diff --git a/keras_nlp/models/__init__.py b/keras_nlp/models/__init__.py
index bdd6042538..21d8394066 100644
--- a/keras_nlp/models/__init__.py
+++ b/keras_nlp/models/__init__.py
@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from keras_nlp.models.albert.albert_preprocessor import AlbertPreprocessor
+from keras_nlp.models.albert.albert_tokenizer import AlbertTokenizer
 from keras_nlp.models.bert.bert_backbone import BertBackbone
 from keras_nlp.models.bert.bert_classifier import BertClassifier
 from keras_nlp.models.bert.bert_preprocessor import BertPreprocessor
diff --git a/keras_nlp/models/albert/__init__.py b/keras_nlp/models/albert/__init__.py
new file mode 100644
index 0000000000..6e4df4e727
--- /dev/null
+++ b/keras_nlp/models/albert/__init__.py
@@ -0,0 +1,13 @@
+# Copyright 2022 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.
diff --git a/keras_nlp/models/albert/albert_preprocessor.py b/keras_nlp/models/albert/albert_preprocessor.py
new file mode 100644
index 0000000000..61d37c34d2
--- /dev/null
+++ b/keras_nlp/models/albert/albert_preprocessor.py
@@ -0,0 +1,201 @@
+# Copyright 2022 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.
+"""ALBERT preprocessor layer."""
+
+from tensorflow import keras
+
+from keras_nlp.layers.multi_segment_packer import MultiSegmentPacker
+from keras_nlp.utils.keras_utils import (
+    convert_inputs_to_list_of_tensor_segments,
+)
+from keras_nlp.utils.keras_utils import pack_x_y_sample_weight
+from keras_nlp.utils.python_utils import classproperty
+
+
+@keras.utils.register_keras_serializable(package="keras_nlp")
+class AlbertPreprocessor(keras.layers.Layer):
+    """An ALBERT preprocessing layer which tokenizes and packs inputs.
+
+    This preprocessing layer will do three things:
+
+     - Tokenize any number of input segments using the `tokenizer`.
+     - Pack the inputs together using a `keras_nlp.layers.MultiSegmentPacker`.
+       with the appropriate `"[CLS]"`, `"[SEP]"` and `"<pad>"` tokens.
+     - Construct a dictionary with keys `"token_ids"`, `"segment_ids"` and
+       `"padding_mask"`, that can be passed directly to
+       `keras_nlp.models.AlbertBackbone`.
+
+    This layer can be used directly with `tf.data.Dataset.map` to preprocess
+    string data in the `(x, y, sample_weight)` format used by
+    `keras.Model.fit`.
+
+    The call method of this layer accepts three arguments, `x`, `y`, and
+    `sample_weight`. `x` can be a python string or tensor representing a single
+    segment, a list of python strings representing a batch of single segments,
+    or a list of tensors representing multiple segments to be packed together.
+    `y` and `sample_weight` are both optional, can have any format, and will be
+    passed through unaltered.
+
+    Special care should be taken when using `tf.data` to map over an unlabeled
+    tuple of string segments. `tf.data.Dataset.map` will unpack this tuple
+    directly into the call arguments of this layer, rather than forward all
+    argument to `x`. To handle this case, it is recommended to  explicitly call
+    the layer, e.g. `ds.map(lambda seg1, seg2: preprocessor(x=(seg1, seg2)))`.
+
+    Args:
+        tokenizer: A `keras_nlp.models.AlbertTokenizer` instance.
+        sequence_length: The length of the packed inputs.
+        truncate: string. The algorithm to truncate a list of batched segments
+            to fit within `sequence_length`. The value can be either
+            `round_robin` or `waterfall`:
+                - `"round_robin"`: Available space is assigned one token at a
+                    time in a round-robin fashion to the inputs that still need
+                    some, until the limit is reached.
+                - `"waterfall"`: The allocation of the budget is done using a
+                    "waterfall" algorithm that allocates quota in a
+                    left-to-right manner and fills up the buckets until we run
+                    out of budget. It supports an arbitrary number of segments.
+
+    Examples:
+    ```python
+    tokenizer = keras_nlp.models.AlbertTokenizer(proto="model.spm")
+    preprocessor = keras_nlp.models.AlbertPreprocessor(
+        tokenizer=tokenizer,
+        sequence_length=10,
+    )
+
+    # Tokenize and pack a single sentence.
