NLLB tokenizer (#18126)

* NLLB tokenizer

* Apply suggestions from code review - Thanks Stefan!
Co-authored-by: Stefan Schweter <stefan@schweter.it>

* Final touches

* Style :)

* Update docs/source/en/model_doc/nllb.mdx
Co-authored-by: Stefan Schweter <stefan@schweter.it>

* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* PR reviews

* Auto models
Co-authored-by: Stefan Schweter <stefan@schweter.it>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

README.md

@@ -304,6 +304,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
 1. **[MVP](https://huggingface.co/docs/transformers/main/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
 1. **[Nezha](https://huggingface.co/docs/transformers/main/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
+1. **[NLLB](https://huggingface.co/docs/transformers/main/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
 1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.

README_ko.md

@@ -285,6 +285,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
 1. **[MVP](https://huggingface.co/docs/transformers/main/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
 1. **[Nezha](https://huggingface.co/docs/transformers/main/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
+1. **[NLLB](https://huggingface.co/docs/transformers/main/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
 1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.

README_zh-hans.md

@@ -309,6 +309,7 @@ conda install -c huggingface transformers
 1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (来自 Google AI) 伴随论文 [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) 由 Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel 发布。
 1. **[MVP](https://huggingface.co/docs/transformers/main/model_doc/mvp)** (来自 中国人民大学 AI Box) 伴随论文 [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) 由 Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen 发布。 
 1. **[Nezha](https://huggingface.co/docs/transformers/main/model_doc/nezha)** (来自华为诺亚方舟实验室) 伴随论文 [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) 由 Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu 发布。
+1. **[NLLB](https://huggingface.co/docs/transformers/main/model_doc/nllb)** (来自 Meta) 伴随论文 [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) 由 the NLLB team 发布。
 1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (来自 the University of Wisconsin - Madison) 伴随论文 [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) 由 Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh 发布。
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (来自 Meta AI) 伴随论文 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 由 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 发布。
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (来自 Google) 伴随论文 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 由 Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 发布。

README_zh-hant.md

@@ -321,6 +321,7 @@ conda install -c huggingface transformers
 1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
 1. **[MVP](https://huggingface.co/docs/transformers/main/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
 1. **[Nezha](https://huggingface.co/docs/transformers/main/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
+1. **[NLLB](https://huggingface.co/docs/transformers/main/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
 1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.

docs/source/en/_toctree.yml

@@ -320,6 +320,8 @@
       title: MVP
     - local: model_doc/nezha
       title: NEZHA
+    - local: model_doc/nllb
+      title: NLLB
     - local: model_doc/nystromformer
       title: Nyströmformer
     - local: model_doc/opt

docs/source/en/index.mdx

@@ -127,6 +127,7 @@ The library currently contains JAX, PyTorch and TensorFlow implementations, pret
 1. **[MT5](model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
 1. **[MVP](model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
 1. **[Nezha](model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
+1. **[NLLB](model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
 1. **[Nyströmformer](model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
 1. **[OPT](master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[Pegasus](model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.

docs/source/en/model_doc/nllb.mdx

+<!--Copyright 2020 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.
+-->
+
+# NLLB
+
+**DISCLAIMER:** If you see something strange, file a [Github Issue](https://github.com/huggingface/transformers/issues/new?assignees=&labels=bug&template=bug-report.yml) and assign
+@LysandreJik
+
+## Overview of NLLB
+
+The NLLB model was presented in [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by Marta R. Costa-jussà, James Cross, Onur Çelebi,
+Maha Elbayad, Kenneth Heafield, Kevin Heffernan, Elahe Kalbassi, Janice Lam, Daniel Licht, Jean Maillard, Anna Sun, Skyler Wang, Guillaume Wenzek, Al Youngblood, Bapi Akula,
+Loic Barrault, Gabriel Mejia Gonzalez, Prangthip Hansanti, John Hoffman, Semarley Jarrett, Kaushik Ram Sadagopan, Dirk Rowe, Shannon Spruit, Chau Tran, Pierre Andrews,
+Necip Fazil Ayan, Shruti Bhosale, Sergey Edunov, Angela Fan, Cynthia Gao, Vedanuj Goswami, Francisco Guzmán, Philipp Koehn, Alexandre Mourachko, Christophe Ropers,
+Safiyyah Saleem, Holger Schwenk, and Jeff Wang.
+
+The abstract of the paper is the following:
+
+*Driven by the goal of eradicating language barriers on a global scale, machine translation has solidified itself as a key focus of artificial intelligence research today.
+However, such efforts have coalesced around a small subset of languages, leaving behind the vast majority of mostly low-resource languages. What does it take to break the
+200 language barrier while ensuring safe, high quality results, all while keeping ethical considerations in mind? In No Language Left Behind, we took on this challenge by
+first contextualizing the need for low-resource language translation support through exploratory interviews with native speakers. Then, we created datasets and models aimed
+at narrowing the performance gap between low and high-resource languages. More specifically, we developed a conditional compute model based on Sparsely Gated Mixture of
+Experts that is trained on data obtained with novel and effective data mining techniques tailored for low-resource languages. We propose multiple architectural and training
+improvements to counteract overfitting while training on thousands of tasks. Critically, we evaluated the performance of over 40,000 different translation directions using
+a human-translated benchmark, Flores-200, and combined human evaluation with a novel toxicity benchmark covering all languages in Flores-200 to assess translation safety.
+Our model achieves an improvement of 44% BLEU relative to the previous state-of-the-art, laying important groundwork towards realizing a universal translation system.*
+
+This implementation contains the dense models available on release. Let us know via a GitHub issue if you would like to see the MoE models as well.
+
+This model was contributed by [Lysandre](https://huggingface.co/lysandre). The authors' code can be found [here](https://github.com/facebookresearch/fairseq/tree/nllb).
+
+## Generating with NLLB
+
+While generating the target text set the `forced_bos_token_id` to the target language id. The following
+example shows how to translate English to French using the *facebook/nllb-200-distilled-600M* model.
+
+Note that we're using the BCP-47 code for French `fra_Latn`. See [here](https://github.com/facebookresearch/flores/blob/main/flores200/README.md#languages-in-flores-200)
+for the list of all BCP-47 in the Flores 200 dataset.
+
+```python
+>>> from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+
+>>> tokenizer = AutoTokenizer.from_pretrained("facebook/nllb-200-distilled-600M")
+>>> model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-600M")
+
+>>> article = "UN Chief says there is no military solution in Syria"
+>>> inputs = tokenizer(article, return_tensors="pt")
+
+>>> translated_tokens = model.generate(
+...     **inputs, forced_bos_token_id=tokenizer.lang_code_to_id["fra_Latn"], max_length=30
+... )
+>>> tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
+Le chef de l'ONU dit qu'il n'y a pas de solution militaire en Syrie
+```
+
+### Generating from any other language than English
+
+English (`eng_Latn`) is set as the default language from which to translate. In order to specify that you'd like to translate from a different language,
+you should specify the BCP-47 code in the `src_lang` keyword argument of the tokenizer initialization.
+
+See example below for a translation from romanian to german:
+
+```py
+>>> from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+
+>>> tokenizer = AutoTokenizer.from_pretrained(
+...     "facebook/nllb-200-distilled-600M", use_auth_token=True, src_lang="ron_Latn"
+... )
+>>> model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-600M", use_auth_token=True)
+
+>>> article = "Şeful ONU spune că nu există o soluţie militară în Siria"
+>>> inputs = tokenizer(article, return_tensors="pt")
+
+>>> translated_tokens = model.generate(
+...     **inputs, forced_bos_token_id=tokenizer.lang_code_to_id["deu_Latn"], max_length=30
+... )
+>>> tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
+UN-Chef sagt, es gibt keine militärische Lösung in Syrien
+```
+
+## NllbTokenizer
+
+[[autodoc]] NllbTokenizer
+    - as_target_tokenizer
+    - build_inputs_with_special_tokens
+
+## NllbTokenizerFast
+
+[[autodoc]] NllbTokenizerFast

