keras-hub 0.21.0.dev0__py3-none-any.whl → 0.21.1.dev0__py3-none-any.whl

keras_hub/src/models/mixtral/__init__.py

@@ -0,0 +1,5 @@
+from keras_hub.src.models.mixtral.mixtral_backbone import MixtralBackbone
+from keras_hub.src.models.mixtral.mixtral_presets import backbone_presets
+from keras_hub.src.utils.preset_utils import register_presets
+
+register_presets(backbone_presets, MixtralBackbone)

keras_hub/src/models/mixtral/mixtral_presets.py

@@ -4,8 +4,8 @@ backbone_presets = {
     "mixtral_8_7b_en": {
         "metadata": {
             "description": (
-                "32-layer Mixtral MoE model with 7 billion",
-                "active parameters and 8 experts per MoE layer.",
+                "32-layer Mixtral MoE model with 7 billion"
+                "active parameters and 8 experts per MoE layer."
             ),
             "params": 46702792704,
             "path": "mixtral",
@@ -15,8 +15,8 @@ backbone_presets = {
     "mixtral_8_instruct_7b_en": {
         "metadata": {
             "description": (
-                "Instruction fine-tuned 32-layer Mixtral MoE model",
-                "with 7 billion active parameters and 8 experts per MoE layer.",
+                "Instruction fine-tuned 32-layer Mixtral MoE model"
+                "with 7 billion active parameters and 8 experts per MoE layer."
             ),
             "params": 46702792704,
             "path": "mixtral",

keras_hub/src/models/moonshine/moonshine_audio_to_text_preprocessor.py

@@ -266,4 +266,7 @@ class MoonshineAudioToTextPreprocessor(AudioToTextPreprocessor):
                 and 0 <= token < vocab_size
             ]
             processed_sequences.append(filtered_tokens)
+        processed_sequences = tf.ragged.constant(
+            processed_sequences, dtype=tf.int32
+        )
         return self.tokenizer.detokenize(processed_sequences)

keras_hub/src/models/qwen/__init__.py

@@ -1 +1,5 @@
 from keras_hub.src.models.qwen.qwen_backbone import QwenBackbone
+from keras_hub.src.models.qwen.qwen_presets import backbone_presets
+from keras_hub.src.utils.preset_utils import register_presets
+
+register_presets(backbone_presets, QwenBackbone)

