keras-hub-nightly 0.16.1.dev202410200345__py3-none-any.whl → 0.19.0.dev202412070351__py3-none-any.whl

keras_hub/src/models/gemma/gemma_presets.py

@@ -6,9 +6,7 @@ backbone_presets = {
         "metadata": {
             "description": "2 billion parameter, 18-layer, base Gemma model.",
             "params": 2506172416,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma/keras/gemma_2b_en/2",
     },
@@ -18,9 +16,7 @@ backbone_presets = {
                 "2 billion parameter, 18-layer, instruction tuned Gemma model."
             ),
             "params": 2506172416,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma/keras/gemma_instruct_2b_en/2",
     },
@@ -31,9 +27,7 @@ backbone_presets = {
                 "The 1.1 update improves model quality."
             ),
             "params": 2506172416,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma/keras/gemma_1.1_instruct_2b_en/3",
     },
@@ -45,9 +39,7 @@ backbone_presets = {
                 "completion. The 1.1 update improves model quality."
             ),
             "params": 2506172416,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/codegemma/keras/code_gemma_1.1_2b_en/1",
     },
@@ -59,9 +51,7 @@ backbone_presets = {
                 "completion."
             ),
             "params": 2506172416,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/codegemma/keras/code_gemma_2b_en/1",
     },
@@ -69,9 +59,7 @@ backbone_presets = {
         "metadata": {
             "description": "7 billion parameter, 28-layer, base Gemma model.",
             "params": 8537680896,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma/keras/gemma_7b_en/2",
     },
@@ -81,9 +69,7 @@ backbone_presets = {
                 "7 billion parameter, 28-layer, instruction tuned Gemma model."
             ),
             "params": 8537680896,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma/keras/gemma_instruct_7b_en/2",
     },
@@ -94,9 +80,7 @@ backbone_presets = {
                 "The 1.1 update improves model quality."
             ),
             "params": 8537680896,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma/keras/gemma_1.1_instruct_7b_en/3",
     },
@@ -108,9 +92,7 @@ backbone_presets = {
                 "completion."
             ),
             "params": 8537680896,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/codegemma/keras/code_gemma_7b_en/1",
     },
@@ -122,9 +104,7 @@ backbone_presets = {
                 "to code."
             ),
             "params": 8537680896,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/codegemma/keras/code_gemma_instruct_7b_en/1",
     },
@@ -136,9 +116,7 @@ backbone_presets = {
                 "to code. The 1.1 update improves model quality."
             ),
             "params": 8537680896,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/codegemma/keras/code_gemma_1.1_instruct_7b_en/1",
     },
@@ -146,9 +124,7 @@ backbone_presets = {
         "metadata": {
             "description": "2 billion parameter, 26-layer, base Gemma model.",
             "params": 2614341888,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma2/keras/gemma2_2b_en/1",
     },
@@ -156,9 +132,7 @@ backbone_presets = {
         "metadata": {
             "description": "2 billion parameter, 26-layer, instruction tuned Gemma model.",
             "params": 2614341888,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma2/keras/gemma2_instruct_2b_en/1",
     },
@@ -166,9 +140,7 @@ backbone_presets = {
         "metadata": {
             "description": "9 billion parameter, 42-layer, base Gemma model.",
             "params": 9241705984,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma2/keras/gemma2_9b_en/2",
     },
@@ -176,9 +148,7 @@ backbone_presets = {
         "metadata": {
             "description": "9 billion parameter, 42-layer, instruction tuned Gemma model.",
             "params": 9241705984,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma2/keras/gemma2_instruct_9b_en/2",
     },
@@ -186,9 +156,7 @@ backbone_presets = {
         "metadata": {
             "description": "27 billion parameter, 42-layer, base Gemma model.",
             "params": 27227128320,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma2/keras/gemma2_27b_en/1",
     },
@@ -196,9 +164,7 @@ backbone_presets = {
         "metadata": {
             "description": "27 billion parameter, 42-layer, instruction tuned Gemma model.",
             "params": 27227128320,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://keras/gemma2/keras/gemma2_instruct_27b_en/1",
     },
@@ -206,9 +172,7 @@ backbone_presets = {
         "metadata": {
             "description": "2 billion parameter, 26-layer, ShieldGemma model.",
             "params": 2614341888,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://google/shieldgemma/keras/shieldgemma_2b_en/1",
     },
@@ -216,9 +180,7 @@ backbone_presets = {
         "metadata": {
             "description": "9 billion parameter, 42-layer, ShieldGemma model.",
             "params": 9241705984,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://google/shieldgemma/keras/shieldgemma_9b_en/1",
     },
@@ -226,9 +188,7 @@ backbone_presets = {
         "metadata": {
             "description": "27 billion parameter, 42-layer, ShieldGemma model.",
             "params": 27227128320,
-            "official_name": "Gemma",
             "path": "gemma",
-            "model_card": "https://www.kaggle.com/models/google/gemma",
         },
         "kaggle_handle": "kaggle://google/shieldgemma/keras/shieldgemma_27b_en/1",
     },

keras_hub/src/models/gpt2/gpt2_presets.py

@@ -9,9 +9,7 @@ backbone_presets = {
                 "Trained on WebText."