+    sentence = tf.constant("The quick brown fox jumped.")
+    preprocessor(sentence)
+    # Same output.
+    preprocessor("The quick brown fox jumped.")
+
+    # Tokenize and a batch of single sentences.
+    sentences = tf.constant(
+        ["The quick brown fox jumped.", "Call me Ishmael."]
+    )
+    preprocessor(sentences)
+    # Same output.
+    preprocessor(
+        ["The quick brown fox jumped.", "Call me Ishmael."]
+    )
+
+    # Tokenize and pack a sentence pair.
+    first_sentence = tf.constant("The quick brown fox jumped.")
+    second_sentence = tf.constant("The fox tripped.")
+    preprocessor((first_sentence, second_sentence))
+
+    # Map a dataset to preprocess a single sentence.
+    features = tf.constant(
+        ["The quick brown fox jumped.", "Call me Ishmael."]
+    )
+    labels = tf.constant([0, 1])
+    ds = tf.data.Dataset.from_tensor_slices((features, labels))
+    ds = ds.map(preprocessor, num_parallel_calls=tf.data.AUTOTUNE)
+
+    # Map a dataset to preprocess sentence pairs.
+    first_sentences = tf.constant(
+        ["The quick brown fox jumped.", "Call me Ishmael."]
+    )
+    second_sentences = tf.constant(
+        ["The fox tripped.", "Oh look, a whale."]
+    )
+    labels = tf.constant([1, 1])
+    ds = tf.data.Dataset.from_tensor_slices(
+        (
+            (first_sentences, second_sentences), labels
+        )
+    )
+    ds = ds.map(preprocessor, num_parallel_calls=tf.data.AUTOTUNE)
+
+    # Map a dataset to preprocess unlabeled sentence pairs.
+    first_sentences = tf.constant(
+        ["The quick brown fox jumped.", "Call me Ishmael."]
+    )
+    second_sentences = tf.constant(
+        ["The fox tripped.", "Oh look, a whale."]
+    )
+    ds = tf.data.Dataset.from_tensor_slices((first_sentences, second_sentences))
+    # Watch out for tf.data's default unpacking of tuples here!
+    # Best to invoke the `preprocessor` directly in this case.
+    ds = ds.map(
+        lambda s1, s2: preprocessor(x=(s1, s2)),
+        num_parallel_calls=tf.data.AUTOTUNE,
+    )
+    ```
+    """
+
+    def __init__(
+        self,
+        tokenizer,
+        sequence_length=512,
+        truncate="round_robin",
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self._tokenizer = tokenizer
+        self.packer = MultiSegmentPacker(
+            start_value=self.tokenizer.cls_token_id,
+            end_value=self.tokenizer.sep_token_id,
+            pad_value=self.tokenizer.pad_token_id,
+            truncate=truncate,
+            sequence_length=sequence_length,
+        )
+
+    @property
+    def tokenizer(self):
+        """The `keras_nlp.models.AlbertTokenizer` used to tokenize strings."""
+        return self._tokenizer
+
+    def get_config(self):
+        config = super().get_config()
+        config.update(
+            {
+                "tokenizer": keras.layers.serialize(self.tokenizer),
+                "sequence_length": self.packer.sequence_length,
+                "truncate": self.packer.truncate,
+            }
+        )
+        return config
+
+    @classmethod
+    def from_config(cls, config):
+        if "tokenizer" in config and isinstance(config["tokenizer"], dict):
+            config["tokenizer"] = keras.layers.deserialize(config["tokenizer"])
+        return cls(**config)
+
+    def call(self, x, y=None, sample_weight=None):
+        x = convert_inputs_to_list_of_tensor_segments(x)
+        x = [self.tokenizer(segment) for segment in x]
+        token_ids, segment_ids = self.packer(x)
+        x = {
+            "token_ids": token_ids,
+            "segment_ids": segment_ids,
+            "padding_mask": token_ids != self.tokenizer.pad_token_id,
+        }
+        return pack_x_y_sample_weight(x, y, sample_weight)
+
+    @classproperty
+    def presets(cls):
+        return {}
+
+    @classmethod
+    def from_preset(
+        cls,
+        preset,
+        sequence_length=None,
+        truncate="round_robin",
+        **kwargs,
+    ):
+        raise NotImplementedError
diff --git a/keras_nlp/models/albert/albert_preprocessor_test.py b/keras_nlp/models/albert/albert_preprocessor_test.py
new file mode 100644
index 0000000000..01267a6edc
--- /dev/null
+++ b/keras_nlp/models/albert/albert_preprocessor_test.py
@@ -0,0 +1,174 @@
+# Copyright 2022 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.
