Add Falcon CausalLM.

keras_nlp/src/models/falcon/falcon_causal_lm.py

@@ -0,0 +1,287 @@
+# Copyright 2024 The KerasNLP Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from keras_nlp.src.api_export import keras_nlp_export
+from keras_nlp.src.backend import ops
+from keras_nlp.src.models.causal_lm import CausalLM
+from keras_nlp.src.models.falcon.falcon_backbone import FalconBackbone
+from keras_nlp.src.models.falcon.falcon_causal_lm_preprocessor import (
+    FalconCausalLMPreprocessor,
+)
+
+
+@keras_nlp_export("keras_nlp.models.FalconCausalLM")
+class FalconCausalLM(CausalLM):
+    """An end-to-end Falcon 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 Falcon 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_nlp.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_nlp.models.FalconBackbone` instance.
+        preprocessor: A `keras_nlp.models.FalconCausalLMPreprocessor` 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
+    falcon_lm = keras_nlp.models.FalconCausalLM.from_preset("falcon_refinedweb_1b_en")
+    falcon_lm.generate("I want to say", max_length=30)
+
+    # Generate with batched prompts.
+    falcon_lm.generate(["This is a", "Where are you"], max_length=30)
+    ```
+
+    Compile the `generate()` function with a custom sampler.
+    ```python
+    falcon_lm = keras_nlp.models.FalconCausalLM.from_preset("falcon_refinedweb_1b_en")
+    falcon_lm.compile(sampler="top_k")
+    falcon_lm.generate("I want to say", max_length=30)
+
+    falcon_lm.compile(sampler=keras_nlp.samplers.BeamSampler(num_beams=2))
+    falcon_lm.generate("I want to say", max_length=30)
+    ```
+
+    Use `generate()` without preprocessing.
+    ```python
+    prompt = {
+        # Token ids for "<bos> Keras is".
+        "token_ids": np.array([[2, 214064, 603, 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),
+    }
+
+    falcon_lm = keras_nlp.models.FalconCausalLM.from_preset(
+        "falcon_refinedweb_1b_en",
+        preprocessor=None,
+    )
+    falcon_lm.generate(prompt)
+    ```
+
+    Call `fit()` on a single batch.
+    ```python
+    features = ["The quick brown fox jumped.", "I forgot my homework."]
+    falcon_lm = keras_nlp.models.FalconCausalLM.from_preset("falcon_refinedweb_1b_en")
+    falcon_lm.fit(x=features, batch_size=2)
+    ```
+
+    Call `fit()` without preprocessing.
+    ```python
+    x = {
+        # Token ids for "<bos> Keras is deep learning library<eos>"
+        "token_ids": np.array([[2, 214064, 603, 5271, 6044, 9581, 1, 0]] * 2),
+        "padding_mask": np.array([[1, 1, 1, 1, 1, 1, 1, 0]] * 2),
+    }
+    y = np.array([[214064, 603, 5271, 6044, 9581, 3, 0, 0]] * 2)
+    sw = np.array([[1, 1, 1, 1, 1, 1, 0, 0]] * 2)
+
+    falcon_lm = keras_nlp.models.FalconCausalLM.from_preset(
+        "falcon_refinedweb_1b_en",
+        preprocessor=None,
+    )
+    falcon_lm.fit(x=x, y=y, sample_weight=sw, batch_size=2)
+    ```
+
+    Custom backbone and vocabulary.
+    ```python
+    tokenizer = keras_nlp.models.FalconTokenizer(
+        proto="proto.spm",
+    )
+    preprocessor = keras_nlp.models.FalconCausalLMPreprocessor(
+        tokenizer=tokenizer,
+        sequence_length=128,
+    )
+    backbone = keras_nlp.models.FalconBackbone(
+        vocabulary_size=30552,
+        num_layers=4,
+        num_heads=4,
+        hidden_dim=256,
+        intermediate_dim=512,
+        max_sequence_length=128,
+    )
+    falcon_lm = keras_nlp.models.FalconCausalLM(
+        backbone=backbone,
+        preprocessor=preprocessor,
+    )
+    falcon_lm.fit(x=features, batch_size=2)
+    ```
+    """
+
+    backbone_cls = FalconBackbone
+    preprocessor_cls = FalconCausalLMPreprocessor
+
+    def __init__(
+        self,
+        backbone,
+        preprocessor=None,
+        **kwargs,
+    ):
+        # === Layers ===
+        self.backbone = backbone
+        self.preprocessor = preprocessor
+
+        # === Functional Model ===
+        inputs = backbone.input
+        hidden_states = backbone(inputs)
+        outputs = backbone.token_embedding(hidden_states, reverse=True)
+        super().__init__(
+            inputs=inputs,
+            outputs=outputs,
+            **kwargs,
+        )
+
+    def call_with_cache(
+        self,
+        token_ids,
+        cache,
+        cache_update_index,
+    ):
+        """Forward pass of `FalconCausalLM` with cache.