keras_hub/src/models/qwen/qwen_causal_lm.py

@@ -17,6 +17,130 @@ from keras_hub.src.utils.tensor_utils import any_equal
     ]
 )
 class QwenCausalLM(CausalLM):
+    """An end-to-end Qwen model for causal language modeling.
+
+    A causal language model (LM) predicts the next token based on previous
+    tokens. This task setup can be used to train the model unsupervised on plain
+    text input, or to autoregressively generate plain text similar to the data
+    used for training. This task can be used for pre-training or fine-tuning a
+    Qwen model, simply by calling `fit()`.
+
+    This model has a `generate()` method, which generates text based on a
+    prompt. The generation strategy used is controlled by an additional
+    `sampler` argument on `compile()`. You can recompile the model with
+    different `keras_hub.samplers` objects to control the generation.
+    By default, `"greedy"` sampling will be used.
+
+    This model can optionally be configured with a `preprocessor` layer, in
+    which case it will automatically apply preprocessing to string inputs during
+    `fit()`, `predict()`, `evaluate()`, and `generate()`. This is done by
+    default when creating the model with `from_preset()`.
+
+    Args:
+        backbone: A `keras_hub.models.QwenBackbone` instance.
+        preprocessor: A `keras_hub.models.QwenCausalLMPreprocessor` or
+            `None`. If `None`, this model will not apply preprocessing, and
+            inputs should be preprocessed before calling the model.
+
+    Examples:
+
+    Use `generate()` to do text generation.
+    ```python
+    qwen_lm = keras_hub.models.QwenCausalLM.from_preset("qwen2.5_0.5b_en")
+    qwen_lm.generate("I want to say", max_length=30)
+
+    # Generate with batched prompts.
+    qwen_lm.generate(["This is a", "Where are you"], max_length=30)
+    ```
+
+    Compile the `generate()` function with a custom sampler.
+    ```python
+    qwen_lm = keras_hub.models.QwenMoeCausalLM.from_preset("qwen2.5_0.5b_en")
+    qwen_lm.compile(sampler="top_k")
+    qwen_lm.generate("I want to say", max_length=30)
+
+    qwen_lm.compile(sampler=keras_hub.samplers.BeamSampler(num_beams=2))
+    qwen_lm.generate("I want to say", max_length=30)
+    ```
+
+    Use `generate()` without preprocessing.
+    ```python
+    prompt = {
+        # Token ids for "<bos> Qwen is".
+        "token_ids": np.array([[2, 12345, 678, 0, 0, 0, 0]] * 2),
+        # Use `"padding_mask"` to indicate values that should not be overridden.
+        "padding_mask": np.array([[1, 1, 1, 0, 0, 0, 0]] * 2),
+    }
+
+    qwen_lm = keras_hub.models.QwenMoeCausalLM.from_preset(
+        "qwen2.5_0.5b_en",
+        preprocessor=None,
+    )
+    qwen_lm.generate(prompt)
+    ```
+
+    Call `fit()` on a single batch.
+    ```python
+    features = ["The quick brown fox jumped.", "I forgot my homework."]
+    qwen_lm = keras_hub.models.QwenMoeCausalLM.from_preset("qwen2.5_0.5b_en")
+    qwen_lm.fit(x=features, batch_size=2)
+    ```
+
+    Call `fit()` with LoRA fine-tuning enabled.
+    ```python
+    features = ["The quick brown fox jumped.", "I forgot my homework."]
+    qwen_lm = keras_hub.models.QwenMoeCausalLM.from_preset("qwen2.5_0.5b_en")
+    qwen_lm.backbone.enable_lora(rank=4)
+    qwen_lm.fit(x=features, batch_size=2)
+    ```
+
+    Call `fit()` without preprocessing.
+    ```python
+    x = {
+        # Token ids for "<bos> Qwen is a language model<eos>"
+        "token_ids": np.array([[2, 12345, 678, 543, 9876, 1, 0, 0]] * 2),
+        "padding_mask": np.array([[1, 1, 1, 1, 1, 1, 0, 0]] * 2),
+    }
+    y = np.array([[12345, 678, 543, 9876, 1, 0, 0, 0]] * 2)
+    sw = np.array([[1, 1, 1, 1, 1, 0, 0, 0]] * 2)
+
+    qwen_lm = keras_hub.models.QwenMoeCausalLM.from_preset(
+        "qwen2.5_0.5b_en",
+        preprocessor=None,
+    )
+    qwen_lm.fit(x=x, y=y, sample_weight=sw, batch_size=2)
+    ```
+
+    Custom backbone and vocabulary.
+    ```python
+    tokenizer = keras_hub.models.QwenMoeTokenizer(
+        proto="qwen_moe_vocab.spm",
+    )
+    preprocessor = keras_hub.models.QwenMoeCausalLMPreprocessor(
+        tokenizer=tokenizer,
+        sequence_length=128,
+    )
+    backbone = keras_hub.models.QwenMoeBackbone(
+        vocabulary_size=151936,
+        num_layers=28,
+        num_query_heads=16,
+        num_key_value_heads=8,
+        hidden_dim=2048,
+        intermediate_dim=4096,
+        moe_intermediate_dim=128,
+        shared_expert_intermediate_dim=4096,
+        num_experts=60,
+        top_k=4,
+        max_sequence_length=4096,
+    )
+    qwen_lm = keras_hub.models.QwenMoeCausalLM(
+        backbone=backbone,
+        preprocessor=preprocessor,
+    )
+    qwen_lm.fit(x=features, batch_size=2)
+    ```
+    """
+
     backbone_cls = QwenBackbone
     preprocessor_cls = QwenCausalLMPreprocessor
 

