keras-hub 0.24.0.dev0__py3-none-any.whl → 0.25.0__py3-none-any.whl

keras_hub/src/models/gpt_oss/gpt_oss_attention.py

@@ -0,0 +1,330 @@
+import math
+
+import keras
+from keras import ops
+
+from keras_hub.src.layers.modeling.rotary_embedding import RotaryEmbedding
+from keras_hub.src.utils.keras_utils import clone_initializer
+
+
+class GptOssAttention(keras.layers.Layer):
+    """A cached attention layer with sliding window and sink tokens.
+
+    This layer implements the attention mechanism described in the GPT-OSS
+    paper. It includes grouped-query attention, rotary position embeddings,
+    sliding window attention, and sink tokens for improved performance on
+    long sequences.
+
+    Args:
+        num_query_heads: int. The number of query attention heads.
+        num_key_value_heads: int. The number of key and value attention
+            heads.
+        rope_max_wavelength: int. The maximum wavelength for the
+            rotary position embedding. Defaults to 10000.
+        rope_scaling_factor: float. The scaling factor for the
+            rotary position embedding. Defaults to 1.0.
+        kernel_initializer: str. The initializer for the kernel
+            weights. Defaults to "glorot_uniform".
+        sliding_window: int. The size of the sliding window.
+            Defaults to 4096.
+        dropout: float. The dropout rate. Defaults to 0.
+        head_dim: int. Head dimension for attention. If None,
+            calculated as hidden_dim // num_query_heads. Defaults to None.
+    """
+
+    def __init__(
+        self,
+        num_query_heads,
+        num_key_value_heads,
+        rope_max_wavelength=10000,
+        rope_scaling_factor=1.0,
+        kernel_initializer="glorot_uniform",
+        sliding_window=4096,
+        dropout=0,
+        head_dim=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.num_query_heads = num_query_heads
+        self.num_key_value_heads = num_key_value_heads
+        self.sliding_window = sliding_window
+        self.dropout = dropout
+        self.head_dim = head_dim
+        self.rope_max_wavelength = rope_max_wavelength
+        self.rope_scaling_factor = rope_scaling_factor
+        self.num_key_value_groups = num_query_heads // num_key_value_heads
+        self._kernel_initializer = keras.initializers.get(
+            clone_initializer(kernel_initializer)
+        )
+
+    def build(self, inputs_shape):
+        # Einsum variables:
+        # b = batch size
+        # q = query length
+        # k = key/value length
+        # m = the model's hidden_dim
+        # u = num query heads
+        # v = num key/value heads
+        # h = head dim
+        self._hidden_dim = inputs_shape[-1]
+
+        if self.head_dim is not None:
+            self._head_dim = self.head_dim
+        else:
+            self._head_dim = self._hidden_dim // self.num_query_heads
+        self._inv_norm_factor = 1.0 / math.sqrt(self._head_dim)
+
+        self._rotary_dim = (self._head_dim // 2) * 2
+
+        self.query_dense = keras.layers.EinsumDense(
+            equation="bqm,muh->bquh",
+            output_shape=(None, self.num_query_heads, self._head_dim),
+            bias_axes="uh",
+            kernel_initializer=self._kernel_initializer,
+            bias_initializer="zeros",
+            dtype=self.dtype_policy,
+            name="query",
+        )
+        self.query_dense.build(inputs_shape)
+
+        self.key_dense = keras.layers.EinsumDense(
+            equation="bkm,mvh->bkvh",
+            output_shape=(
+                None,
+                self.num_key_value_heads,
+                self._head_dim,
+            ),
+            bias_axes="vh",