             ),
             "params": 124439808,
-            "official_name": "GPT-2",
             "path": "gpt2",
-            "model_card": "https://github.com/openai/gpt-2/blob/master/model_card.md",
         },
         "kaggle_handle": "kaggle://keras/gpt2/keras/gpt2_base_en/2",
     },
@@ -22,9 +20,7 @@ backbone_presets = {
                 "Trained on WebText."
             ),
             "params": 354823168,
-            "official_name": "GPT-2",
             "path": "gpt2",
-            "model_card": "https://github.com/openai/gpt-2/blob/master/model_card.md",
         },
         "kaggle_handle": "kaggle://keras/gpt2/keras/gpt2_medium_en/2",
     },
@@ -35,9 +31,7 @@ backbone_presets = {
                 "Trained on WebText."
             ),
             "params": 774030080,
-            "official_name": "GPT-2",
             "path": "gpt2",
-            "model_card": "https://github.com/openai/gpt-2/blob/master/model_card.md",
         },
         "kaggle_handle": "kaggle://keras/gpt2/keras/gpt2_large_en/2",
     },
@@ -48,9 +42,7 @@ backbone_presets = {
                 "Trained on WebText."
             ),
             "params": 1557611200,
-            "official_name": "GPT-2",
             "path": "gpt2",
-            "model_card": "https://github.com/openai/gpt-2/blob/master/model_card.md",
         },
         "kaggle_handle": "kaggle://keras/gpt2/keras/gpt2_extra_large_en/2",
     },
@@ -61,7 +53,6 @@ backbone_presets = {
                 "Finetuned on the CNN/DailyMail summarization dataset."
             ),
             "params": 124439808,
-            "official_name": "GPT-2",
             "path": "gpt2",
         },
         "kaggle_handle": "kaggle://keras/gpt2/keras/gpt2_base_en_cnn_dailymail/2",

keras_hub/src/models/image_object_detector.py

@@ -0,0 +1,87 @@
+import keras
+
+from keras_hub.src.api_export import keras_hub_export
+from keras_hub.src.models.task import Task
+
+
+@keras_hub_export("keras_hub.models.ImageObjectDetector")
+class ImageObjectDetector(Task):
+    """Base class for all image object detection tasks.
+
+    The `ImageObjectDetector` tasks wrap a `keras_hub.models.Backbone` and
+    a `keras_hub.models.Preprocessor` to create a model that can be used for
+    object detection. `ImageObjectDetector` tasks take an additional
+    `num_classes` argument, controlling the number of predicted output classes.
+
+    To fine-tune with `fit()`, pass a dataset containing tuples of `(x, y)`
+    labels where `x` is a string and `y` is dictionary with `boxes` and
+    `classes`.
+
+    All `ImageObjectDetector` tasks include a `from_preset()` constructor which
+    can be used to load a pre-trained config and weights.