keras_hub/src/models/qwen/qwen_causal_lm_preprocessor.py

@@ -11,6 +11,72 @@ from keras_hub.src.models.qwen.qwen_tokenizer import QwenTokenizer
     ]
 )
 class QwenCausalLMPreprocessor(CausalLMPreprocessor):
+    """Qwen Causal LM preprocessor.
+
+    This preprocessing layer is meant for use with
+    `keras_hub.models.QwenCausalLM`. By default, it will take in batches of
+    strings, and return outputs in a `(x, y, sample_weight)` format, where the
+    `y` label is the next token id in the `x` sequence.
+
+    For use with generation, the layer also exposes two methods
+    `generate_preprocess()` and `generate_postprocess()`. When this preprocessor
+    is attached to a `keras_hub.models.QwenCausalLM` instance, these methods
+    will be called implicitly in `generate()`. They can also be called
+    standalone (e.g. to precompute preprocessing inputs for generation in a
+    separate process).
+
+    Args:
+        tokenizer: A `keras_hub.models.QwenTokenizer` instance.
+        sequence_length: The length of the packed inputs.
+        add_start_token: If `True`, the preprocessor will prepend the tokenizer
+            start token to each input sequence. Default is `True`.
+        add_end_token: If `True`, the preprocessor will append the tokenizer
+            end token to each input sequence. Default is `False`.
+
+    Call arguments:
+        x: A string, `tf.Tensor` or list of python strings.
+        y: Label data. Should always be `None` as the layer generates labels.
+        sample_weight: Label weights. Should always be `None` as the layer
+            generates label weights.
+        sequence_length: Pass to override the configured `sequence_length` of
+            the layer.
+
+    Examples:
+    ```python
+    # Load the preprocessor from a preset.
+    preprocessor = keras_hub.models.QwenCausalLMPreprocessor.from_preset(
+        "qwen2.5_0.5b_en"
+    )
+
+    # Tokenize and pack a single sentence.
+    sentence = tf.constant("League of legends")
+    preprocessor(sentence)
+    # Same output.
+    preprocessor("League of legends")
+
+    # Tokenize a batch of sentences.
+    sentences = tf.constant(["Taco tuesday", "Fish taco please!"])
+    preprocessor(sentences)
+    # Same output.
+    preprocessor(["Taco tuesday", "Fish taco please!"])
+
+    # Map a dataset to preprocess a single sentence.
+    features = tf.constant(
+        [
+            "Avatar 2 is amazing!",
+            "Well, I am not sure.",
+        ]
+    )
+    labels = tf.constant([1, 0])
+    ds = tf.data.Dataset.from_tensor_slices((features, labels))
+    ds = ds.map(preprocessor, num_parallel_calls=tf.data.AUTOTUNE)
+
+    # Map a dataset to preprocess unlabled sentences.
+    ds = tf.data.Dataset.from_tensor_slices(features)
+    ds = ds.map(preprocessor, num_parallel_calls=tf.data.AUTOTUNE)
+    ```
+    """
+
     backbone_cls = QwenBackbone
     tokenizer_cls = QwenTokenizer
 

keras_hub/src/models/qwen/qwen_presets.py

@@ -28,8 +28,8 @@ backbone_presets = {
     "qwen2.5_instruct_0.5b_en": {
         "metadata": {
             "description": (
-                "Instruction fine-tuned 24-layer Qwen model with 0.5 ",
-                "billion parameters.",
+                "Instruction fine-tuned 24-layer Qwen model with 0.5 "
+                "billion parameters."
             ),
             "params": 494032768,
             "path": "qwen",
@@ -39,8 +39,8 @@ backbone_presets = {
     "qwen2.5_instruct_32b_en": {
         "metadata": {
             "description": (
-                "Instruction fine-tuned 64-layer Qwen model with 32 ",
-                "billion parameters.",
+                "Instruction fine-tuned 64-layer Qwen model with 32 "
+                "billion parameters."
             ),
             "params": 32763876352,
             "path": "qwen",
@@ -50,8 +50,8 @@ backbone_presets = {
     "qwen2.5_instruct_72b_en": {
         "metadata": {
             "description": (
-                "Instruction fine-tuned 80-layer Qwen model with 72 ",
-                "billion parameters.",
+                "Instruction fine-tuned 80-layer Qwen model with 72 "
+                "billion parameters."
             ),
             "params": 72706203648,
             "path": "qwen",

keras_hub/src/models/qwen_moe/__init__.py

@@ -0,0 +1,5 @@
+from keras_hub.src.models.qwen_moe.qwen_moe_backbone import QwenMoeBackbone
+from keras_hub.src.models.qwen_moe.qwen_moe_presets import backbone_presets
+from keras_hub.src.utils.preset_utils import register_presets
+
+register_presets(backbone_presets, QwenMoeBackbone)

keras_hub/src/models/qwen_moe/qwen_moe_causal_lm_preprocessor.py

@@ -4,12 +4,74 @@ from keras_hub.src.models.qwen_moe.qwen_moe_backbone import QwenMoeBackbone
 from keras_hub.src.models.qwen_moe.qwen_moe_tokenizer import QwenMoeTokenizer
 