+            kernel_initializer=self._kernel_initializer,
+            bias_initializer="zeros",
+            dtype=self.dtype_policy,
+            name="key",
+        )
+        self.key_dense.build(inputs_shape)
+
+        self.value_dense = keras.layers.EinsumDense(
+            equation="bkm,mvh->bkvh",
+            output_shape=(
+                None,
+                self.num_key_value_heads,
+                self._head_dim,
+            ),
+            bias_axes="vh",
+            kernel_initializer=self._kernel_initializer,
+            bias_initializer="zeros",
+            dtype=self.dtype_policy,
+            name="value",
+        )
+        self.value_dense.build(inputs_shape)
+
+        self.dropout_layer = keras.layers.Dropout(
+            rate=self.dropout,
+            dtype=self.dtype_policy,
+        )
+
+        self.output_dense = keras.layers.EinsumDense(
+            equation="bquh,uhm->bqm",
+            output_shape=(None, self._hidden_dim),
+            bias_axes="m",
+            kernel_initializer=self._kernel_initializer,
+            bias_initializer="zeros",
+            dtype=self.dtype_policy,
+            name="attention_output",
+        )
+        self.output_dense.build(
+            (None, None, self.num_query_heads, self._head_dim)
+        )
+
+        self.rotary_embedding_layer = RotaryEmbedding(
+            max_wavelength=self.rope_max_wavelength,
+            scaling_factor=self.rope_scaling_factor,  # YaRN scaling factor
+            rope_type="yarn",
+            beta_fast=32.0,
+            beta_slow=1.0,
+            original_max_position_embeddings=4096,
+            dtype=self.dtype_policy,
+        )
+
+        self.sinks = self.add_weight(
+            shape=(self.num_query_heads,),
+            initializer="random_normal",
+            dtype=self.dtype,
+            name="sinks",
+        )
+
+        self._dot_product_equation = "bquh,bkuh->buqk"
+        self._combine_equation = "buqk,bkuh->bquh"
+
+        self.built = True
+
+    def call(
+        self,
+        hidden_states,
+        attention_mask=None,
+        cache=None,
+        cache_update_index=None,
+        training=None,
+    ):
+        start_index = (
+            cache_update_index if cache_update_index is not None else 0
+        )
+
+        query = self.query_dense(hidden_states)
+
+        # Compute RoPE for queries (only
+        #  to first _rotary_dim dimensions)
+        if self._rotary_dim < self._head_dim:
+            query_rot = query[..., : self._rotary_dim]
+            query_rot = self.rotary_embedding_layer(
+                query_rot, start_index=start_index
+            )
+            query = ops.concatenate(
+                [query_rot, query[..., self._rotary_dim :]], axis=-1
+            )
+        else:
+            query = self.rotary_embedding_layer(query, start_index=start_index)
+
+        def _compute_key_value(x):
+            key, value = self.key_dense(x), self.value_dense(x)
+            # Compute RoPE for keys (only apply to first _rotary_dim dimensions)
+            if self._rotary_dim < self._head_dim:
+                key_rot = key[..., : self._rotary_dim]
+                key_rot = self.rotary_embedding_layer(
+                    key_rot, start_index=start_index
+                )
+                key = ops.concatenate(
+                    [key_rot, key[..., self._rotary_dim :]], axis=-1
+                )
+            else:
+                key = self.rotary_embedding_layer(key, start_index=start_index)
+            return key, value
+
+        if cache is not None:
+            key_cache = cache[:, 0, ...]
+            value_cache = cache[:, 1, ...]