+    """
+
+    def compile(
+        self,
+        optimizer="auto",
+        box_loss="auto",
+        classification_loss="auto",
+        metrics=None,
+        **kwargs,
+    ):
+        """Configures the `ImageObjectDetector` task for training.
+
+        The `ImageObjectDetector` task extends the default compilation signature of
+        `keras.Model.compile` with defaults for `optimizer`, `loss`, and
+        `metrics`. To override these defaults, pass any value
+        to these arguments during compilation.
+
+        Args:
+            optimizer: `"auto"`, an optimizer name, or a `keras.Optimizer`
+                instance. Defaults to `"auto"`, which uses the default optimizer
+                for the given model and task. See `keras.Model.compile` and
+                `keras.optimizers` for more info on possible `optimizer` values.
+            box_loss: `"auto"`, a loss name, or a `keras.losses.Loss` instance.
+                Defaults to `"auto"`, where a
+                `keras.losses.Huber` loss will be
+                applied for the object detector task. See
+                `keras.Model.compile` and `keras.losses` for more info on
+                possible `loss` values.
+            classification_loss: `"auto"`, a loss name, or a `keras.losses.Loss`
+                instance. Defaults to `"auto"`, where a
+                `keras.losses.BinaryFocalCrossentropy` loss will be
+                applied for the object detector task. See
+                `keras.Model.compile` and `keras.losses` for more info on
+                possible `loss` values.
+            metrics: `a list of metrics to be evaluated by
+                the model during training and testing. Defaults to `None`.
+                See `keras.Model.compile` and `keras.metrics` for
+                more info on possible `metrics` values.
+            **kwargs: See `keras.Model.compile` for a full list of arguments
+                supported by the compile method.
+        """
+        if optimizer == "auto":
+            optimizer = keras.optimizers.Adam(5e-5)
+        if box_loss == "auto":
+            box_loss = keras.losses.Huber(reduction="sum")
+        if classification_loss == "auto":
+            activation = getattr(self, "activation", None)
+            activation = keras.activations.get(activation)
+            from_logits = activation != keras.activations.sigmoid
+            classification_loss = keras.losses.BinaryFocalCrossentropy(
+                from_logits=from_logits, reduction="sum"
+            )
+        if metrics is not None:
+            raise ValueError("User metrics not yet supported")
+
+        losses = {
+            "bbox_regression": box_loss,
+            "cls_logits": classification_loss,
+        }
+
+        super().compile(
+            optimizer=optimizer,
+            loss=losses,
+            metrics=metrics,
+            **kwargs,
+        )

keras_hub/src/models/image_object_detector_preprocessor.py

@@ -0,0 +1,57 @@
+import keras
+
+from keras_hub.src.api_export import keras_hub_export
+from keras_hub.src.models.preprocessor import Preprocessor
+from keras_hub.src.utils.tensor_utils import preprocessing_function
+
+
+@keras_hub_export("keras_hub.models.ImageObjectDetectorPreprocessor")
+class ImageObjectDetectorPreprocessor(Preprocessor):
+    """Base class for object detector preprocessing layers.
+
+    `ImageObjectDetectorPreprocessor` tasks wraps a
+    `keras_hub.layers.Preprocessor` to create a preprocessing layer for
+    object detection tasks. It is intended to be paired with a
+    `keras_hub.models.ImageObjectDetector` task.
+
+    All `ImageObjectDetectorPreprocessor` take three inputs, `x`, `y`, and
+    `sample_weight`. `x`, the first input, should always be included. It can
+    be a image or batch of images. See examples below. `y` and `sample_weight`
+    are optional inputs that will be passed through unaltered. Usually, `y` will
+    be the a dict of `{"boxes": Tensor(batch_size, num_boxes, 4),
+    "classes": (batch_size, num_boxes)}.
+
+    The layer will returns either `x`, an `(x, y)` tuple if labels were provided,
+    or an `(x, y, sample_weight)` tuple if labels and sample weight were
+    provided. `x` will be the input images after all model preprocessing has
+    been applied.