 
-@keras_hub_export(
-    [
-        "keras_hub.models.QwenMoeCausalLMPreprocessor",
-    ]
-)
+@keras_hub_export("keras_hub.models.QwenMoeCausalLMPreprocessor")
 class QwenMoeCausalLMPreprocessor(CausalLMPreprocessor):
+    """Qwen-Moe Causal LM preprocessor.
+
+    This preprocessing layer is meant for use with
+    `keras_hub.models.QwenMoeCausalLM`. By default, it will take in batches of
+    strings, and return outputs in a `(x, y, sample_weight)` format, where the
+    `y` label is the next token id in the `x` sequence.
+
+    For use with generation, the layer also exposes two methods
+    `generate_preprocess()` and `generate_postprocess()`. When this preprocessor
+    is attached to a `keras_hub.models.QwenMoeCausalLM` instance, these methods
+    will be called implicitly in `generate()`. They can also be called
+    standalone (e.g. to precompute preprocessing inputs for generation in a
+    separate process).
+
+    Args:
+        tokenizer: A `keras_hub.models.QwenMoeTokenizer` instance.
+        sequence_length: The length of the packed inputs.
+        add_start_token: If `True`, the preprocessor will prepend the tokenizer
+            start token to each input sequence. Default is `True`.
+        add_end_token: If `True`, the preprocessor will append the tokenizer
+            end token to each input sequence. Default is `False`.
+
+    Call arguments:
+        x: A string, `tf.Tensor` or list of python strings.
+        y: Label data. Should always be `None` as the layer generates labels.
+        sample_weight: Label weights. Should always be `None` as the layer
+            generates label weights.
+        sequence_length: Pass to override the configured `sequence_length` of
+            the layer.
+
+    Examples:
+    ```python
+    # Load the preprocessor from a preset.
+    preprocessor = keras_hub.models.QwenMoeCausalLMPreprocessor.from_preset(
+        "qwen2.5_0.5b_en"
+    )
+
+    # Tokenize and pack a single sentence.
+    sentence = tf.constant("League of legends")
+    preprocessor(sentence)
+    # Same output.
+    preprocessor("League of legends")
+
+    # Tokenize a batch of sentences.
+    sentences = tf.constant(["Taco tuesday", "Fish taco please!"])
+    preprocessor(sentences)
+    # Same output.
+    preprocessor(["Taco tuesday", "Fish taco please!"])
+
+    # Map a dataset to preprocess a single sentence.
+    features = tf.constant(
+        [
+            "Avatar 2 is amazing!",
+            "Well, I am not sure.",
+        ]
+    )
+    labels = tf.constant([1, 0])
+    ds = tf.data.Dataset.from_tensor_slices((features, labels))
+    ds = ds.map(preprocessor, num_parallel_calls=tf.data.AUTOTUNE)
+
+    # Map a dataset to preprocess unlabled sentences.
+    ds = tf.data.Dataset.from_tensor_slices(features)
+    ds = ds.map(preprocessor, num_parallel_calls=tf.data.AUTOTUNE)
+    ```
+    """
+
     backbone_cls = QwenMoeBackbone
     tokenizer_cls = QwenMoeTokenizer
 

keras_hub/src/models/qwen_moe/qwen_moe_presets.py

@@ -4,8 +4,8 @@ backbone_presets = {
     "qwen1.5_moe_2.7b_en": {
         "metadata": {
             "description": (
-                "24-layer Qwen MoE model with 2.7 billion active parameters ",
-                "and 8 experts per MoE layer.",
+                "24-layer Qwen MoE model with 2.7 billion active parameters "
+                "and 8 experts per MoE layer."
             ),
             "params": 14315784192,
             "path": "qwen-1.5-moe",