+
+"""Tests for ALBERT preprocessor layer."""
+
+import io
+import os
+
+import sentencepiece
+import tensorflow as tf
+from absl.testing import parameterized
+from tensorflow import keras
+
+from keras_nlp.models.albert.albert_preprocessor import AlbertPreprocessor
+from keras_nlp.models.albert.albert_tokenizer import AlbertTokenizer
+
+
+class AlbertPreprocessorTest(tf.test.TestCase, parameterized.TestCase):
+    def setUp(self):
+        bytes_io = io.BytesIO()
+        vocab_data = tf.data.Dataset.from_tensor_slices(
+            ["the quick brown fox", "the earth is round"]
+        )
+        sentencepiece.SentencePieceTrainer.train(
+            sentence_iterator=vocab_data.as_numpy_iterator(),
+            model_writer=bytes_io,
+            vocab_size=10,
+            model_type="WORD",
+            pad_id=0,
+            unk_id=1,
+            bos_id=2,
+            eos_id=3,
+            pad_piece="<pad>",
+            unk_piece="<unk>",
+            bos_piece="[CLS]",
+            eos_piece="[SEP]",
+        )
+        self.proto = bytes_io.getvalue()
+
+        self.preprocessor = AlbertPreprocessor(
+            tokenizer=AlbertTokenizer(proto=self.proto),
+            sequence_length=12,
+        )
+
+    def test_tokenize_strings(self):
+        input_data = "the quick brown fox"
+        output = self.preprocessor(input_data)
+        self.assertAllEqual(
+            output["token_ids"], [2, 4, 9, 5, 7, 3, 0, 0, 0, 0, 0, 0]
+        )
+        self.assertAllEqual(
+            output["segment_ids"], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
+        )
+        self.assertAllEqual(
+            output["padding_mask"], [1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0]
+        )
+
+    def test_tokenize_list_of_strings(self):
+        # We should handle a list of strings as as batch.
+        input_data = ["the quick brown fox"] * 4
+        output = self.preprocessor(input_data)
+        self.assertAllEqual(
+            output["token_ids"],
+            [[2, 4, 9, 5, 7, 3, 0, 0, 0, 0, 0, 0]] * 4,
+        )
+        self.assertAllEqual(
+            output["segment_ids"], [[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]] * 4
+        )
+        self.assertAllEqual(
+            output["padding_mask"], [[1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0]] * 4
+        )
+
+    def test_tokenize_labeled_batch(self):
+        x = tf.constant(["the quick brown fox"] * 4)
+        y = tf.constant([1] * 4)
+        sw = tf.constant([1.0] * 4)
+        x_out, y_out, sw_out = self.preprocessor(x, y, sw)
+        self.assertAllEqual(
+            x_out["token_ids"],
+            [[2, 4, 9, 5, 7, 3, 0, 0, 0, 0, 0, 0]] * 4,
+        )
+        self.assertAllEqual(
+            x_out["segment_ids"], [[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]] * 4
+        )
+        self.assertAllEqual(
+            x_out["padding_mask"], [[1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0]] * 4
+        )
+        self.assertAllEqual(y_out, y)
+        self.assertAllEqual(sw_out, sw)
+
+    def test_tokenize_labeled_dataset(self):
+        x = tf.constant(["the quick brown fox"] * 4)
+        y = tf.constant([1] * 4)
+        sw = tf.constant([1.0] * 4)
+        ds = tf.data.Dataset.from_tensor_slices((x, y, sw))
+        ds = ds.map(self.preprocessor)
+        x_out, y_out, sw_out = ds.batch(4).take(1).get_single_element()
+        self.assertAllEqual(
+            x_out["token_ids"],
+            [[2, 4, 9, 5, 7, 3, 0, 0, 0, 0, 0, 0]] * 4,
+        )
+        self.assertAllEqual(
+            x_out["segment_ids"], [[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]] * 4
+        )
+        self.assertAllEqual(
+            x_out["padding_mask"], [[1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0]] * 4
+        )
+        self.assertAllEqual(y_out, y)
+        self.assertAllEqual(sw_out, sw)
+
+    def test_tokenize_multiple_sentences(self):
+        sentence_one = tf.constant("the quick brown fox")
+        sentence_two = tf.constant("the earth")
+        output = self.preprocessor((sentence_one, sentence_two))
+        self.assertAllEqual(
+            output["token_ids"],
+            [2, 4, 9, 5, 7, 3, 4, 6, 3, 0, 0, 0],
+        )
+        self.assertAllEqual(
+            output["segment_ids"], [0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0]
+        )
+        self.assertAllEqual(
+            output["padding_mask"], [1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0]
+        )
+
+    def test_tokenize_multiple_batched_sentences(self):
+        sentence_one = tf.constant(["the quick brown fox"] * 4)
+        sentence_two = tf.constant(["the earth"] * 4)
+        # The first tuple or list is always interpreted as an enumeration of
+        # separate sequences to concatenate.