+
+    All `ImageObjectDetectorPreprocessor` tasks include a `from_preset()`
+    constructor which can be used to load a pre-trained config and vocabularies.
+    You can call the `from_preset()` constructor directly on this base class, in
+    which case the correct class for your model will be automatically
+    instantiated.
+
+    Args:
+        image_converter: Preprocessing pipeline for images.
+
+    Examples.
+    ```python
+    preprocessor = keras_hub.models.ImageObjectDetectorPreprocessor.from_preset(
+        "retinanet_resnet50",
+    )
+    """
+
+    def __init__(
+        self,
+        image_converter=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.image_converter = image_converter
+
+    @preprocessing_function
+    def call(self, x, y=None, sample_weight=None):
+        if self.image_converter:
+            x = self.image_converter(x)
+        return keras.utils.pack_x_y_sample_weight(x, y, sample_weight)

keras_hub/src/models/image_to_image.py

@@ -234,7 +234,7 @@ class ImageToImage(Task):
                 input_is_scalar = True
             x = ops.image.resize(
                 x,
-                (self.backbone.height, self.backbone.width),
+                (self.backbone.image_shape[0], self.backbone.image_shape[1]),
                 interpolation="nearest",
                 data_format=data_format,
             )
@@ -284,8 +284,8 @@ class ImageToImage(Task):
         self,
         inputs,
         num_steps,
-        guidance_scale,
         strength,
+        guidance_scale=None,
         seed=None,
     ):
         """Generate image based on the provided `inputs`.
@@ -313,30 +313,36 @@ class ImageToImage(Task):
                 - A `tf.data.Dataset` with `"images"`, `"prompts"` and/or
                     `"negative_prompts"` keys.
             num_steps: int. The number of diffusion steps to take.
-            guidance_scale: float. The classifier free guidance scale defined in
-                [Classifier-Free Diffusion Guidance](
-                https://arxiv.org/abs/2207.12598). A higher scale encourages
-                generating images more closely related to the prompts, typically
-                at the cost of lower image quality.
             strength: float. Indicates the extent to which the reference
                 `images` are transformed. Must be between `0.0` and `1.0`. When
                 `strength=1.0`, `images` is essentially ignore and added noise
                 is maximum and the denoising process runs for the full number of
                 iterations specified in `num_steps`.
+            guidance_scale: Optional float. The classifier free guidance scale
+                defined in [Classifier-Free Diffusion Guidance](
+                https://arxiv.org/abs/2207.12598). A higher scale encourages
+                generating images more closely related to the prompts, typically
+                at the cost of lower image quality. Note that some models don't
+                utilize classifier-free guidance.
             seed: optional int. Used as a random seed.
         """
         num_steps = int(num_steps)
-        guidance_scale = float(guidance_scale)
         strength = float(strength)
+        guidance_scale = (
+            float(guidance_scale) if guidance_scale is not None else None
+        )
         if strength < 0.0 or strength > 1.0:
             raise ValueError(
                 "`strength` must be between `0.0` and `1.0`. "
                 f"Received strength={strength}."
             )
+        if guidance_scale is not None and guidance_scale > 1.0:
+            guidance_scale = ops.convert_to_tensor(float(guidance_scale))
+        else:
+            guidance_scale = None
         starting_step = int(num_steps * (1.0 - strength))
         starting_step = ops.convert_to_tensor(starting_step, "int32")
-        num_steps = ops.convert_to_tensor(num_steps, "int32")
-        guidance_scale = ops.convert_to_tensor(guidance_scale)
+        num_steps = ops.convert_to_tensor(int(num_steps), "int32")
 