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__ = "0.21.0.dev0"
+__version__ = "0.21.1.dev0"
 
 
 @keras_hub_export("keras_hub.version")

{keras_hub-0.21.0.dev0.dist-info → keras_hub-0.21.1.dev0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: keras-hub
-Version: 0.21.0.dev0
+Version: 0.21.1.dev0
 Summary: Pretrained models for Keras.
 Author-email: Keras team <keras-users@googlegroups.com>
 License-Expression: Apache-2.0

{keras_hub-0.21.0.dev0.dist-info → keras_hub-0.21.1.dev0.dist-info}/RECORD

@@ -5,7 +5,7 @@ keras_hub/models/__init__.py,sha256=itSzodVUeuX6HQnmsSXY0Wv-5Htbu397410R-SFW_4I,
 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=ac8MCBjym07GRYtzUbXheZDZ7VsOHeic5uu3xKGLmM8,211
+keras_hub/src/version.py,sha256=9-4pakvS-20qqoJLPAxs1MpQcXpHuTo0hZidVMHPOG0,211
 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
@@ -252,13 +252,14 @@ keras_hub/src/models/mit/mit_image_classifier_preprocessor.py,sha256=oNYs-pUK8Vn
 keras_hub/src/models/mit/mit_image_converter.py,sha256=Mw7nV-OzyBveGuZUNFsPPKyq9jXJVW2_cVH024CNkXM,311
 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/mixtral/__init__.py,sha256=Bo-r5hI-ptOoUmz2HN9h9oZBN-C-XeWEL19QblUN0Cs,263
 keras_hub/src/models/mixtral/mixtral_attention.py,sha256=f5aiTtstWeKG_ZwumAlYIzjIN08CpnxNdenxWNJSwZw,8713
 keras_hub/src/models/mixtral/mixtral_backbone.py,sha256=vUAFXvqwVBgKxYbOsqIHzPN59bhaDrGWwOnBCzeUtt0,8034
 keras_hub/src/models/mixtral/mixtral_causal_lm.py,sha256=JA1t6xTeaYX_fNo9ftRyvzdRDG3vndC-Rlwn5fnsbQo,12001
 keras_hub/src/models/mixtral/mixtral_causal_lm_preprocessor.py,sha256=q2qXa9QAUWBvOWv9DeNvwsBNXSORJAbQFoQsWQ7e8V8,3079
 keras_hub/src/models/mixtral/mixtral_decoder.py,sha256=CvOjhTxPnGQ_HNknZXRI6Cx1kpuHG99_TiOh-mNcsDw,18190
 keras_hub/src/models/mixtral/mixtral_layer_norm.py,sha256=zfbDKZEb45FTwP0zQd7WPPp8tuiGoSNfS-DRYWkZyWw,1031
-keras_hub/src/models/mixtral/mixtral_presets.py,sha256=AteLrYXyVjooz_DHLnBA1OMlZS6LMu7Y7gGUWddn6go,856
+keras_hub/src/models/mixtral/mixtral_presets.py,sha256=pi5hHcwVSqr7ytf4dSnU_ew_t7NYw7EsZrmklQDqDVo,852
 keras_hub/src/models/mixtral/mixtral_tokenizer.py,sha256=Kc233k879QMyX164X_CzWbqpnqEkKWNqa648guTGkBk,661
 keras_hub/src/models/mobilenet/__init__.py,sha256=hxkNGGj_iAMu62iooUDEPA818sNOIgjG7pXMLEMOsAE,275
 keras_hub/src/models/mobilenet/mobilenet_backbone.py,sha256=aZBSFeLUObYYoi3od9DI1KfgPCqh5GHTcAI8Y2ZHShA,29536
@@ -270,7 +271,7 @@ keras_hub/src/models/mobilenet/util.py,sha256=S7j4UacmVIJ3fU8cymyAoK49eHcpWIKTOy
 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
