[WIP] Add Tokenizer for MyT5 Model (#31286)

* Initial commit for MyT5 model

* custom implementation of MyT5 tokenizer, unused files deleted

* unittest for myt5 tokenizer

* upadate of import structure and style

* removed remmanents of MyT5Config

* fixed docstrings

* Updates after review: filled documentaion file, new docstrings and tests added

* Fixed code style issues

* fixed copied from to refer to function

* updated loading myt5 tokenizer in tests, added sample byte map file to fixtures

* changes after review

* removed redundant copied from

* removed redundant copied from

* optimalization and loading model from hf

* [run_slow] myt5

* [run-slow] myt5

* Updated en documentation for myt5

Co-authored-by: Arthur <[email protected]>

---------

Co-authored-by: Arthur <[email protected]>

docs/source/en/_toctree.yml

@@ -494,6 +494,8 @@
         title: MT5
       - local: model_doc/mvp
         title: MVP
+      - local: model_doc/myt5
+        title: myt5
       - local: model_doc/nemotron
         title: Nemotron
       - local: model_doc/nezha

docs/source/en/model_doc/myt5.md

@@ -0,0 +1,46 @@
+<!--Copyright 2024 The HuggingFace Team. All rights reserved.
+
+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
+
+http://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.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# myt5
+
+## Overview
+
+The myt5 model was proposed in [MYTE: Morphology-Driven Byte Encoding for Better and Fairer Multilingual Language Modeling](https://arxiv.org/pdf/2403.10691.pdf) by Tomasz Limisiewicz, Terra Blevins, Hila Gonen, Orevaoghene Ahia, and Luke Zettlemoyer.
+MyT5 (**My**te **T5**) is a multilingual language model based on T5 architecture.
+The model uses a **m**orphologically-driven **byte** (**MYTE**) representation described in our paper.
+**MYTE** uses codepoints corresponding to morphemes in contrast to characters used in UTF-8 encoding.
+As a pre-requisite, we used unsupervised morphological segmentation ([Morfessor](https://aclanthology.org/E14-2006.pdf)) to obtain morpheme inventories for 99 languages.
+However, the morphological segmentation step is not needed when using the pre-defined morpheme inventory from the hub (see: [Tomli/myt5-base](https://huggingface.co/Tomlim/myt5-base)).
+
+The abstract from the paper is the following:
+
+*A major consideration in multilingual language modeling is how to best represent languages with diverse vocabularies and scripts. Although contemporary text encoding methods cover most of the world’s writing systems, they exhibit bias towards the high-resource languages of the Global West. As a result, texts of underrepresented languages tend to be segmented into long sequences of linguistically meaningless units. To address the disparities, we introduce a new paradigm that encodes the same information with segments of consistent size across diverse languages. Our encoding convention (MYTE) is based on morphemes, as their inventories are more balanced across languages than characters, which are used in previous methods. We show that MYTE produces shorter encodings for all 99 analyzed languages, with the most notable improvements for non-European languages and non-Latin scripts. This, in turn, improves multilingual LM performance and diminishes the perplexity gap throughout diverse languages.*
+
+This model was contributed by [Tomasz Limisiewicz](https://huggingface.co/Tomlim).
+The original code can be found [here](https://github.com/tomlimi/MYTE).
+
+## MyT5Tokenizer
+
+[[autodoc]] MyT5Tokenizer
+    - build_inputs_with_special_tokens
+    - get_special_tokens_mask
+    - create_token_type_ids_from_sequences
+    - save_vocabulary
+
+## MyT5Tokenizer
+
+[[autodoc]] MyT5Tokenizer
+

