizer import (
+    SentencePieceTokenizer,
+)
+
+
+@keras_hub_export(
+    [
+        "keras_hub.tokenizers.T5GemmaTokenizer",
+        "keras_hub.models.T5GemmaTokenizer",
+    ]
+)
+class T5GemmaTokenizer(SentencePieceTokenizer):
+    """T5Gemma tokenizer layer based on SentencePiece.
+
+    This tokenizer class will tokenize raw strings into integer sequences and
+    is based on `keras_hub.tokenizers.SentencePieceTokenizer`. Unlike the
+    underlying tokenizer, it will check for all special tokens needed by
+    T5Gemma models and provides a `from_preset()` method to automatically
+    download a matching vocabulary for a T5Gemma preset.
+
+    If input is a batch of strings (rank > 0), the layer will output a
+    `tf.RaggedTensor` where the last dimension of the output is ragged.
+
+    If input is a scalar string (rank == 0), the layer will output a dense
+    `tf.Tensor` with static shape `[None]`.
+
+    Args:
+        proto: Either a `string` path to a SentencePiece proto file, or a
+            `bytes` object with a serialized SentencePiece proto. See the
+            [SentencePiece repository](https://github.com/google/sentencepiece)
+            for more details on the format.
+
+    Examples:
+
+    ```python
+    import io
+    import tensorflow as tf
+    import sentencepiece
+
+    # Unbatched input.
+    tokenizer = keras_hub.models.T5GemmaTokenizer.from_preset(
+        "t5gemma_b_b_prefixlm_it"
+    )
+    tokenizer("The quick brown fox jumped.")
+
+    # Batched input.
+    tokenizer(["The quick brown fox jumped.", "The fox slept."])
+
+    # Detokenization.
+    tokenizer.detokenize(tokenizer("The quick brown fox jumped."))
+
+    # Custom vocabulary.
+    bytes_io = io.BytesIO()
+    ds = tf.data.Dataset.from_tensor_slices(["The quick brown fox jumped."])
+    sentencepiece.SentencePieceTrainer.train(
+        sentence_iterator=ds.as_numpy_iterator(),
+        model_writer=bytes_io,
+        vocab_size=8,
+        model_type="WORD",
+        pad_id=0,
+        bos_id=1,
+        eos_id=2,
+        unk_id=3,
+        pad_piece="<pad>",
+        bos_piece="<bos>",
+        eos_piece="<eos>",
+        unk_piece="<unk>",
+    )
+    tokenizer = keras_hub.models.T5GemmaTokenizer(
+        proto=bytes_io.getvalue(),
+    )
+    tokenizer("The quick brown fox jumped.")
+    ```
+    """
+
+    backbone_cls = T5GemmaBackbone
+
+    def __init__(self, proto, **kwargs):
+        self._add_special_token("<bos>", "start_token")
+        self._add_special_token("<eos>", "end_token")
+        self._add_special_token("<pad>", "pad_token")
+        super().__init__(proto=proto, **kwargs)