src/transformers/__init__.py

@@ -266,6 +266,7 @@ _import_structure = {
     "models.mt5": ["MT5Config"],
     "models.mvp": ["MvpConfig", "MvpTokenizer"],
     "models.nezha": ["NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP", "NezhaConfig"],
+    "models.nllb": [],
     "models.nystromformer": [
         "NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
         "NystromformerConfig",
@@ -479,6 +480,7 @@ else:
     _import_structure["models.m2m_100"].append("M2M100Tokenizer")
     _import_structure["models.marian"].append("MarianTokenizer")
     _import_structure["models.mbart"].append("MBartTokenizer")
+    _import_structure["models.nllb"].append("NllbTokenizer")
     _import_structure["models.mbart50"].append("MBart50Tokenizer")
     _import_structure["models.mluke"].append("MLukeTokenizer")
     _import_structure["models.mt5"].append("MT5Tokenizer")
@@ -543,6 +545,7 @@ else:
     _import_structure["models.mpnet"].append("MPNetTokenizerFast")
     _import_structure["models.mt5"].append("MT5TokenizerFast")
     _import_structure["models.mvp"].append("MvpTokenizerFast")
+    _import_structure["models.nllb"].append("NllbTokenizerFast")
     _import_structure["models.openai"].append("OpenAIGPTTokenizerFast")
     _import_structure["models.pegasus"].append("PegasusTokenizerFast")
     _import_structure["models.realm"].append("RealmTokenizerFast")
@@ -3188,6 +3191,7 @@ if TYPE_CHECKING:
         from .models.mbart import MBart50Tokenizer, MBartTokenizer
         from .models.mluke import MLukeTokenizer
         from .models.mt5 import MT5Tokenizer
+        from .models.nllb import NllbTokenizer
         from .models.pegasus import PegasusTokenizer
         from .models.plbart import PLBartTokenizer
         from .models.reformer import ReformerTokenizer
@@ -3242,6 +3246,7 @@ if TYPE_CHECKING:
         from .models.mpnet import MPNetTokenizerFast
         from .models.mt5 import MT5TokenizerFast
         from .models.mvp import MvpTokenizerFast
+        from .models.nllb import NllbTokenizerFast
         from .models.openai import OpenAIGPTTokenizerFast
         from .models.pegasus import PegasusTokenizerFast
         from .models.realm import RealmTokenizerFast

src/transformers/convert_slow_tokenizer.py

@@ -694,6 +694,37 @@ class MBart50Converter(SpmConverter):
         )
 