src/transformers/__init__.py

@@ -607,6 +607,7 @@
         "MusicgenMelodyDecoderConfig",
     ],
     "models.mvp": ["MvpConfig", "MvpTokenizer"],
+    "models.myt5": ["MyT5Tokenizer"],
     "models.nemotron": ["NemotronConfig"],
     "models.nllb": [],
     "models.nllb_moe": ["NllbMoeConfig"],
@@ -5457,6 +5458,7 @@
         MusicgenMelodyDecoderConfig,
     )
     from .models.mvp import MvpConfig, MvpTokenizer
+    from .models.myt5 import MyT5Tokenizer
     from .models.nemotron import NemotronConfig
     from .models.nllb_moe import NllbMoeConfig
     from .models.nougat import NougatProcessor

src/transformers/models/__init__.py

@@ -168,6 +168,7 @@
     musicgen,
     musicgen_melody,
     mvp,
+    myt5,
     nemotron,
     nllb,
     nllb_moe,

src/transformers/models/auto/configuration_auto.py

@@ -497,6 +497,7 @@
         ("musicgen", "MusicGen"),
         ("musicgen_melody", "MusicGen Melody"),
         ("mvp", "MVP"),
+        ("myt5", "myt5"),
         ("nat", "NAT"),
         ("nemotron", "Nemotron"),
         ("nezha", "Nezha"),

src/transformers/models/auto/tokenization_auto.py

@@ -322,6 +322,7 @@
             ("musicgen", ("T5Tokenizer", "T5TokenizerFast" if is_tokenizers_available() else None)),
             ("musicgen_melody", ("T5Tokenizer", "T5TokenizerFast" if is_tokenizers_available() else None)),
             ("mvp", ("MvpTokenizer", "MvpTokenizerFast" if is_tokenizers_available() else None)),
+            ("myt5", ("MyT5Tokenizer", None)),
             ("nezha", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
             (
                 "nllb",

src/transformers/models/myt5/__init__.py

@@ -0,0 +1,29 @@
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# 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
+#
+#     http://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 typing import TYPE_CHECKING
+
+from ...utils import _LazyModule
+
+
+_import_structure = {"tokenization_myt5": ["MyT5Tokenizer"]}
+
+
+if TYPE_CHECKING:
+    from .tokenization_myt5 import MyT5Tokenizer
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/myt5/convert_myt5_original_tf_checkpoint_to_pytorch.py

@@ -0,0 +1,60 @@
+# coding=utf-8
+# Copyright 2024 The MyT5 authors and HuggingFace Inc. team.
+#
+# 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
+#
+#     http://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.
+"""Convert MyT5 checkpoint."""
+
+import argparse
+
+from transformers import T5Config, T5ForConditionalGeneration, load_tf_weights_in_t5
+from transformers.utils import logging
+
+
+logging.set_verbosity_info()
+
+
+# Copied from transformers.models.t5.convert_t5_original_tf_checkpoint_to_pytorch.convert_tf_checkpoint_to_pytorch
+def convert_tf_checkpoint_to_pytorch(tf_checkpoint_path, config_file, pytorch_dump_path):
+    # Initialise PyTorch model
+    config = T5Config.from_json_file(config_file)
+    print(f"Building PyTorch model from configuration: {config}")
+    model = T5ForConditionalGeneration(config)
+
+    # Load weights from tf checkpoint
+    load_tf_weights_in_t5(model, config, tf_checkpoint_path)
+
+    # Save pytorch-model
+    print(f"Save PyTorch model to {pytorch_dump_path}")
+    model.save_pretrained(pytorch_dump_path)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    # Required parameters
+    parser.add_argument(
+        "--tf_checkpoint_path", default=None, type=str, required=True, help="Path to the TensorFlow checkpoint path."
+    )
+    parser.add_argument(
+        "--config_file",
+        default=None,
+        type=str,
+        required=True,
+        help=(
+            "The config json file corresponding to the pre-trained MyT5 model. \nThis specifies the model architecture."
+        ),
+    )
+    parser.add_argument(
+        "--pytorch_dump_path", default=None, type=str, required=True, help="Path to the output PyTorch model."
+    )
+    args = parser.parse_args()
+    convert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.config_file, args.pytorch_dump_path)