keras_hub/src/utils/transformers/convert_t5gemma.py

@@ -0,0 +1,229 @@
+from keras_hub.src.models.t5gemma.t5gemma_backbone import T5GemmaBackbone
+from keras_hub.src.utils.preset_utils import get_file
+
+backbone_cls = T5GemmaBackbone
+
+
+def convert_backbone_config(transformers_config):
+    """Convert a Hugging Face T5Gemma config to a KerasHub backbone config."""
+    encoder_config = transformers_config["encoder"]
+    decoder_config = transformers_config["decoder"]
+
+    if decoder_config.get("hidden_activation") == "gelu_pytorch_tanh":
+        decoder_config["hidden_activation"] = "gelu_approximate"
+    if encoder_config.get("hidden_activation") == "gelu_pytorch_tanh":
+        encoder_config["hidden_activation"] = "gelu_approximate"
+
+    backbone_config = {
+        "vocabulary_size": decoder_config["vocab_size"],
+        "encoder_hidden_dim": encoder_config["hidden_size"],
+        "encoder_intermediate_dim": encoder_config["intermediate_size"],
+        "encoder_num_layers": encoder_config["num_hidden_layers"],
+        "encoder_num_attention_heads": encoder_config["num_attention_heads"],
+        "encoder_num_key_value_heads": encoder_config["num_key_value_heads"],
+        "encoder_head_dim": encoder_config["head_dim"],
+        "encoder_layer_types": encoder_config["layer_types"],
+        "decoder_hidden_dim": decoder_config["hidden_size"],
+        "decoder_intermediate_dim": decoder_config["intermediate_size"],
+        "decoder_num_layers": decoder_config["num_hidden_layers"],
+        "decoder_num_attention_heads": decoder_config["num_attention_heads"],
+        "decoder_num_key_value_heads": decoder_config["num_key_value_heads"],
+        "decoder_head_dim": decoder_config["head_dim"],
+        "decoder_layer_types": decoder_config["layer_types"],
+        "dropout_rate": decoder_config["dropout_rate"],
+        "rms_norm_eps": decoder_config["rms_norm_eps"],
+        "query_pre_attn_scalar": decoder_config["query_pre_attn_scalar"],
+        "tie_word_embeddings": transformers_config.get(
+            "tie_word_embeddings", True
+        ),
+        "attention_bias": decoder_config["attention_bias"],
+        "hidden_activation": decoder_config["hidden_activation"],
+        "initializer_range": decoder_config["initializer_range"],
+        "attention_dropout": decoder_config["attention_dropout"],
+        "sliding_window": decoder_config["sliding_window"],
+        "cross_attention_hidden_size": encoder_config["hidden_size"],
+        "attn_logit_softcapping": decoder_config["attn_logit_softcapping"],
+        "final_logit_softcapping": decoder_config["final_logit_softcapping"],
+        "rope_max_wavelength": decoder_config["rope_theta"],
+    }
+    return backbone_config
+
+
+def convert_weights(backbone, loader, transformers_config):
+    """Convert T5Gemma from Hugging Face to KerasHub."""
+    # Token embeddings.
+    loader.port_weight(
+        keras_variable=backbone.token_embedding.embeddings,
+        hf_weight_key="encoder.embed_tokens.weight",
+    )
+    loader.port_weight(
+        keras_variable=backbone.decoder_token_embedding.embeddings,
+        hf_weight_key="decoder.embed_tokens.weight",
+    )
+
+    # Encoder.
+    loader.port_weight(
+        keras_variable=backbone.encoder_norm.scale,
+        hf_weight_key="encoder.norm.weight",
+    )
+    for i in range(backbone.encoder_num_layers):
+        layer = backbone.get_layer(f"encoder_layer_{i}")
+        hf_prefix = f"encoder.layers.{i}"
+
+        # Self-attention.
+        loader.port_weight(
+            keras_variable=layer.self_attn.query_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.q_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.self_attn.key_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.k_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.self_attn.value_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.v_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.self_attn.output_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.o_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+
+        # MLP.
+        loader.port_weight(
+            keras_variable=layer.mlp.gate_proj.kernel,
+            hf_weight_key=f"{hf_prefix}.mlp.gate_proj.weight",
+            hook_fn=lambda w, s: w.T,
+        )
+        loader.port_weight(
+            keras_variable=layer.mlp.up_proj.kernel,
+            hf_weight_key=f"{hf_prefix}.mlp.up_proj.weight",
+            hook_fn=lambda w, s: w.T,
+        )
+        loader.port_weight(
+            keras_variable=layer.mlp.down_proj.kernel,
+            hf_weight_key=f"{hf_prefix}.mlp.down_proj.weight",
+            hook_fn=lambda w, s: w.T,
+        )
+
+        # Layer norm.
+        loader.port_weight(
+            keras_variable=layer.pre_self_attn_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.pre_self_attn_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.post_self_attn_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.post_self_attn_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.pre_feedforward_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.pre_feedforward_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.post_feedforward_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.post_feedforward_layernorm.weight",
+        )
+
+    # Decoder.
+    loader.port_weight(
+        keras_variable=backbone.decoder_norm.scale,
+        hf_weight_key="decoder.norm.weight",
+    )
+    for i in range(backbone.decoder_num_layers):
+        layer = backbone.get_layer(f"decoder_layer_{i}")
+        hf_prefix = f"decoder.layers.{i}"
+
+        # Self-attention.
+        loader.port_weight(
+            keras_variable=layer.self_attn.query_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.q_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.self_attn.key_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.k_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.self_attn.value_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.v_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.self_attn.output_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.self_attn.o_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+
+        # Cross-attention.
+        loader.port_weight(
+            keras_variable=layer.cross_attn.query_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.cross_attn.q_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.cross_attn.key_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.cross_attn.k_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.cross_attn.value_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.cross_attn.v_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+        loader.port_weight(
+            keras_variable=layer.cross_attn.output_dense.kernel,
+            hf_weight_key=f"{hf_prefix}.cross_attn.o_proj.weight",
+            hook_fn=lambda w, s: w.T.reshape(s),
+        )
+
+        # MLP.
+        loader.port_weight(
+            keras_variable=layer.mlp.gate_proj.kernel,
+            hf_weight_key=f"{hf_prefix}.mlp.gate_proj.weight",
+            hook_fn=lambda w, s: w.T,
+        )
+        loader.port_weight(
+            keras_variable=layer.mlp.up_proj.kernel,
+            hf_weight_key=f"{hf_prefix}.mlp.up_proj.weight",
+            hook_fn=lambda w, s: w.T,
+        )
+        loader.port_weight(
+            keras_variable=layer.mlp.down_proj.kernel,
+            hf_weight_key=f"{hf_prefix}.mlp.down_proj.weight",
+            hook_fn=lambda w, s: w.T,
+        )
+
+        # Layer norm.
+        loader.port_weight(
+            keras_variable=layer.pre_self_attn_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.pre_self_attn_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.post_self_attn_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.post_self_attn_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.pre_cross_attn_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.pre_cross_attn_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.post_cross_attn_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.post_cross_attn_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.pre_feedforward_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.pre_feedforward_layernorm.weight",
+        )
+        loader.port_weight(
+            keras_variable=layer.post_feedforward_layernorm.scale,
+            hf_weight_key=f"{hf_prefix}.post_feedforward_layernorm.weight",
+        )
+
+
+def convert_tokenizer(cls, preset, **kwargs):
+    """Convert a T5Gemma tokenizer."""
+    return cls(get_file(preset, "tokenizer.model"), **kwargs)