 
+class NllbConverter(SpmConverter):
+    def vocab(self, proto):
+        vocab = [
+            ("<s>", 0.0),
+            ("<pad>", 0.0),
+            ("</s>", 0.0),
+            ("<unk>", 0.0),
+        ]
+        vocab += [(piece.piece, piece.score) for piece in proto.pieces[3:]]
+        vocab += [
+            # fmt: off
+            ('ace_Arab', 0.0), ('ace_Latn', 0.0), ('acm_Arab', 0.0), ('acq_Arab', 0.0), ('aeb_Arab', 0.0), ('afr_Latn', 0.0), ('ajp_Arab', 0.0), ('aka_Latn', 0.0), ('amh_Ethi', 0.0), ('apc_Arab', 0.0), ('arb_Arab', 0.0), ('ars_Arab', 0.0), ('ary_Arab', 0.0), ('arz_Arab', 0.0), ('asm_Beng', 0.0), ('ast_Latn', 0.0), ('awa_Deva', 0.0), ('ayr_Latn', 0.0), ('azb_Arab', 0.0), ('azj_Latn', 0.0), ('bak_Cyrl', 0.0), ('bam_Latn', 0.0), ('ban_Latn', 0.0), ('bel_Cyrl', 0.0), ('bem_Latn', 0.0), ('ben_Beng', 0.0), ('bho_Deva', 0.0), ('bjn_Arab', 0.0), ('bjn_Latn', 0.0), ('bod_Tibt', 0.0), ('bos_Latn', 0.0), ('bug_Latn', 0.0), ('bul_Cyrl', 0.0), ('cat_Latn', 0.0), ('ceb_Latn', 0.0), ('ces_Latn', 0.0), ('cjk_Latn', 0.0), ('ckb_Arab', 0.0), ('crh_Latn', 0.0), ('cym_Latn', 0.0), ('dan_Latn', 0.0), ('deu_Latn', 0.0), ('dik_Latn', 0.0), ('dyu_Latn', 0.0), ('dzo_Tibt', 0.0), ('ell_Grek', 0.0), ('eng_Latn', 0.0), ('epo_Latn', 0.0), ('est_Latn', 0.0), ('eus_Latn', 0.0), ('ewe_Latn', 0.0), ('fao_Latn', 0.0), ('pes_Arab', 0.0), ('fij_Latn', 0.0), ('fin_Latn', 0.0), ('fon_Latn', 0.0), ('fra_Latn', 0.0), ('fur_Latn', 0.0), ('fuv_Latn', 0.0), ('gla_Latn', 0.0), ('gle_Latn', 0.0), ('glg_Latn', 0.0), ('grn_Latn', 0.0), ('guj_Gujr', 0.0), ('hat_Latn', 0.0), ('hau_Latn', 0.0), ('heb_Hebr', 0.0), ('hin_Deva', 0.0), ('hne_Deva', 0.0), ('hrv_Latn', 0.0), ('hun_Latn', 0.0), ('hye_Armn', 0.0), ('ibo_Latn', 0.0), ('ilo_Latn', 0.0), ('ind_Latn', 0.0), ('isl_Latn', 0.0), ('ita_Latn', 0.0), ('jav_Latn', 0.0), ('jpn_Jpan', 0.0), ('kab_Latn', 0.0), ('kac_Latn', 0.0), ('kam_Latn', 0.0), ('kan_Knda', 0.0), ('kas_Arab', 0.0), ('kas_Deva', 0.0), ('kat_Geor', 0.0), ('knc_Arab', 0.0), ('knc_Latn', 0.0), ('kaz_Cyrl', 0.0), ('kbp_Latn', 0.0), ('kea_Latn', 0.0), ('khm_Khmr', 0.0), ('kik_Latn', 0.0), ('kin_Latn', 0.0), ('kir_Cyrl', 0.0), ('kmb_Latn', 0.0), ('kon_Latn', 0.0), ('kor_Hang', 0.0), ('kmr_Latn', 0.0), ('lao_Laoo', 0.0), ('lvs_Latn', 0.0), ('lij_Latn', 0.0), ('lim_Latn', 0.0), ('lin_Latn', 0.0), ('lit_Latn', 0.0), ('lmo_Latn', 0.0), ('ltg_Latn', 0.0), ('ltz_Latn', 0.0), ('lua_Latn', 0.0), ('lug_Latn', 0.0), ('luo_Latn', 0.0), ('lus_Latn', 0.0), ('mag_Deva', 0.0), ('mai_Deva', 0.0), ('mal_Mlym', 0.0), ('mar_Deva', 0.0), ('min_Latn', 0.0), ('mkd_Cyrl', 0.0), ('plt_Latn', 0.0), ('mlt_Latn', 0.0), ('mni_Beng', 0.0), ('khk_Cyrl', 0.0), ('mos_Latn', 0.0), ('mri_Latn', 0.0), ('zsm_Latn', 0.0), ('mya_Mymr', 0.0), ('nld_Latn', 0.0), ('nno_Latn', 0.0), ('nob_Latn', 0.0), ('npi_Deva', 0.0), ('nso_Latn', 0.0), ('nus_Latn', 0.0), ('nya_Latn', 0.0), ('oci_Latn', 0.0), ('gaz_Latn', 0.0), ('ory_Orya', 0.0), ('pag_Latn', 0.0), ('pan_Guru', 0.0), ('pap_Latn', 0.0), ('pol_Latn', 0.0), ('por_Latn', 0.0), ('prs_Arab', 0.0), ('pbt_Arab', 0.0), ('quy_Latn', 0.0), ('ron_Latn', 0.0), ('run_Latn', 0.0), ('rus_Cyrl', 0.0), ('sag_Latn', 0.0), ('san_Deva', 0.0), ('sat_Beng', 0.0), ('scn_Latn', 0.0), ('shn_Mymr', 0.0), ('sin_Sinh', 0.0), ('slk_Latn', 0.0), ('slv_Latn', 0.0), ('smo_Latn', 0.0), ('sna_Latn', 0.0), ('snd_Arab', 0.0), ('som_Latn', 0.0), ('sot_Latn', 0.0), ('spa_Latn', 0.0), ('als_Latn', 0.0), ('srd_Latn', 0.0), ('srp_Cyrl', 0.0), ('ssw_Latn', 0.0), ('sun_Latn', 0.0), ('swe_Latn', 0.0), ('swh_Latn', 0.0), ('szl_Latn', 0.0), ('tam_Taml', 0.0), ('tat_Cyrl', 0.0), ('tel_Telu', 0.0), ('tgk_Cyrl', 0.0), ('tgl_Latn', 0.0), ('tha_Thai', 0.0), ('tir_Ethi', 0.0), ('taq_Latn', 0.0), ('taq_Tfng', 0.0), ('tpi_Latn', 0.0), ('tsn_Latn', 0.0), ('tso_Latn', 0.0), ('tuk_Latn', 0.0), ('tum_Latn', 0.0), ('tur_Latn', 0.0), ('twi_Latn', 0.0), ('tzm_Tfng', 0.0), ('uig_Arab', 0.0), ('ukr_Cyrl', 0.0), ('umb_Latn', 0.0), ('urd_Arab', 0.0), ('uzn_Latn', 0.0), ('vec_Latn', 0.0), ('vie_Latn', 0.0), ('war_Latn', 0.0), ('wol_Latn', 0.0), ('xho_Latn', 0.0), ('ydd_Hebr', 0.0), ('yor_Latn', 0.0), ('yue_Hant', 0.0), ('zho_Hans', 0.0), ('zho_Hant', 0.0), ('zul_Latn', 0.0)
+            # fmt: on
+        ]
+        vocab += [("<mask>", 0.0)]
+        return vocab
+
+    def unk_id(self, proto):
+        return 3
+
+    def post_processor(self):
+        return processors.TemplateProcessing(
+            single="eng_Latn $A </s>",
+            pair="eng_Latn $A $B </s>",
+            special_tokens=[
+                ("eng_Latn", self.original_tokenizer.convert_tokens_to_ids("eng_Latn")),
+                ("</s>", self.original_tokenizer.convert_tokens_to_ids("</s>")),
+            ],
+        )
+
+
 class XLMRobertaConverter(SpmConverter):
     def vocab(self, proto):
         vocab = [
@@ -1034,6 +1065,7 @@ SLOW_TO_FAST_CONVERTERS = {
     "MPNetTokenizer": MPNetConverter,
     "MobileBertTokenizer": BertConverter,
     "MvpTokenizer": RobertaConverter,
+    "NllbTokenizer": NllbConverter,
     "OpenAIGPTTokenizer": OpenAIGPTConverter,
     "PegasusTokenizer": PegasusConverter,
     "RealmTokenizer": BertConverter,

src/transformers/models/__init__.py

@@ -97,6 +97,7 @@ from . import (
     mt5,
     mvp,
     nezha,
+    nllb,
     nystromformer,
     openai,
     opt,

src/transformers/models/auto/configuration_auto.py

@@ -342,6 +342,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("mt5", "MT5"),
         ("mvp", "MVP"),
         ("nezha", "Nezha"),
+        ("nllb", "NLLB"),
         ("nystromformer", "Nyströmformer"),
         ("openai-gpt", "OpenAI GPT"),
         ("opt", "OPT"),

src/transformers/models/auto/modeling_auto.py

@@ -94,6 +94,7 @@ MODEL_MAPPING_NAMES = OrderedDict(
         ("mt5", "MT5Model"),
         ("mvp", "MvpModel"),
         ("nezha", "NezhaModel"),
+        ("nllb", "M2M100Model"),
         ("nystromformer", "NystromformerModel"),
         ("openai-gpt", "OpenAIGPTModel"),
         ("opt", "OPTModel"),
@@ -234,6 +235,7 @@ MODEL_WITH_LM_HEAD_MAPPING_NAMES = OrderedDict(
         ("mpnet", "MPNetForMaskedLM"),
         ("mvp", "MvpForConditionalGeneration"),
         ("nezha", "NezhaForMaskedLM"),
+        ("nllb", "M2M100ForConditionalGeneration"),
         ("nystromformer", "NystromformerForMaskedLM"),
         ("openai-gpt", "OpenAIGPTLMHeadModel"),
         ("plbart", "PLBartForConditionalGeneration"),
@@ -447,6 +449,7 @@ MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
         ("mbart", "MBartForConditionalGeneration"),
         ("mt5", "MT5ForConditionalGeneration"),
         ("mvp", "MvpForConditionalGeneration"),
+        ("nllb", "M2M100ForConditionalGeneration"),
         ("pegasus", "PegasusForConditionalGeneration"),
         ("plbart", "PLBartForConditionalGeneration"),
         ("prophetnet", "ProphetNetForConditionalGeneration"),