+            if cache_update_index is None:
+                key = key_cache
+                value = value_cache
+            else:
+                key_update, value_update = _compute_key_value(hidden_states)
+                start = [0, cache_update_index, 0, 0]
+                key = ops.slice_update(key_cache, start, key_update)
+                value = ops.slice_update(value_cache, start, value_update)
+                cache = ops.stack((key, value), axis=1)
+        else:
+            if cache_update_index is not None:
+                raise ValueError(
+                    "`cache_update_index` should not be set if `cache` is "
+                    f"`None`. Received: cache={cache}, "
+                    f"cache_update_index={cache_update_index}"
+                )
+            key, value = _compute_key_value(hidden_states)
+
+        # [batch_shape, seq_len, num_key_value_heads, head_dim]
+        # -> [batch_shape, seq_len, num_heads, head_dim]
+        key = ops.repeat(key, repeats=self.num_key_value_groups, axis=2)
+        value = ops.repeat(value, repeats=self.num_key_value_groups, axis=2)
+
+        attention_output = self._compute_attention(
+            query, key, value, attention_mask, start_index
+        )
+
+        attention_output = self.dropout_layer(
+            attention_output, training=training
+        )
+
+        attention_output = self.output_dense(attention_output)
+
+        if cache is not None:
+            return attention_output, cache
+        return attention_output
+
+    def _compute_attention(
+        self, query, key, value, attention_mask=None, start_index=0
+    ):
+        attention_scores = ops.einsum(self._dot_product_equation, query, key)
+        attention_scores = ops.multiply(
+            attention_scores,
+            ops.cast(self._inv_norm_factor, self.compute_dtype),
+        )
+
+        # Apply sliding window mask if specified
+        if self.sliding_window is not None and self.sliding_window > 0:
+            q_len = ops.shape(attention_scores)[-2]
+            kv_len = ops.shape(attention_scores)[-1]
+
+            # Query positions are offset by start_index during generation
+            q_positions = ops.arange(q_len) + start_index
+            kv_positions = ops.arange(kv_len)
+
+            # Mask true for positions outside sliding window
+            # For causal attention: mask if kv_pos < q_pos - sliding_window
+            mask = (
+                kv_positions[None, :]
+                >= q_positions[:, None] - self.sliding_window
+            )
+            if self.compute_dtype == "float32":
+                sliding_adder = ops.cast(-1e9, self.compute_dtype)
+            else:
+                sliding_adder = ops.cast(-1e4, self.compute_dtype)
+            attention_scores = ops.where(
+                mask[None, None, :, :], attention_scores, sliding_adder
+            )
+
+        if attention_mask is not None:
+            # The mask is a boolean tensor, True for positions to be masked.
+            # We add a large negative number to the masked positions.
+            # Use a large negative value for masking
+            if self.compute_dtype == "float32":
+                adder = ops.cast(-1e9, self.compute_dtype)
+            else:
+                adder = ops.cast(-1e4, self.compute_dtype)
+            attention_scores = ops.where(
+                attention_mask[:, None, :, :], attention_scores, adder
+            )
+
+        # Handle sink tokens by concatenating them to the logits.
+        b = ops.shape(attention_scores)[0]
+        q = ops.shape(attention_scores)[2]
+
+        sinks = ops.reshape(self.sinks, (1, self.num_query_heads, 1, 1))
+        sinks = ops.broadcast_to(sinks, (b, self.num_query_heads, q, 1))
+        # attention_scores shape: [b, num_heads, q, k]
+        # sinks shape: [b, num_heads, q, 1]
+        # We need to concatenate along the last dimension
+        combined_logits = ops.concatenate([attention_scores, sinks], axis=-1)
+
+        # Stabilize logits before softmax for numerical stability.
+        max_logits = ops.max(combined_logits, axis=-1, keepdims=True)
+        max_logits = ops.stop_gradient(max_logits)
+        combined_logits = combined_logits - max_logits
+
+        probs = ops.softmax(combined_logits, axis=-1)
+
+        # Remove the sink probabilities before computing the output.
+        attention_scores = probs[..., :-1]
+        attention_scores = ops.cast(attention_scores, self.compute_dtype)
+
+        attention_output = ops.einsum(
+            self._combine_equation, attention_scores, value
+        )
+
+        return attention_output
+
+    def get_config(self):
+        config = super().get_config()
+        config.update(
+            {
+                "num_query_heads": self.num_query_heads,
+                "num_key_value_heads": self.num_key_value_heads,
+                "rope_max_wavelength": self.rope_max_wavelength,
+                "rope_scaling_factor": self.rope_scaling_factor,
+                "kernel_initializer": keras.initializers.serialize(
+                    self._kernel_initializer
+                ),
+                "sliding_window": self.sliding_window,
+                "dropout": self.dropout,
+                "head_dim": self.head_dim,
+            }
+        )
+        return config

keras_hub/src/models/gpt_oss/gpt_oss_backbone.py

@@ -0,0 +1,221 @@
+import keras
+
+from keras_hub.src.api_export import keras_hub_export
+from keras_hub.src.layers.modeling.reversible_embedding import (
+    ReversibleEmbedding,
+)
+from keras_hub.src.models.backbone import Backbone
+from keras_hub.src.models.gpt_oss.gpt_oss_decoder import (
+    GptOssTransformerDecoder,
+)
+from keras_hub.src.models.gpt_oss.gpt_oss_layer_norm import (
+    GptOssLayerNormalization,
+)
+
+
+def _gpt_oss_kernel_initializer(stddev=0.02):
+    return keras.initializers.RandomNormal(stddev=stddev)
+
+
+@keras_hub_export("keras_hub.models.GptOssBackbone")
+class GptOssBackbone(Backbone):
+    """A GPT-style Transformer with a Mixture of Experts.