         # Check `inputs` format.
         required_keys = ["images", "prompts"]

keras_hub/src/models/inpaint.py

@@ -202,7 +202,7 @@ class Inpaint(Task):
                 input_is_scalar = True
             x = ops.image.resize(
                 x,
-                (self.backbone.height, self.backbone.width),
+                (self.backbone.image_shape[0], self.backbone.image_shape[1]),
                 interpolation="nearest",
                 data_format=data_format,
             )
@@ -240,7 +240,7 @@ class Inpaint(Task):
                 x = ops.cast(x, "float32")
             x = ops.image.resize(
                 x,
-                (self.backbone.height, self.backbone.width),
+                (self.backbone.image_shape[0], self.backbone.image_shape[1]),
                 interpolation="nearest",
                 data_format=data_format,
             )
@@ -303,7 +303,7 @@ class Inpaint(Task):
                 input_is_scalar = True
             x = ops.image.resize(
                 x,
-                (self.backbone.height, self.backbone.width),
+                (self.backbone.image_shape[0], self.backbone.image_shape[1]),
                 interpolation="nearest",
                 data_format=data_format,
             )
@@ -323,7 +323,7 @@ class Inpaint(Task):
                 x = ops.cast(x, "float32")
             x = ops.image.resize(
                 x,
-                (self.backbone.height, self.backbone.width),
+                (self.backbone.image_shape[0], self.backbone.image_shape[1]),
                 interpolation="nearest",
                 data_format=data_format,
             )
@@ -376,16 +376,16 @@ class Inpaint(Task):
         self,
         inputs,
         num_steps,
-        guidance_scale,
         strength,
+        guidance_scale=None,
         seed=None,
     ):
         """Generate image based on the provided `inputs`.
 
         Typically, `inputs` is a dict with `"images"` `"masks"` and `"prompts"`
         keys. `"images"` are reference images within a value range of
-        `[-1.0, 1.0]`, which will be resized to `self.backbone.height` and
-        `self.backbone.width`, then encoded into latent space by the VAE
+        `[-1.0, 1.0]`, which will be resized to height and width from
+        `self.backbone.image_shape`, then encoded into latent space by the VAE
         encoder. `"masks"` are mask images with a boolean dtype, where white
         pixels are repainted while black pixels are preserved. `"prompts"` are
         strings that will be tokenized and encoded by the text encoder.
@@ -406,26 +406,33 @@ class Inpaint(Task):
                 - A `tf.data.Dataset` with `"images"`, `"masks"`, `"prompts"`
                     and/or `"negative_prompts"` keys.
             num_steps: int. The number of diffusion steps to take.
-            guidance_scale: float. The classifier free guidance scale defined in
-                [Classifier-Free Diffusion Guidance](
-                https://arxiv.org/abs/2207.12598). A higher scale encourages
-                generating images more closely related to the prompts, typically
-                at the cost of lower image quality.
             strength: float. Indicates the extent to which the reference
                 `images` are transformed. Must be between `0.0` and `1.0`. When
                 `strength=1.0`, `images` is essentially ignore and added noise
                 is maximum and the denoising process runs for the full number of
                 iterations specified in `num_steps`.
+            guidance_scale: Optional float. The classifier free guidance scale
+                defined in [Classifier-Free Diffusion Guidance](
+                https://arxiv.org/abs/2207.12598). A higher scale encourages
+                generating images more closely related to the prompts, typically
+                at the cost of lower image quality. Note that some models don't
+                utilize classifier-free guidance.
             seed: optional int. Used as a random seed.
         """
         num_steps = int(num_steps)
-        guidance_scale = float(guidance_scale)
         strength = float(strength)
+        guidance_scale = (
+            float(guidance_scale) if guidance_scale is not None else None
+        )
         if strength < 0.0 or strength > 1.0:
             raise ValueError(
                 "`strength` must be between `0.0` and `1.0`. "
                 f"Received strength={strength}."