-keras_hub/src/models/moonshine/moonshine_audio_to_text_preprocessor.py,sha256=hTw941ww8cJrP5DRrxv2DtZUNLJ9A3cayFhnsG5Ef4g,10016
+keras_hub/src/models/moonshine/moonshine_audio_to_text_preprocessor.py,sha256=Syv_rF81hmN63fnkC1ALiXZbpEK8aoFHOiNvELF3iw0,10124
 keras_hub/src/models/moonshine/moonshine_backbone.py,sha256=XtRUBe_VusXsFRk7-t1JNXM0lxp2UBOJk9v7gfTNDhA,19623
 keras_hub/src/models/moonshine/moonshine_decoder.py,sha256=Exf5Gg1gsCBST53wxOgBetKkhjS8E8QIUIlUwHlOkIY,11816
 keras_hub/src/models/moonshine/moonshine_encoder.py,sha256=NjjMO_FEBlWFSv6Appv8a3V7XovW2afvxxjXwQRgV60,8148
@@ -303,23 +304,23 @@ keras_hub/src/models/phi3/phi3_layernorm.py,sha256=Oqu81tGd97Lzx3kG1QEtZ0S6gbfn3
 keras_hub/src/models/phi3/phi3_presets.py,sha256=sb2ce7Gq1OikFEf2KIYG69rFKHYKj8qhlN-Ea8d6J7k,1366
 keras_hub/src/models/phi3/phi3_rotary_embedding.py,sha256=wqiRn8nETNcLc5Vsm_d_8s11Ro6ibWZbWvODdLqIOo4,5013
 keras_hub/src/models/phi3/phi3_tokenizer.py,sha256=bOPH14wTVVHJHq8mgzXLjsgvKMNhfO8eayevAPpjYVA,1992
-keras_hub/src/models/qwen/__init__.py,sha256=hskG3tZUY_AYZPp0WVzbCtw37AIYENyp3DOnqHmdRBw,65
+keras_hub/src/models/qwen/__init__.py,sha256=C2NncZC-NIg29cQwbHUsrLsA6JAFXtco3hwP7VFgy9M,245
 keras_hub/src/models/qwen/qwen_attention.py,sha256=SrUYESCg27ksuDKZHKJ5Wmnkbr6WZdF7nHv0AHFfWR8,13014
 keras_hub/src/models/qwen/qwen_backbone.py,sha256=i39_LoKu6hcYWV6KFh2OzUDaXjV7g1WLNGF2-JD_tqI,13015
-keras_hub/src/models/qwen/qwen_causal_lm.py,sha256=_f-UHaKHp0ncxknpkpEJiW3jlng3E4CmddjQfz2QzJo,12249
-keras_hub/src/models/qwen/qwen_causal_lm_preprocessor.py,sha256=Va-4TLJD3ycEnkS41rF3dVj4_6K0j-gxLTrREFRcyr0,609
+keras_hub/src/models/qwen/qwen_causal_lm.py,sha256=2ugwuJsoxO8WpKX8iuWQPwc7SOAks7YHNvLeaJfLqG4,16722
+keras_hub/src/models/qwen/qwen_causal_lm_preprocessor.py,sha256=Va6LPa0GoLSxPripWOngw6jVNI-nQSFMdaja_BcfWmY,3195
 keras_hub/src/models/qwen/qwen_decoder.py,sha256=utmAvZlU7_nP-6pjGPDinK4JaMzsQSwOARG0ote-jAg,11771
 keras_hub/src/models/qwen/qwen_layernorm.py,sha256=DS35r3qd6g5ocL7Nhf_vNzLLMo1aI9VCSmL64dgNOYI,924
-keras_hub/src/models/qwen/qwen_presets.py,sha256=DpRplWNwktM4KDgIP495PTUBJxQE_mS6KQSK5LGWOyc,1998
+keras_hub/src/models/qwen/qwen_presets.py,sha256=1FkKV6M3yqJz4EP1xa7bEvfIQ721xXT-_ikjWX0xvww,1992
 keras_hub/src/models/qwen/qwen_tokenizer.py,sha256=LCv3IyiDDHqVnM9N3lf5-BE3iwicIh0nKS1hjoPw9lE,1532
-keras_hub/src/models/qwen_moe/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+keras_hub/src/models/qwen_moe/__init__.py,sha256=5D8GUmVDsJs0J4sVZHcXOLkZf12U96l-WtwyVee4lu8,267