+
+    This network implements a GPT-style decoder network with Mixture of Expert
+    (MoE) layers, similar to the architecture described in
+    ["Mixtral of Experts"](https://arxiv.org/pdf/2401.04088) but with
+    customizations found in some open-source GPT models. It includes the
+    embedding lookups and transformer layers.
+
+    The default constructor gives a fully customizable, randomly initialized
+    GptOss model with any number of layers, heads, and embedding
+    dimensions. To load preset architectures and weights, use the `from_preset`
+    constructor.
+
+    Args:
+        vocabulary_size: int. The size of the token vocabulary.
+        num_layers: int. The number of transformer layers.
+        num_query_heads: int. The number of query attention heads for
+            each transformer.
+        hidden_dim: int. The size of the transformer encoding and pooling
+            layers.
+        intermediate_dim: int. The output dimension of the first Dense layer
+            in a three-layer feedforward network for each transformer.
+        num_key_value_heads: int. The number of key and value attention heads
+            for each transformer.
+        num_experts: int. The number of experts for the MoE layers.
+        top_k: int. The number of experts to use for each token.
+            Defaults to `2`.
+        rope_max_wavelength: int. The maximum angular wavelength of
+            the sine/cosine curves, for rotary embeddings. Defaults to `10000`.
+        rope_scaling_factor: float. The scaling factor for
+            calculation of roatary embedding. Defaults to `1.0`.
+        layer_norm_epsilon: float. Epsilon for the layer
+            normalization layers in the transformer decoder. Defaults to `1e-6`.
+        sliding_window: int. The sliding window for the attention
+            layers. This controls the maximum cache size for the attention
+            layers in each transformer decoder. Only `sliding_window` number
+            of tokens are saved in the cache and used to generate the next
+            token. Defaults to `4096`.
+        head_dim: int. Head dimension for attention layers. This
+            parameter is accepted for HuggingFace compatibility but ignored.
+            The head dimension is calculated dynamically as hidden_dim //
+            num_query_heads. Defaults to `None`.
+        dropout: float. Attention dropout probability.
+        dtype: string or `keras.mixed_precision.DTypePolicy`. The dtype to use
+            for model computations and weights. Note that some computations,
+            such as softmax and layer normalization, will always be done at
+            `float32` precision regardless of dtype.
+
+    Examples:
+
+    ```python
+    import numpy as np
+    import keras_hub
+
+    # Load a pretrained GptOss backbone from a preset.
+    model = keras_hub.models.GptOssBackbone.from_preset("gpt_oss_20b_en")
+
+    input_data = {
+        "token_ids": np.ones(shape=(1, 12), dtype="int32"),
+        "padding_mask": np.array(
+            [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0]], dtype="int32"
+        ),
+    }
+
+    model(input_data)
+
+    # Randomly initialized GptOss decoder with custom config.