keras_hub/src/utils/transformers/preset_loader.py

@@ -19,6 +19,7 @@ from keras_hub.src.utils.transformers import convert_pali_gemma
 from keras_hub.src.utils.transformers import convert_qwen
 from keras_hub.src.utils.transformers import convert_qwen3
 from keras_hub.src.utils.transformers import convert_qwen_moe
+from keras_hub.src.utils.transformers import convert_t5gemma
 from keras_hub.src.utils.transformers import convert_vit
 from keras_hub.src.utils.transformers.safetensor_utils import SafetensorLoader
 
@@ -62,6 +63,8 @@ class TransformersPresetLoader(PresetLoader):
             self.converter = convert_qwen_moe
         elif model_type == "qwen3":
             self.converter = convert_qwen3
+        elif model_type == "t5gemma":
+            self.converter = convert_t5gemma
         else:
             raise ValueError(
                 "KerasHub has no converter for huggingface/transformers models "

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.dev202508250413"
+__version__ = "0.23.0.dev202508260411"
 
 
 @keras_hub_export("keras_hub.version")

keras_hub/tokenizers/__init__.py

@@ -88,6 +88,9 @@ from keras_hub.src.models.siglip.siglip_tokenizer import (
     SigLIPTokenizer as SigLIPTokenizer,
 )
 from keras_hub.src.models.t5.t5_tokenizer import T5Tokenizer as T5Tokenizer
+from keras_hub.src.models.t5gemma.t5gemma_tokenizer import (
+    T5GemmaTokenizer as T5GemmaTokenizer,
+)
 from keras_hub.src.models.whisper.whisper_tokenizer import (
     WhisperTokenizer as WhisperTokenizer,
 )