src/transformers/models/auto/tokenization_auto.py

@@ -177,6 +177,13 @@ else:
             ),
             ("mvp", ("MvpTokenizer", "MvpTokenizerFast" if is_tokenizers_available() else None)),
             ("nezha", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
+            (
+                "nllb",
+                (
+                    "NllbTokenizer" if is_sentencepiece_available() else None,
+                    "NllbTokenizerFast" if is_tokenizers_available() else None,
+                ),
+            ),
             (
                 "nystromformer",
                 (

src/transformers/models/nllb/__init__.py

+# flake8: noqa
+# There's no way to ignore "F401 '...' imported but unused" warnings in this
+# module, but to preserve other warnings. So, don't check this module at all.
+
+# Copyright 2022 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 (
+    OptionalDependencyNotAvailable,
+    _LazyModule,
+    is_sentencepiece_available,
+    is_tokenizers_available,
+    is_torch_available,
+)
+
+
+_import_structure = {}
+
+try:
+    if not is_sentencepiece_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["tokenization_nllb"] = ["NllbTokenizer"]
+
+try:
+    if not is_tokenizers_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["tokenization_nllb_fast"] = ["NllbTokenizerFast"]
+
+
+if TYPE_CHECKING:
+    try:
+        if not is_sentencepiece_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .tokenization_nllb import NllbTokenizer
+
+    try:
+        if not is_tokenizers_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .tokenization_nllb_fast import NllbTokenizerFast
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/nllb/tokenization_nllb.py