 keras_hub/src/models/qwen_moe/qwen_moe_attention.py,sha256=pE79_iHUm2LGkoWL6zMJw_pNfzIvmyq3yJaiq47W2TY,13242
 keras_hub/src/models/qwen_moe/qwen_moe_backbone.py,sha256=nrfELvIvRLmrgKrUNXci2CrecmeI6bWzJj7HH-RcWJA,15341
 keras_hub/src/models/qwen_moe/qwen_moe_causal_lm.py,sha256=MeP60v7GcN_SmH5_ULRpqgmFVgaYAosSecZiSQVlJvU,13256
-keras_hub/src/models/qwen_moe/qwen_moe_causal_lm_preprocessor.py,sha256=uKaXRrJs02vkVudjdehzJPp0B84tPMkxNHlp166kceE,589
+keras_hub/src/models/qwen_moe/qwen_moe_causal_lm_preprocessor.py,sha256=9P6TT7W_fqf4HsXcmlHF-DW_anR-XoDrRN2ZFGA7Ai4,3168
 keras_hub/src/models/qwen_moe/qwen_moe_decoder.py,sha256=kmUjLpYTbJQ3J_31qWhLOd0Dg2_9cl_JX_zM8ZMH1Qo,23130
 keras_hub/src/models/qwen_moe/qwen_moe_layernorm.py,sha256=DbkWJo7U0-cwdZwHPeAnFznYwtao6o0fjpoDJ9UWnpc,927
-keras_hub/src/models/qwen_moe/qwen_moe_presets.py,sha256=uKrA9xLV3P3jtYUUsqdhKq_HPkB4lXmOYseB1wXTZnI,457
+keras_hub/src/models/qwen_moe/qwen_moe_presets.py,sha256=LhOA3Ow-z3cNTan4AOrtyCXS58EgfvO_gtqiZt5cUQc,455
 keras_hub/src/models/qwen_moe/qwen_moe_tokenizer.py,sha256=2c3X8jNGO0q0UL5NtUqSgHWLqhyJGi2ohNcTeOGhd84,1407
 keras_hub/src/models/resnet/__init__.py,sha256=C5UqlQ6apm8WSp1bnrxB6Bi3BGaknxRQs-r3b2wpaGA,257
 keras_hub/src/models/resnet/resnet_backbone.py,sha256=Q7nlqcTXZzjqd0e-DsjHC4ok58yOX7qxseotym3uZpM,31276
@@ -501,7 +502,7 @@ keras_hub/src/utils/transformers/preset_loader.py,sha256=1nfS5xVsl-JROGXJXltTqV1
 keras_hub/src/utils/transformers/safetensor_utils.py,sha256=CYUHyA4y-B61r7NDnCsFb4t_UmSwZ1k9L-8gzEd6KRg,3339
 keras_hub/tokenizers/__init__.py,sha256=uMjjm0mzUkRb0e4Ac_JK8aJ9cKGUi5UqmzWoWAFJprE,4164
 keras_hub/utils/__init__.py,sha256=jXPqVGBpJr_PpYmqD8aDG-fRMlxH-ulqCR2SZMn288Y,646
-keras_hub-0.21.0.dev0.dist-info/METADATA,sha256=5v0xL50HM0QeIox1_e2ZHoT0ZlHk3YigEM8NjvOaycI,7374
-keras_hub-0.21.0.dev0.dist-info/WHEEL,sha256=zaaOINJESkSfm_4HQVc5ssNzHCPXhJm0kEUakpsEHaU,91
-keras_hub-0.21.0.dev0.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
-keras_hub-0.21.0.dev0.dist-info/RECORD,,
+keras_hub-0.21.1.dev0.dist-info/METADATA,sha256=hPQjOgZ3VecWAQ8VAKLvr8CyBwQA4zW53rlD-xJEGjM,7374
+keras_hub-0.21.1.dev0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+keras_hub-0.21.1.dev0.dist-info/top_level.txt,sha256=N4J6piIWBKa38A4uV-CnIopnOEf8mHAbkNXafXm_CuA,10
+keras_hub-0.21.1.dev0.dist-info/RECORD,,

{keras_hub-0.21.0.dev0.dist-info → keras_hub-0.21.1.dev0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.8.0)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any