+        output = self.preprocessor((sentence_one, sentence_two))
+        self.assertAllEqual(
+            output["token_ids"],
+            [[2, 4, 9, 5, 7, 3, 4, 6, 3, 0, 0, 0]] * 4,
+        )
+        self.assertAllEqual(
+            output["segment_ids"], [[0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0]] * 4
+        )
+        self.assertAllEqual(
+            output["padding_mask"], [[1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0]] * 4
+        )
+
+    def test_errors_for_2d_list_input(self):
+        ambiguous_input = [["one", "two"], ["three", "four"]]
+        with self.assertRaises(ValueError):
+            self.preprocessor(ambiguous_input)
+
+    @parameterized.named_parameters(
+        ("tf_format", "tf", "model"),
+        ("keras_format", "keras_v3", "model.keras"),
+    )
+    def test_saved_model(self, save_format, filename):
+        input_data = tf.constant(["the quick brown fox"])
+        inputs = keras.Input(dtype="string", shape=())
+        outputs = self.preprocessor(inputs)
+        model = keras.Model(inputs, outputs)
+        path = os.path.join(self.get_temp_dir(), filename)
+        model.save(path, save_format=save_format)
+        restored_model = keras.models.load_model(path)
+        self.assertAllEqual(
+            model(input_data)["token_ids"],
+            restored_model(input_data)["token_ids"],
+        )
diff --git a/keras_nlp/models/albert/albert_tokenizer.py b/keras_nlp/models/albert/albert_tokenizer.py
new file mode 100644
index 0000000000..e9e5ac19fc
--- /dev/null
+++ b/keras_nlp/models/albert/albert_tokenizer.py
@@ -0,0 +1,95 @@
+# Copyright 2022 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.
+
+"""ALBERT tokenizer."""
+
+
+from tensorflow import keras
+
+from keras_nlp.tokenizers.sentence_piece_tokenizer import SentencePieceTokenizer
+from keras_nlp.utils.python_utils import classproperty
+
+
+@keras.utils.register_keras_serializable(package="keras_nlp")
+class AlbertTokenizer(SentencePieceTokenizer):
+    """ALBERT tokenizer layer based on SentencePiece.
+
+    This tokenizer class will tokenize raw strings into integer sequences and
+    is based on `keras_nlp.tokenizers.SentencePieceTokenizer`. Unlike the
+    underlying tokenizer, it will check for all special tokens needed by
+    ALBERT models and provides a `from_preset()` method to automatically
+    download a matching vocabulary for a ALBERT preset.
+
+    This tokenizer does not provide truncation or padding of inputs. It can be
+    combined with a `keras_nlp.models.AlbertPreprocessor` layer for input
+    packing.
+
+    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
+    tokenizer = keras_nlp.models.AlbertTokenizer(proto="model.spm")
+
+    # Batched inputs.
+    tokenizer(["the quick brown fox", "the earth is round"])
+
+    # Unbatched inputs.
+    tokenizer("the quick brown fox")
+
+    # Detokenization.