+# coding=utf-8
+# Copyright 2022 The Facebook AI Research Team Authors and The 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.
+
+import os
+from contextlib import contextmanager
+from shutil import copyfile
+from typing import Any, Dict, List, Optional, Tuple
+
+import sentencepiece as spm
+
+from ...tokenization_utils import AddedToken, BatchEncoding, PreTrainedTokenizer
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+SPIECE_UNDERLINE = "▁"
+
+VOCAB_FILES_NAMES = {"vocab_file": "sentencepiece.bpe.model"}
+
+PRETRAINED_VOCAB_FILES_MAP = {
+    "vocab_file": {
+        "facebook/nllb-200-distilled-600M": (
+            "https://huggingface.co/facebook/nllb-200-distilled-600M/blob/main/sentencepiece.bpe.model"
+        ),
+    }
+}
+
+PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
+    "facebook/nllb-200-distilled-600M": 1024,
+}
+
+# fmt: off
+FAIRSEQ_LANGUAGE_CODES = ['ace_Arab', 'ace_Latn', 'acm_Arab', 'acq_Arab', 'aeb_Arab', 'afr_Latn', 'ajp_Arab', 'aka_Latn', 'amh_Ethi', 'apc_Arab', 'arb_Arab', 'ars_Arab', 'ary_Arab', 'arz_Arab', 'asm_Beng', 'ast_Latn', 'awa_Deva', 'ayr_Latn', 'azb_Arab', 'azj_Latn', 'bak_Cyrl', 'bam_Latn', 'ban_Latn', 'bel_Cyrl', 'bem_Latn', 'ben_Beng', 'bho_Deva', 'bjn_Arab', 'bjn_Latn', 'bod_Tibt', 'bos_Latn', 'bug_Latn', 'bul_Cyrl', 'cat_Latn', 'ceb_Latn', 'ces_Latn', 'cjk_Latn', 'ckb_Arab', 'crh_Latn', 'cym_Latn', 'dan_Latn', 'deu_Latn', 'dik_Latn', 'dyu_Latn', 'dzo_Tibt', 'ell_Grek', 'eng_Latn', 'epo_Latn', 'est_Latn', 'eus_Latn', 'ewe_Latn', 'fao_Latn', 'pes_Arab', 'fij_Latn', 'fin_Latn', 'fon_Latn', 'fra_Latn', 'fur_Latn', 'fuv_Latn', 'gla_Latn', 'gle_Latn', 'glg_Latn', 'grn_Latn', 'guj_Gujr', 'hat_Latn', 'hau_Latn', 'heb_Hebr', 'hin_Deva', 'hne_Deva', 'hrv_Latn', 'hun_Latn', 'hye_Armn', 'ibo_Latn', 'ilo_Latn', 'ind_Latn', 'isl_Latn', 'ita_Latn', 'jav_Latn', 'jpn_Jpan', 'kab_Latn', 'kac_Latn', 'kam_Latn', 'kan_Knda', 'kas_Arab', 'kas_Deva', 'kat_Geor', 'knc_Arab', 'knc_Latn', 'kaz_Cyrl', 'kbp_Latn', 'kea_Latn', 'khm_Khmr', 'kik_Latn', 'kin_Latn', 'kir_Cyrl', 'kmb_Latn', 'kon_Latn', 'kor_Hang', 'kmr_Latn', 'lao_Laoo', 'lvs_Latn', 'lij_Latn', 'lim_Latn', 'lin_Latn', 'lit_Latn', 'lmo_Latn', 'ltg_Latn', 'ltz_Latn', 'lua_Latn', 'lug_Latn', 'luo_Latn', 'lus_Latn', 'mag_Deva', 'mai_Deva', 'mal_Mlym', 'mar_Deva', 'min_Latn', 'mkd_Cyrl', 'plt_Latn', 'mlt_Latn', 'mni_Beng', 'khk_Cyrl', 'mos_Latn', 'mri_Latn', 'zsm_Latn', 'mya_Mymr', 'nld_Latn', 'nno_Latn', 'nob_Latn', 'npi_Deva', 'nso_Latn', 'nus_Latn', 'nya_Latn', 'oci_Latn', 'gaz_Latn', 'ory_Orya', 'pag_Latn', 'pan_Guru', 'pap_Latn', 'pol_Latn', 'por_Latn', 'prs_Arab', 'pbt_Arab', 'quy_Latn', 'ron_Latn', 'run_Latn', 'rus_Cyrl', 'sag_Latn', 'san_Deva', 'sat_Beng', 'scn_Latn', 'shn_Mymr', 'sin_Sinh', 'slk_Latn', 'slv_Latn', 'smo_Latn', 'sna_Latn', 'snd_Arab', 'som_Latn', 'sot_Latn', 'spa_Latn', 'als_Latn', 'srd_Latn', 'srp_Cyrl', 'ssw_Latn', 'sun_Latn', 'swe_Latn', 'swh_Latn', 'szl_Latn', 'tam_Taml', 'tat_Cyrl', 'tel_Telu', 'tgk_Cyrl', 'tgl_Latn', 'tha_Thai', 'tir_Ethi', 'taq_Latn', 'taq_Tfng', 'tpi_Latn', 'tsn_Latn', 'tso_Latn', 'tuk_Latn', 'tum_Latn', 'tur_Latn', 'twi_Latn', 'tzm_Tfng', 'uig_Arab', 'ukr_Cyrl', 'umb_Latn', 'urd_Arab', 'uzn_Latn', 'vec_Latn', 'vie_Latn', 'war_Latn', 'wol_Latn', 'xho_Latn', 'ydd_Hebr', 'yor_Latn', 'yue_Hant', 'zho_Hans', 'zho_Hant', 'zul_Latn']
+# fmt: on
+
+
+class NllbTokenizer(PreTrainedTokenizer):
+    """
+    Construct an NLLB tokenizer.
+
+    Adapted from [`RobertaTokenizer`] and [`XLNetTokenizer`]. Based on
+    [SentencePiece](https://github.com/google/sentencepiece).
+
+    The tokenization method is `<tokens> <eos> <language code>` for source language documents, and `<language code>
+    <tokens> <eos>` for target language documents.
+
+    Examples:
+
+    ```python
+    >>> from transformers import NllbTokenizer
+
+    >>> tokenizer = NllbTokenizer.from_pretrained(
+    ...     "facebook/nllb-200-distilled-600M", src_lang="eng_Latn", tgt_lang="fra_Latn"
+    ... )
+    >>> example_english_phrase = " UN Chief Says There Is No Military Solution in Syria"
+    >>> expected_translation_french = "Le chef de l'ONU affirme qu'il n'y a pas de solution militaire en Syrie."
+    >>> inputs = tokenizer(example_english_phrase, return_tensors="pt")
+    >>> with tokenizer.as_target_tokenizer():
+    ...     labels = tokenizer(expected_translation_french, return_tensors="pt")
+    >>> inputs["labels"] = labels["input_ids"]
+    ```
+
+    Args:
+        vocab_file (`str`):
+            Path to the vocabulary file.
+        bos_token (`str`, *optional*, defaults to `"<s>"`):
+            The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token.
+
+            <Tip>
+
+            When building a sequence using special tokens, this is not the token that is used for the beginning of
+            sequence. The token used is the `cls_token`.
+
+            </Tip>
+
+        eos_token (`str`, *optional*, defaults to `"</s>"`):
+            The end of sequence token.
+
+            <Tip>
+
+            When building a sequence using special tokens, this is not the token that is used for the end of sequence.
+            The token used is the `sep_token`.
+
+            </Tip>
+
+        sep_token (`str`, *optional*, defaults to `"</s>"`):
+            The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
+            sequence classification or for a text and a question for question answering. It is also used as the last
+            token of a sequence built with special tokens.
+        cls_token (`str`, *optional*, defaults to `"<s>"`):
+            The classifier token which is used when doing sequence classification (classification of the whole sequence
+            instead of per-token classification). It is the first token of the sequence when built with special tokens.
+        unk_token (`str`, *optional*, defaults to `"<unk>"`):
+            The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
+            token instead.
+        pad_token (`str`, *optional*, defaults to `"<pad>"`):
+            The token used for padding, for example when batching sequences of different lengths.
+        mask_token (`str`, *optional*, defaults to `"<mask>"`):
+            The token used for masking values. This is the token used when training this model with masked language
+            modeling. This is the token which the model will try to predict.
+        tokenizer_file (`str`, *optional*):
+            The path to a tokenizer file to use instead of the vocab file.
+        src_lang (`str`, *optional*):
+            The language to use as source language for translation.
+        tgt_lang (`str`, *optional*):
+            The language to use as target language for translation.
+        sp_model_kwargs (`Dict[str, str]`):
+            Additional keyword arguments to pass to the model initialization.
+    """
+
+    vocab_files_names = VOCAB_FILES_NAMES
+    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
+    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
+    model_input_names = ["input_ids", "attention_mask"]
+
+    prefix_tokens: List[int] = []
+    suffix_tokens: List[int] = []
+
+    def __init__(
+        self,
+        vocab_file,
+        bos_token="<s>",
+        eos_token="</s>",
+        sep_token="</s>",
+        cls_token="<s>",
+        unk_token="<unk>",
+        pad_token="<pad>",
+        mask_token="<mask>",
+        tokenizer_file=None,
+        src_lang=None,
+        tgt_lang=None,
+        sp_model_kwargs: Optional[Dict[str, Any]] = None,
+        additional_special_tokens=None,
+        **kwargs
+    ):
+
+        # Mask token behave like a normal word, i.e. include the space before it
+        mask_token = AddedToken(mask_token, lstrip=True, rstrip=False) if isinstance(mask_token, str) else mask_token
+
+        self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
+
+        super().__init__(
+            bos_token=bos_token,
+            eos_token=eos_token,
+            unk_token=unk_token,
+            sep_token=sep_token,
+            cls_token=cls_token,
+            pad_token=pad_token,
+            mask_token=mask_token,
+            tokenizer_file=tokenizer_file,
+            src_lang=src_lang,
+            tgt_lang=tgt_lang,
+            additional_special_tokens=additional_special_tokens,
+            sp_model_kwargs=self.sp_model_kwargs,
+            **kwargs,
+        )
+
+        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
+        self.sp_model.Load(str(vocab_file))
+        self.vocab_file = vocab_file
+
+        # Original fairseq vocab and spm vocab must be "aligned":
+        # Vocab    |    0    |    1    |   2    |    3    |  4   |  5   |  6   |   7  |   8  |  9
+        # -------- | ------- | ------- | ------ | ------- | ---- | ---- | ---- | ---- | ---- | ----
+        # fairseq  | '<s>'   | '<pad>' | '</s>' | '<unk>' | 'an' | '▁n' | '▁m' | '▁t' | '▁k' | '▁a'
+        # spm      | '<unk>' | '<s>'   | '</s>' | 'an'    | '▁n' | '▁m' | '▁t' | '▁k' | '▁a' | '▁s'