             )
+        if guidance_scale is not None and guidance_scale > 1.0:
+            guidance_scale = ops.convert_to_tensor(guidance_scale)
+        else:
+            guidance_scale = None
         starting_step = int(num_steps * (1.0 - strength))
         starting_step = ops.convert_to_tensor(starting_step, "int32")
         num_steps = ops.convert_to_tensor(num_steps, "int32")

keras_hub/src/models/llama/llama_backbone.py

@@ -59,7 +59,7 @@ class LlamaBackbone(Backbone):
     }
 
     # Pretrained Llama decoder.
-    model = keras_hub.models.LlamaBackbone.from_preset("llama7b_base_en")
+    model = keras_hub.models.LlamaBackbone.from_preset("llama2_7b_en")
     model(input_data)
 
     # Randomly initialized Llama decoder with custom config.

keras_hub/src/models/llama/llama_presets.py

@@ -6,9 +6,7 @@ backbone_presets = {
         "metadata": {
             "description": "7 billion parameter, 32-layer, base LLaMA 2 model.",
             "params": 6738415616,
-            "official_name": "LLaMA 2",
-            "path": "llama2",
-            "model_card": "https://github.com/meta-llama/llama",
+            "path": "llama",
         },
         "kaggle_handle": "kaggle://keras/llama2/keras/llama2_7b_en/1",
     },
@@ -19,9 +17,7 @@ backbone_presets = {
                 "activation and weights quantized to int8."
             ),
             "params": 6739839488,
-            "official_name": "LLaMA 2",
-            "path": "llama2",
-            "model_card": "https://github.com/meta-llama/llama",
+            "path": "llama",
         },
         "kaggle_handle": "kaggle://keras/llama2/keras/llama2_7b_en_int8/1",
     },
@@ -32,9 +28,7 @@ backbone_presets = {
                 "model."
             ),
             "params": 6738415616,
-            "official_name": "LLaMA 2",
-            "path": "llama2",
-            "model_card": "https://github.com/meta-llama/llama",
+            "path": "llama",
         },
         "kaggle_handle": "kaggle://keras/llama2/keras/llama2_instruct_7b_en/1",
     },
@@ -45,9 +39,7 @@ backbone_presets = {
                 "model with activation and weights quantized to int8."
             ),
             "params": 6739839488,
-            "official_name": "LLaMA 2",
-            "path": "llama2",
-            "model_card": "https://github.com/meta-llama/llama",
+            "path": "llama",
         },
         "kaggle_handle": "kaggle://keras/llama2/keras/llama2_instruct_7b_en_int8/1",
     },
@@ -58,9 +50,7 @@ backbone_presets = {
                 "model."
             ),
             "params": 6738415616,
-            "official_name": "Vicuna",
-            "path": "vicuna",
-            "model_card": "https://github.com/lm-sys/FastChat",
+            "path": "llama",
         },
         "kaggle_handle": "kaggle://keras/vicuna/keras/vicuna_1.5_7b_en/1",
     },

keras_hub/src/models/llama3/llama3_presets.py

@@ -6,9 +6,7 @@ backbone_presets = {
         "metadata": {
             "description": "8 billion parameter, 32-layer, base LLaMA 3 model.",
             "params": 8030261248,
-            "official_name": "LLaMA 3",
             "path": "llama3",
-            "model_card": "https://github.com/meta-llama/llama3",
         },
         "kaggle_handle": "kaggle://keras/llama3/keras/llama3_8b_en/3",
     },
@@ -19,9 +17,7 @@ backbone_presets = {
                 "activation and weights quantized to int8."
             ),
             "params": 8031894016,
-            "official_name": "LLaMA 3",
             "path": "llama3",
-            "model_card": "https://github.com/meta-llama/llama3",
         },
         "kaggle_handle": "kaggle://keras/llama3/keras/llama3_8b_en_int8/1",
     },
@@ -32,9 +28,7 @@ backbone_presets = {
                 "model."
             ),
             "params": 8030261248,
-            "official_name": "LLaMA 3",
             "path": "llama3",
-            "model_card": "https://github.com/meta-llama/llama3",
         },
         "kaggle_handle": "kaggle://keras/llama3/keras/llama3_instruct_8b_en/3",
     },
@@ -45,9 +39,7 @@ backbone_presets = {
                 "model with activation and weights quantized to int8."
             ),
             "params": 8031894016,
-            "official_name": "LLaMA 3",
             "path": "llama3",
-            "model_card": "https://github.com/meta-llama/llama3",
         },
         "kaggle_handle": (
             "kaggle://keras/llama3/keras/llama3_instruct_8b_en_int8/1"