+    tokenizer.detokenize(tf.constant([[[2, 14, 2231, 886, 2385, 3]]))
+    ```
+    """
+
+    def __init__(self, proto, **kwargs):
+        super().__init__(proto=proto, **kwargs)
+
+        # Check for necessary special tokens.
+        cls_token = "[CLS]"
+        sep_token = "[SEP]"
+        pad_token = "<pad>"
+        for token in [cls_token, sep_token, pad_token]:
+            if token not in self.get_vocabulary():
+                raise ValueError(
+                    f"Cannot find token `'{token}'` in the provided "
+                    f"`vocabulary`. Please provide `'{token}'` in your "
+                    "`vocabulary` or use a pretrained `vocabulary` name."
+                )
+
+        self.cls_token_id = self.token_to_id(cls_token)
+        self.sep_token_id = self.token_to_id(sep_token)
+        self.pad_token_id = self.token_to_id(pad_token)
+
+    @classproperty
+    def presets(cls):
+        return {}
+
+    @classmethod
+    def from_preset(
+        cls,
+        preset,
+        **kwargs,
+    ):
+        raise NotImplementedError
diff --git a/keras_nlp/models/albert/albert_tokenizer_test.py b/keras_nlp/models/albert/albert_tokenizer_test.py
new file mode 100644
index 0000000000..0116f11e22
--- /dev/null
+++ b/keras_nlp/models/albert/albert_tokenizer_test.py
@@ -0,0 +1,89 @@
+# Copyright 2022 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.
+
+"""Tests for ALBERT tokenizer."""
+
+import io
+import os
+
+import sentencepiece
+import tensorflow as tf
+from absl.testing import parameterized
+from tensorflow import keras
+
+from keras_nlp.models.albert.albert_tokenizer import AlbertTokenizer
+
+
+class AlbertTokenizerTest(tf.test.TestCase, parameterized.TestCase):
+    def setUp(self):
+        bytes_io = io.BytesIO()
+        vocab_data = tf.data.Dataset.from_tensor_slices(
+            ["the quick brown fox", "the earth is round"]
+        )
+        sentencepiece.SentencePieceTrainer.train(
+            sentence_iterator=vocab_data.as_numpy_iterator(),
+            model_writer=bytes_io,
+            vocab_size=10,
+            model_type="WORD",
+            pad_id=0,
+            unk_id=1,
+            bos_id=2,
+            eos_id=3,
+            pad_piece="<pad>",
+            unk_piece="<unk>",
+            bos_piece="[CLS]",
+            eos_piece="[SEP]",
+        )
+        self.proto = bytes_io.getvalue()
+
+        self.tokenizer = AlbertTokenizer(proto=self.proto)
+
+    def test_tokenize(self):
+        input_data = "the quick brown fox"
+        output = self.tokenizer(input_data)
+        self.assertAllEqual(output, [4, 9, 5, 7])
+
+    def test_tokenize_batch(self):
+        input_data = tf.constant(["the quick brown fox", "the earth is round"])
+        output = self.tokenizer(input_data)
+        self.assertAllEqual(output, [[4, 9, 5, 7], [4, 6, 8, 1]])
+
+    def test_detokenize(self):
+        input_data = tf.constant([[4, 9, 5, 7]])
+        output = self.tokenizer.detokenize(input_data)
+        self.assertEqual(output, tf.constant(["the quick brown fox"]))
+
+    def test_vocabulary_size(self):
+        tokenizer = AlbertTokenizer(proto=self.proto)
+        self.assertEqual(tokenizer.vocabulary_size(), 10)
+
+    @parameterized.named_parameters(
+        ("tf_format", "tf", "model"),
+        ("keras_format", "keras_v3", "model.keras"),
+    )
+    def test_saved_model(self, save_format, filename):
+        input_data = tf.constant(["the quick brown fox"])
+
+        inputs = keras.Input(dtype="string", shape=())
+        outputs = self.tokenizer(inputs)
+        model = keras.Model(inputs, outputs)
+
+        path = os.path.join(self.get_temp_dir(), filename)
+        model.save(path, save_format=save_format)
+
+        restored_model = keras.models.load_model(path)
+        self.assertAllEqual(
+            model(input_data),
+            restored_model(input_data),
+        )