+
+        # Mimic fairseq token-to-id alignment for the first 4 token
+        self.fairseq_tokens_to_ids = {"<s>": 0, "<pad>": 1, "</s>": 2, "<unk>": 3}
+
+        # The first "real" token "," has position 4 in the original fairseq vocab and position 3 in the spm vocab
+        self.fairseq_offset = 1
+
+        self.sp_model_size = len(self.sp_model)
+        self.lang_code_to_id = {
+            code: self.sp_model_size + i + self.fairseq_offset for i, code in enumerate(FAIRSEQ_LANGUAGE_CODES)
+        }
+        self.id_to_lang_code = {v: k for k, v in self.lang_code_to_id.items()}
+        self.fairseq_tokens_to_ids["<mask>"] = len(self.sp_model) + len(self.lang_code_to_id) + self.fairseq_offset
+
+        self.fairseq_tokens_to_ids.update(self.lang_code_to_id)
+        self.fairseq_ids_to_tokens = {v: k for k, v in self.fairseq_tokens_to_ids.items()}
+        self._additional_special_tokens = list(self.lang_code_to_id.keys())
+
+        if additional_special_tokens is not None:
+            # Only add those special tokens if they are not already there.
+            self._additional_special_tokens.extend(
+                [t for t in additional_special_tokens if t not in self._additional_special_tokens]
+            )
+
+        self._src_lang = src_lang if src_lang is not None else "eng_Latn"
+        self.cur_lang_code_id = self.lang_code_to_id[self._src_lang]
+        self.tgt_lang = tgt_lang
+        self.set_src_lang_special_tokens(self._src_lang)
+
+    def __getstate__(self):
+        state = self.__dict__.copy()
+        state["sp_model"] = None
+        state["sp_model_proto"] = self.sp_model.serialized_model_proto()
+        return state
+
+    def __setstate__(self, d):
+        self.__dict__ = d
+
+        # for backward compatibility
+        if not hasattr(self, "sp_model_kwargs"):
+            self.sp_model_kwargs = {}
+
+        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
+        self.sp_model.LoadFromSerializedProto(self.sp_model_proto)
+
+    @property
+    def vocab_size(self):
+        return len(self.sp_model) + len(self.lang_code_to_id) + self.fairseq_offset + 1  # Plus 1 for the mask token
+
+    @property
+    def src_lang(self) -> str:
+        return self._src_lang
+
+    @src_lang.setter
+    def src_lang(self, new_src_lang: str) -> None:
+        self._src_lang = new_src_lang
+        self.set_src_lang_special_tokens(self._src_lang)
+
+    def get_special_tokens_mask(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
+    ) -> List[int]:
+        """
+        Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
+        special tokens using the tokenizer `prepare_for_model` method.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
+                Whether or not the token list is already formatted with special tokens for the model.
+
+        Returns:
+            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
+        """
+
+        if already_has_special_tokens:
+            return super().get_special_tokens_mask(
+                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
+            )
+
+        prefix_ones = [1] * len(self.prefix_tokens)
+        suffix_ones = [1] * len(self.suffix_tokens)
+        if token_ids_1 is None:
+            return prefix_ones + ([0] * len(token_ids_0)) + suffix_ones
+        return prefix_ones + ([0] * len(token_ids_0)) + ([0] * len(token_ids_1)) + suffix_ones
+
+    def build_inputs_with_special_tokens(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
+        adding special tokens. An NLLB sequence has the following format, where `X` represents the sequence:
+
+        - `input_ids` (for encoder) `X [eos, src_lang_code]`
+        - `decoder_input_ids`: (for decoder) `X [eos, tgt_lang_code]`
+
+        BOS is never used. Pairs of sequences are not the expected use case, but they will be handled without a
+        separator.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs to which the special tokens will be added.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
+        """
+        if token_ids_1 is None:
+            return self.prefix_tokens + token_ids_0 + self.suffix_tokens
+        # We don't expect to process pairs, but leave the pair logic for API consistency
+        return self.prefix_tokens + token_ids_0 + token_ids_1 + self.suffix_tokens
+
+    def create_token_type_ids_from_sequences(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Create a mask from the two sequences passed to be used in a sequence-pair classification task. nllb does not
+        make use of token type ids, therefore a list of zeros is returned.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of zeros.
+
+        """
+
+        sep = [self.sep_token_id]
+        cls = [self.cls_token_id]
+
+        if token_ids_1 is None:
+            return len(cls + token_ids_0 + sep) * [0]
+        return len(cls + token_ids_0 + sep + sep + token_ids_1 + sep) * [0]
+
+    def _build_translation_inputs(
+        self, raw_inputs, return_tensors: str, src_lang: Optional[str], tgt_lang: Optional[str], **extra_kwargs
+    ):
+        """Used by translation pipeline, to prepare inputs for the generate function"""
+        if src_lang is None or tgt_lang is None:
+            raise ValueError("Translation requires a `src_lang` and a `tgt_lang` for this model")
+        self.src_lang = src_lang
+        inputs = self(raw_inputs, add_special_tokens=True, return_tensors=return_tensors, **extra_kwargs)
+        tgt_lang_id = self.convert_tokens_to_ids(tgt_lang)
+        inputs["forced_bos_token_id"] = tgt_lang_id
+        return inputs
+
+    def get_vocab(self):
+        vocab = {self.convert_ids_to_tokens(i): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text: str) -> List[str]:
+        return self.sp_model.encode(text, out_type=str)
+
+    def _convert_token_to_id(self, token):
+        """Converts a token (str) in an id using the vocab."""
+        if token in self.fairseq_tokens_to_ids:
+            return self.fairseq_tokens_to_ids[token]
+        spm_id = self.sp_model.PieceToId(token)
+
+        # Need to return unknown token if the SP model returned 0
+        return spm_id + self.fairseq_offset if spm_id else self.unk_token_id
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        if index in self.fairseq_ids_to_tokens:
+            return self.fairseq_ids_to_tokens[index]
+        return self.sp_model.IdToPiece(index - self.fairseq_offset)
+
+    def convert_tokens_to_string(self, tokens):
+        """Converts a sequence of tokens (strings for sub-words) in a single string."""
+        out_string = "".join(tokens).replace(SPIECE_UNDERLINE, " ").strip()
+        return out_string
+
+    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
+        if not os.path.isdir(save_directory):
+            logger.error(f"Vocabulary path ({save_directory}) should be a directory")
+            return
+        out_vocab_file = os.path.join(
+            save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
+        )
+
+        if os.path.abspath(self.vocab_file) != os.path.abspath(out_vocab_file) and os.path.isfile(self.vocab_file):
+            copyfile(self.vocab_file, out_vocab_file)
+        elif not os.path.isfile(self.vocab_file):
+            with open(out_vocab_file, "wb") as fi:
+                content_spiece_model = self.sp_model.serialized_model_proto()
+                fi.write(content_spiece_model)
+
+        return (out_vocab_file,)
+
+    def prepare_seq2seq_batch(
+        self,
+        src_texts: List[str],
+        src_lang: str = "eng_Latn",
+        tgt_texts: Optional[List[str]] = None,
+        tgt_lang: str = "fra_Latn",
+        **kwargs,
+    ) -> BatchEncoding:
+        self.src_lang = src_lang
+        self.tgt_lang = tgt_lang
+        return super().prepare_seq2seq_batch(src_texts, tgt_texts, **kwargs)
+
+    @contextmanager
+    def as_target_tokenizer(self):
+        """
+        Temporarily sets the tokenizer for encoding the targets. Useful for tokenizer associated to
+        sequence-to-sequence models that need a slightly different processing for the labels.
+        """
+        self.set_tgt_lang_special_tokens(self.tgt_lang)
+        yield
+        self.set_src_lang_special_tokens(self.src_lang)
+
+    def set_src_lang_special_tokens(self, src_lang) -> None:
+        """Reset the special tokens to the source lang setting. No prefix and suffix=[eos, src_lang_code]."""
+        self.cur_lang_code = self.lang_code_to_id[src_lang]
+        self.prefix_tokens = []
+        self.suffix_tokens = [self.eos_token_id, self.cur_lang_code]
+
+    def set_tgt_lang_special_tokens(self, lang: str) -> None:
+        """Reset the special tokens to the target language setting. No prefix and suffix=[eos, tgt_lang_code]."""
+        self.cur_lang_code = self.lang_code_to_id[lang]
+        self.prefix_tokens = []
+        self.suffix_tokens = [self.eos_token_id, self.cur_lang_code]