+    model = keras_hub.models.GptOssBackbone(
+        vocabulary_size=10,
+        hidden_dim=512,
+        num_layers=2,
+        num_query_heads=32,
+        num_key_value_heads=8,
+        intermediate_dim=1024,
+        num_experts=4,
+        top_k=2,
+        sliding_window=256,
+        layer_norm_epsilon=1e-6,
+        dtype="float32"
+    )
+    model(input_data)
+    ```
+    """
+
+    def __init__(
+        self,
+        vocabulary_size,
+        num_layers,
+        num_query_heads,
+        hidden_dim,
+        intermediate_dim,
+        num_key_value_heads,
+        num_experts,
+        top_k=2,
+        rope_max_wavelength=10000,
+        rope_scaling_factor=1.0,
+        layer_norm_epsilon=1e-6,
+        sliding_window=4096,
+        head_dim=None,
+        dropout=0,
+        output_router_logits=False,
+        dtype=None,
+        **kwargs,
+    ):
+        # === Layers ===
+        self.token_embedding = ReversibleEmbedding(
+            input_dim=vocabulary_size,
+            output_dim=hidden_dim,
+            tie_weights=False,
+            embeddings_initializer=_gpt_oss_kernel_initializer(stddev=0.01),
+            dtype=dtype,
+            name="token_embedding",
+        )
+        self.transformer_layers = []
+        for i in range(num_layers):
+            layer = GptOssTransformerDecoder(
+                intermediate_dim=intermediate_dim,
+                num_query_heads=num_query_heads,
+                num_key_value_heads=num_key_value_heads,
+                num_experts=num_experts,
+                top_k=top_k,
+                output_router_logits=output_router_logits,
+                rope_max_wavelength=rope_max_wavelength,
+                rope_scaling_factor=rope_scaling_factor,
+                layer_norm_epsilon=layer_norm_epsilon,
+                kernel_initializer=_gpt_oss_kernel_initializer(stddev=0.02),
+                # GPT-OSS uses SW attention in every other layer
+                sliding_window=sliding_window if i % 2 == 1 else None,
+                dropout=dropout,
+                head_dim=head_dim,
+                dtype=dtype,
+                name=f"transformer_layer_{i}",
+            )
+            self.transformer_layers.append(layer)
+        self.layer_norm = GptOssLayerNormalization(
+            epsilon=layer_norm_epsilon,
+            dtype=dtype,
+            name="sequence_output_layernorm",
+        )
+
+        # === Functional Model ===
+        token_id_input = keras.Input(
+            shape=(None,), dtype="int32", name="token_ids"
+        )
+        padding_mask_input = keras.Input(
+            shape=(None,), dtype="int32", name="padding_mask"
+        )
+        x = self.token_embedding(token_id_input)
+        for transformer_layer in self.transformer_layers:
+            x = transformer_layer(x, decoder_padding_mask=padding_mask_input)
+        sequence_output = self.layer_norm(x)
+        super().__init__(
+            inputs={
+                "token_ids": token_id_input,
+                "padding_mask": padding_mask_input,
+            },
+            outputs=sequence_output,
+            dtype=dtype,
+            **kwargs,
+        )
+
+        # === Config ===
+        self.vocabulary_size = vocabulary_size
+        self.num_layers = num_layers
+        self.num_query_heads = num_query_heads
+        self.hidden_dim = hidden_dim
+        self.intermediate_dim = intermediate_dim
+        self.num_key_value_heads = num_key_value_heads
+        self.num_experts = num_experts
+        self.top_k = top_k
+        self.rope_max_wavelength = rope_max_wavelength
+        self.rope_scaling_factor = rope_scaling_factor
+        self.sliding_window = sliding_window
+        self.layer_norm_epsilon = layer_norm_epsilon
+        self.dropout = dropout
+        self.output_router_logits = output_router_logits
+        self.head_dim = head_dim
+
+    def get_config(self):
+        config = super().get_config()
+        config.update(
+            {
+                "vocabulary_size": self.vocabulary_size,
+                "num_layers": self.num_layers,
+                "num_query_heads": self.num_query_heads,
+                "hidden_dim": self.hidden_dim,
+                "intermediate_dim": self.intermediate_dim,
+                "num_experts": self.num_experts,
+                "top_k": self.top_k,
+                "rope_max_wavelength": self.rope_max_wavelength,
+                "rope_scaling_factor": self.rope_scaling_factor,
+                "num_key_value_heads": self.num_key_value_heads,
+                "sliding_window": self.sliding_window,
+                "layer_norm_epsilon": self.layer_norm_epsilon,
+                "dropout": self.dropout,
+                "output_router_logits": self.output_router_logits,
+                "head_dim": self.head_dim,
+            }
+        )
+        return config