+
+        `call_with_cache` adds an additional forward pass for the model for
+        autoregressive inference. Unlike calling the model directly, this method
+        allows caching previous key/value Tensors in multi-head attention layer,
+        and avoids recomputing the outputs of seen tokens.
+
+        Args:
+            token_ids: a dense int Tensor with shape `(batch_size, max_length)`.
+            cache: a dense float Tensor, the cache of key and value.
+            cache_update_index: int, or int Tensor. The index of current inputs in the
+                whole sequence.
+
+        Returns:
+            A (logits, hidden_states, cache) tuple. Where `logits` is the
+            language model logits for the input token_ids, `hidden_states` is
+            the final hidden representation of the input tokens, and `cache` is
+            the decoding cache.
+        """
+        x = self.backbone.token_embedding(token_ids)
+        # Each decoder layer has a cache; we update them separately.
+        caches = []
+        for i, transformer_layer in enumerate(self.backbone.transformer_layers):
+            current_cache = cache[:, i, ...]
+            x, next_cache = transformer_layer(
+                x,
+                attention_cache=current_cache,
+                attention_cache_update_index=cache_update_index,
+            )
+            caches.append(next_cache)
+        cache = ops.stack(caches, axis=1)
+        hidden_states = x = self.backbone.final_layernorm(x)
+        logits = self.backbone.token_embedding(x, reverse=True)
+        return logits, hidden_states, cache
+
+    def _build_cache(self, token_ids):
+        """Build an empty cache for use with `call_with_cache()`."""
+        batch_size = ops.shape(token_ids)[0]
+        max_length = ops.shape(token_ids)[1]
+        num_layers = self.backbone.num_layers
+        num_heads = self.backbone.num_attention_heads
+        head_dim = self.backbone.hidden_dim // self.backbone.num_attention_heads
+        shape = [batch_size, num_layers, 2, max_length, num_heads, head_dim]
+        cache = ops.zeros(shape, dtype=self.compute_dtype)
+        # Seed the cache.
+        _, hidden_states, cache = self.call_with_cache(token_ids, cache, 0)
+        return hidden_states, cache
+
+    def generate_step(
+        self,
+        inputs,
+        stop_token_ids=None,
+    ):
+        """A compilable generation function for a single batch of inputs.
+
+        This function represents the inner, XLA-compilable, generation function
+        for a single batch of inputs. Inputs should have the same structure as
+        model inputs, a dictionary with keys `"token_ids"` and `"padding_mask"`.
+
+        Args:
+            inputs: A dictionary with two keys `"token_ids"` and
+                `"padding_mask"` and batched tensor values.
+            stop_token_ids: Tuple of id's of end token's to stop on. If all
+                sequences have produced a new stop token, generation
+                will stop.
+        """
+        token_ids, padding_mask = inputs["token_ids"], inputs["padding_mask"]
+        # Create and seed cache with a single forward pass.
+        hidden_states, cache = self._build_cache(token_ids)
+        # Compute the lengths of all user inputted tokens ids.
+        row_lengths = ops.sum(ops.cast(padding_mask, "int32"), axis=-1)
+        # Start at the first index that has no user inputted id.
+        index = ops.min(row_lengths)
+
+        def next(prompt, cache, index):
+            # The cache index is the index of our previous token.
+            cache_update_index = index - 1
+            batch_size = ops.shape(prompt)[0]
+            prompt = ops.slice(prompt, [0, cache_update_index], [batch_size, 1])
+            logits, hidden_states, cache = self.call_with_cache(
+                prompt,
+                cache,
+                cache_update_index,
+            )
+            return (
+                ops.squeeze(logits, axis=1),
+                ops.squeeze(hidden_states, axis=1),
+                cache,
+            )
+
+        token_ids = self.sampler(
+            next=next,
+            prompt=token_ids,
+            cache=cache,
+            index=index,
+            mask=padding_mask,
+            stop_token_ids=stop_token_ids,
+            hidden_states=hidden_states,
+            model=self,
+        )
+
+        # Compute an output padding mask with the token ids we updated.
+        if stop_token_ids is not None:
+            # Build a mask of `end_token_id` locations not in the original
+            # prompt (not in locations where `padding_mask` is True).
+            end_locations = ops.logical_and(
+                ops.equal(token_ids, stop_token_ids),
+                ops.logical_not(padding_mask),
+            )
+            end_locations = ops.cast(end_locations, "int32")
+            # Use cumsum to get ones in all locations after end_locations.
+            cumsum = ops.cast(ops.cumsum(end_locations, axis=-1), "int32")
+            overflow = cumsum - end_locations
+            # Our padding mask is the inverse of these overflow locations.
+            padding_mask = ops.logical_not(ops.cast(overflow, "bool"))
+        else:
+            # Without early stopping, all locations will have been updated.
+            padding_mask = ops.ones_like(token_ids, dtype="bool")
+        return {
+            "token_ids": token_ids,
+            "padding_mask": padding_mask,
+        }