src/transformers/models/nllb/tokenization_nllb_fast.py

+# coding=utf-8
+# Copyright 2022 The Facebook AI Research Team Authors and The 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.
+
+import os
+from contextlib import contextmanager
+from shutil import copyfile
+from typing import List, Optional, Tuple
+
+from tokenizers import processors
+
+from ...tokenization_utils import AddedToken, BatchEncoding
+from ...tokenization_utils_fast import PreTrainedTokenizerFast
+from ...utils import is_sentencepiece_available, logging
+
+
+if is_sentencepiece_available():
+    from .tokenization_nllb import NllbTokenizer
+else:
+    NllbTokenizer = None
+
+
+logger = logging.get_logger(__name__)
+
+
+VOCAB_FILES_NAMES = {"vocab_file": "sentencepiece.bpe.model", "tokenizer_file": "tokenizer.json"}
+
+PRETRAINED_VOCAB_FILES_MAP = {
+    "vocab_file": {
+        "facebook/nllb-200-distilled-600M": (
+            "https://huggingface.co/facebook/nllb-200-distilled-600M/resolve/main/sentencepiece.bpe.model"
+        ),
+    },
+    "tokenizer_file": {
+        "facebook/nllb-200-distilled-600M": (
+            "https://huggingface.co/facebook/nllb-200-distilled-600M/resolve/main/tokenizer.json"
+        ),
+    },
+}
+
+PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
+    "facebook/nllb-large-en-ro": 1024,
+    "facebook/nllb-200-distilled-600M": 1024,
+}
+
+# fmt: off
+FAIRSEQ_LANGUAGE_CODES = ['ace_Arab', 'ace_Latn', 'acm_Arab', 'acq_Arab', 'aeb_Arab', 'afr_Latn', 'ajp_Arab', 'aka_Latn', 'amh_Ethi', 'apc_Arab', 'arb_Arab', 'ars_Arab', 'ary_Arab', 'arz_Arab', 'asm_Beng', 'ast_Latn', 'awa_Deva', 'ayr_Latn', 'azb_Arab', 'azj_Latn', 'bak_Cyrl', 'bam_Latn', 'ban_Latn', 'bel_Cyrl', 'bem_Latn', 'ben_Beng', 'bho_Deva', 'bjn_Arab', 'bjn_Latn', 'bod_Tibt', 'bos_Latn', 'bug_Latn', 'bul_Cyrl', 'cat_Latn', 'ceb_Latn', 'ces_Latn', 'cjk_Latn', 'ckb_Arab', 'crh_Latn', 'cym_Latn', 'dan_Latn', 'deu_Latn', 'dik_Latn', 'dyu_Latn', 'dzo_Tibt', 'ell_Grek', 'eng_Latn', 'epo_Latn', 'est_Latn', 'eus_Latn', 'ewe_Latn', 'fao_Latn', 'pes_Arab', 'fij_Latn', 'fin_Latn', 'fon_Latn', 'fra_Latn', 'fur_Latn', 'fuv_Latn', 'gla_Latn', 'gle_Latn', 'glg_Latn', 'grn_Latn', 'guj_Gujr', 'hat_Latn', 'hau_Latn', 'heb_Hebr', 'hin_Deva', 'hne_Deva', 'hrv_Latn', 'hun_Latn', 'hye_Armn', 'ibo_Latn', 'ilo_Latn', 'ind_Latn', 'isl_Latn', 'ita_Latn', 'jav_Latn', 'jpn_Jpan', 'kab_Latn', 'kac_Latn', 'kam_Latn', 'kan_Knda', 'kas_Arab', 'kas_Deva', 'kat_Geor', 'knc_Arab', 'knc_Latn', 'kaz_Cyrl', 'kbp_Latn', 'kea_Latn', 'khm_Khmr', 'kik_Latn', 'kin_Latn', 'kir_Cyrl', 'kmb_Latn', 'kon_Latn', 'kor_Hang', 'kmr_Latn', 'lao_Laoo', 'lvs_Latn', 'lij_Latn', 'lim_Latn', 'lin_Latn', 'lit_Latn', 'lmo_Latn', 'ltg_Latn', 'ltz_Latn', 'lua_Latn', 'lug_Latn', 'luo_Latn', 'lus_Latn', 'mag_Deva', 'mai_Deva', 'mal_Mlym', 'mar_Deva', 'min_Latn', 'mkd_Cyrl', 'plt_Latn', 'mlt_Latn', 'mni_Beng', 'khk_Cyrl', 'mos_Latn', 'mri_Latn', 'zsm_Latn', 'mya_Mymr', 'nld_Latn', 'nno_Latn', 'nob_Latn', 'npi_Deva', 'nso_Latn', 'nus_Latn', 'nya_Latn', 'oci_Latn', 'gaz_Latn', 'ory_Orya', 'pag_Latn', 'pan_Guru', 'pap_Latn', 'pol_Latn', 'por_Latn', 'prs_Arab', 'pbt_Arab', 'quy_Latn', 'ron_Latn', 'run_Latn', 'rus_Cyrl', 'sag_Latn', 'san_Deva', 'sat_Beng', 'scn_Latn', 'shn_Mymr', 'sin_Sinh', 'slk_Latn', 'slv_Latn', 'smo_Latn', 'sna_Latn', 'snd_Arab', 'som_Latn', 'sot_Latn', 'spa_Latn', 'als_Latn', 'srd_Latn', 'srp_Cyrl', 'ssw_Latn', 'sun_Latn', 'swe_Latn', 'swh_Latn', 'szl_Latn', 'tam_Taml', 'tat_Cyrl', 'tel_Telu', 'tgk_Cyrl', 'tgl_Latn', 'tha_Thai', 'tir_Ethi', 'taq_Latn', 'taq_Tfng', 'tpi_Latn', 'tsn_Latn', 'tso_Latn', 'tuk_Latn', 'tum_Latn', 'tur_Latn', 'twi_Latn', 'tzm_Tfng', 'uig_Arab', 'ukr_Cyrl', 'umb_Latn', 'urd_Arab', 'uzn_Latn', 'vec_Latn', 'vie_Latn', 'war_Latn', 'wol_Latn', 'xho_Latn', 'ydd_Hebr', 'yor_Latn', 'yue_Hant', 'zho_Hans', 'zho_Hant', 'zul_Latn']
+# fmt: on
+
+
+class NllbTokenizerFast(PreTrainedTokenizerFast):
+    """
+    Construct a "fast" NLLB tokenizer (backed by HuggingFace's *tokenizers* library). Based on
+    [BPE](https://huggingface.co/docs/tokenizers/python/latest/components.html?highlight=BPE#models).
+
+    This tokenizer inherits from [`PreTrainedTokenizerFast`] which contains most of the main methods. Users should
+    refer to this superclass for more information regarding those methods.
+
+    The tokenization method is `<tokens> <eos> <language code>` for source language documents, and `<language code>
+    <tokens> <eos>` for target language documents.
+
+    Examples:
+
+    ```python
+    >>> from transformers import NllbTokenizerFast
+
+    >>> tokenizer = NllbTokenizerFast.from_pretrained(
+    ...     "facebook/nllb-200-distilled-600M", src_lang="eng_Latn", tgt_lang="fra_Latn"
+    ... )
+    >>> example_english_phrase = " UN Chief Says There Is No Military Solution in Syria"
+    >>> expected_translation_french = "Le chef de l'ONU affirme qu'il n'y a pas de solution militaire en Syrie."
+    >>> inputs = tokenizer(example_english_phrase, return_tensors="pt")
+    >>> with tokenizer.as_target_tokenizer():
+    ...     labels = tokenizer(expected_translation_french, return_tensors="pt")
+    >>> inputs["labels"] = labels["input_ids"]
+    ```
+
+    Args:
+        vocab_file (`str`):
+            Path to the vocabulary file.
+        bos_token (`str`, *optional*, defaults to `"<s>"`):
+            The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token.
+
+            <Tip>
+
+            When building a sequence using special tokens, this is not the token that is used for the beginning of
+            sequence. The token used is the `cls_token`.
+
+            </Tip>
+
+        eos_token (`str`, *optional*, defaults to `"</s>"`):
+            The end of sequence token.
+
+            <Tip>
+
+            When building a sequence using special tokens, this is not the token that is used for the end of sequence.
+            The token used is the `sep_token`.
+
+            </Tip>
+
+        sep_token (`str`, *optional*, defaults to `"</s>"`):
+            The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
+            sequence classification or for a text and a question for question answering. It is also used as the last