{keras_hub_nightly-0.23.0.dev202508250413.dist-info → keras_hub_nightly-0.23.0.dev202508260411.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: keras-hub-nightly
-Version: 0.23.0.dev202508250413
+Version: 0.23.0.dev202508260411
 Summary: Pretrained models for Keras.
 Author-email: Keras team <keras-users@googlegroups.com>
 License-Expression: Apache-2.0

{keras_hub_nightly-0.23.0.dev202508250413.dist-info → keras_hub_nightly-0.23.0.dev202508260411.dist-info}/RECORD

@@ -1,11 +1,11 @@
 keras_hub/__init__.py,sha256=bJbUZkqwhZvTb1Tqx1fbkq6mzBYiEyq-Hin3oQIkhdE,558
 keras_hub/layers/__init__.py,sha256=SMkchjCbNydCBULOFC1pzZRaD-KWZ2CaH6CEVf1MRWE,5428
 keras_hub/metrics/__init__.py,sha256=KYalsMPBnfwim9BdGHFfJ5WxUKFXOQ1QoKIMT_0lwlM,439
-keras_hub/models/__init__.py,sha256=UXMwKVZ7bg-AOrq2xsl8M0idUAS89pkdCvQKhzL-D3I,28439
+keras_hub/models/__init__.py,sha256=BIiZQysBWPgBaShlvPu2wGvDOeHZyJ5IEcaGmdO3KJA,28899
 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=hda9lLVTfIS7cJbaqRyEGv8ngWNeZ8dwdekqAtgdTnA,222
+keras_hub/src/version.py,sha256=g1tcd00bjQZuGngpSoTU2YlVzqJrCZYrV6wu_r4dQOA,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
@@ -437,6 +437,16 @@ keras_hub/src/models/t5/t5_preprocessor.py,sha256=UVOnCHUJF_MBcOyfR9G9oeRUEoN3Xo
 keras_hub/src/models/t5/t5_presets.py,sha256=I9rOBMG4dcBaSK3UHRcaJHUuVHeXsez60TYRqXZKL-A,3173
 keras_hub/src/models/t5/t5_tokenizer.py,sha256=pLTu15JeYSpVmy-2600vBc-Mxn_uHyTKts4PI2MxxBM,2517
 keras_hub/src/models/t5/t5_transformer_layer.py,sha256=uDeP84F1x7xJxki5iKe12Zn6eWD_4yVjoFXMuod-a3A,5347
+keras_hub/src/models/t5gemma/__init__.py,sha256=KFXci5c05Fi2xqmdn8wj0oz38GeCcIksaYo9bQWGOLA,263
+keras_hub/src/models/t5gemma/t5gemma_attention.py,sha256=j1BqTCCEfeRiwYOlwrBDrLoIPB5mRaCMRS9J5bWobWA,15232
+keras_hub/src/models/t5gemma/t5gemma_backbone.py,sha256=wV5UTSlHm9P5AhsK-Bnab_myyTaDEOtI-0h5Glhf_VA,16574
+keras_hub/src/models/t5gemma/t5gemma_decoder.py,sha256=BHzdk5akm7sVbEyL7e176YYeuT2gVtSW7ol41b0PdSM,14375
+keras_hub/src/models/t5gemma/t5gemma_encoder.py,sha256=KW5xZTVS9UgzoQspHwKcYkqKWYxob2wACZKQUv-zIC0,8675
+keras_hub/src/models/t5gemma/t5gemma_layers.py,sha256=19_CLs6_lYTqdQJQTlalI50VEI8F3buNgXWoBoIgjas,4381
+keras_hub/src/models/t5gemma/t5gemma_presets.py,sha256=tfAu9fOxrrvF6vf8_zOY8be2RKU-0BoQpshCupOqH80,532
+keras_hub/src/models/t5gemma/t5gemma_seq_2_seq_lm.py,sha256=-dRXqt1DbKQVUKqUqafBft2rJUB89tEj7NuRMlhX5og,17836
+keras_hub/src/models/t5gemma/t5gemma_seq_2_seq_lm_preprocessor.py,sha256=AXjmd0vOQ2J__E9GACeKWTosGrlkzcriC2OstQi0-x0,8186
+keras_hub/src/models/t5gemma/t5gemma_tokenizer.py,sha256=4EUX_kUEDqB6QAKSv2VxBVUVrF16TIBBX34Dir7f-70,2740
 keras_hub/src/models/vae/__init__.py,sha256=i3UaSW4IJf76O7lSPE1dyxOVjuHx8iAYKivqvUbDHOw,62
 keras_hub/src/models/vae/vae_backbone.py,sha256=Yk0srJhB-zfxQeAoyZdNzvxfxPxPMVie0nqKU7cp-2M,7033
 keras_hub/src/models/vae/vae_layers.py,sha256=NaPjITYTvMcEOrtCQZXxqtkbDzB7odtiNxJx-YpvYy8,27751
@@ -539,14 +549,15 @@ keras_hub/src/utils/transformers/convert_pali_gemma.py,sha256=B1leeDw96Yvu81hYum
 keras_hub/src/utils/transformers/convert_qwen.py,sha256=WUxMAEFVqRs7TRw7QU5TH3_ev4yf02R1xFVliMvTQqg,5886
 keras_hub/src/utils/transformers/convert_qwen3.py,sha256=LIormvCMWPq6X9Wo2eNbADjtFZ0nI7tFGZFBxmo4GKw,5700
 keras_hub/src/utils/transformers/convert_qwen_moe.py,sha256=a7R28aln-PdAcNuKAXdrtzvslho2Co6GypChxLMKPpc,10618
+keras_hub/src/utils/transformers/convert_t5gemma.py,sha256=DPOwd61UhjspKuCsk3_EaNvSADGP_f8KLcZARHYVk5Y,9490
 keras_hub/src/utils/transformers/convert_vit.py,sha256=9SUZ9utNJhW_5cj3acMn9cRy47u2eIcDsrhmzj77o9k,5187
-keras_hub/src/utils/transformers/preset_loader.py,sha256=YZVpeNhFITHdauY3MWESrZLNUIJt9ilHJ1jUhvITNT8,4781
+keras_hub/src/utils/transformers/preset_loader.py,sha256=JZn5mfKnVTN5aAvdZ6GWbS_CK3wP42iDkEJsmA58BVw,4925
 keras_hub/src/utils/transformers/safetensor_utils.py,sha256=CYUHyA4y-B61r7NDnCsFb4t_UmSwZ1k9L-8gzEd6KRg,3339
 keras_hub/src/utils/transformers/export/gemma.py,sha256=xX_vfQwvFZ_-lQX4kgMNOGKL7fL_1yk7QyGYV2Qyly4,4699
 keras_hub/src/utils/transformers/export/hf_exporter.py,sha256=Qk52c6LIA2eMHUNY9Vy4STJSpnhLMdJ_t-3ljqhSr4k,5081
-keras_hub/tokenizers/__init__.py,sha256=gQIESc4erRLuwxHyxtYy_Z0ePQXw_uhXAa4GVHMffYk,4244
+keras_hub/tokenizers/__init__.py,sha256=B-LdN3lKiXaqv9MV7JiVaKiVqs8SL_1sqSpdbg_qtJY,4349
 keras_hub/utils/__init__.py,sha256=jXPqVGBpJr_PpYmqD8aDG-fRMlxH-ulqCR2SZMn288Y,646
-keras_hub_nightly-0.23.0.dev202508250413.dist-info/METADATA,sha256=-mvp1ZJsgMdQ3PlKEf9_5htPCfHjyZyw0TfWwa3C-yY,7395
-keras_hub_nightly-0.23.0.dev202508250413.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-keras_hub_nightly-0.23.0.dev202508250413.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
-keras_hub_nightly-0.23.0.dev202508250413.dist-info/RECORD,,
+keras_hub_nightly-0.23.0.dev202508260411.dist-info/METADATA,sha256=KdNoOY6Bw-w1ZrGexPSltZfYkpESpzxpf9YfkLvENiY,7395
+keras_hub_nightly-0.23.0.dev202508260411.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+keras_hub_nightly-0.23.0.dev202508260411.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
+keras_hub_nightly-0.23.0.dev202508260411.dist-info/RECORD,,