+            token of a sequence built with special tokens.
+        cls_token (`str`, *optional*, defaults to `"<s>"`):
+            The classifier token which is used when doing sequence classification (classification of the whole sequence
+            instead of per-token classification). It is the first token of the sequence when built with special tokens.
+        unk_token (`str`, *optional*, defaults to `"<unk>"`):
+            The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
+            token instead.
+        pad_token (`str`, *optional*, defaults to `"<pad>"`):
+            The token used for padding, for example when batching sequences of different lengths.
+        mask_token (`str`, *optional*, defaults to `"<mask>"`):
+            The token used for masking values. This is the token used when training this model with masked language
+            modeling. This is the token which the model will try to predict.
+        tokenizer_file (`str`, *optional*):
+            The path to a tokenizer file to use instead of the vocab file.
+        src_lang (`str`, *optional*):
+            The language to use as source language for translation.
+        tgt_lang (`str`, *optional*):
+            The language to use as target language for translation.
+    """
+
+    vocab_files_names = VOCAB_FILES_NAMES
+    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
+    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
+    model_input_names = ["input_ids", "attention_mask"]
+    slow_tokenizer_class = NllbTokenizer
+
+    prefix_tokens: List[int] = []
+    suffix_tokens: List[int] = []
+
+    def __init__(
+        self,
+        vocab_file=None,
+        tokenizer_file=None,
+        bos_token="<s>",
+        eos_token="</s>",
+        sep_token="</s>",
+        cls_token="<s>",
+        unk_token="<unk>",
+        pad_token="<pad>",
+        mask_token="<mask>",
+        src_lang=None,
+        tgt_lang=None,
+        additional_special_tokens=None,
+        **kwargs
+    ):
+        # Mask token behave like a normal word, i.e. include the space before it
+        mask_token = AddedToken(mask_token, lstrip=True, rstrip=False) if isinstance(mask_token, str) else mask_token
+
+        super().__init__(
+            vocab_file=vocab_file,
+            tokenizer_file=tokenizer_file,
+            bos_token=bos_token,
+            eos_token=eos_token,
+            sep_token=sep_token,
+            cls_token=cls_token,
+            unk_token=unk_token,
+            pad_token=pad_token,
+            mask_token=mask_token,
+            src_lang=src_lang,
+            tgt_lang=tgt_lang,
+            additional_special_tokens=additional_special_tokens,
+            **kwargs,
+        )
+
+        self.vocab_file = vocab_file
+        self.can_save_slow_tokenizer = False if not self.vocab_file else True
+
+        _additional_special_tokens = FAIRSEQ_LANGUAGE_CODES.copy()
+
+        if additional_special_tokens is not None:
+            # Only add those special tokens if they are not already there.
+            _additional_special_tokens.extend(
+                [t for t in additional_special_tokens if t not in _additional_special_tokens]
+            )
+
+        self.add_special_tokens({"additional_special_tokens": _additional_special_tokens})
+        self.lang_code_to_id = {
+            lang_code: self.convert_tokens_to_ids(lang_code) for lang_code in FAIRSEQ_LANGUAGE_CODES
+        }
+
+        self._src_lang = src_lang if src_lang is not None else "eng_Latn"
+        self.cur_lang_code = self.convert_tokens_to_ids(self._src_lang)
+        self.tgt_lang = tgt_lang
+        self.set_src_lang_special_tokens(self._src_lang)
+
+    @property
+    def src_lang(self) -> str:
+        return self._src_lang
+
+    @src_lang.setter
+    def src_lang(self, new_src_lang: str) -> None:
+        self._src_lang = new_src_lang
+        self.set_src_lang_special_tokens(self._src_lang)
+
+    def build_inputs_with_special_tokens(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
+        adding special tokens. The special tokens depend on calling set_lang.
+
+        An NLLB sequence has the following format, where `X` represents the sequence:
+
+        - `input_ids` (for encoder) `X [eos, src_lang_code]`
+        - `decoder_input_ids`: (for decoder) `X [eos, tgt_lang_code]`
+
+        BOS is never used. Pairs of sequences are not the expected use case, but they will be handled without a
+        separator.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs to which the special tokens will be added.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: list of [input IDs](../glossary#input-ids) with the appropriate special tokens.
+        """
+        if token_ids_1 is None:
+            return self.prefix_tokens + token_ids_0 + self.suffix_tokens
+        # We don't expect to process pairs, but leave the pair logic for API consistency
+        return self.prefix_tokens + token_ids_0 + token_ids_1 + self.suffix_tokens
+
+    def create_token_type_ids_from_sequences(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Create a mask from the two sequences passed to be used in a sequence-pair classification task. nllb does not
+        make use of token type ids, therefore a list of zeros is returned.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of zeros.
+
+        """
+
+        sep = [self.sep_token_id]
+        cls = [self.cls_token_id]
+
+        if token_ids_1 is None:
+            return len(cls + token_ids_0 + sep) * [0]
+        return len(cls + token_ids_0 + sep + sep + token_ids_1 + sep) * [0]
+
+    def _build_translation_inputs(
+        self, raw_inputs, return_tensors: str, src_lang: Optional[str], tgt_lang: Optional[str], **extra_kwargs
+    ):
+        """Used by translation pipeline, to prepare inputs for the generate function"""
+        if src_lang is None or tgt_lang is None:
+            raise ValueError("Translation requires a `src_lang` and a `tgt_lang` for this model")
+        self.src_lang = src_lang
+        inputs = self(raw_inputs, add_special_tokens=True, return_tensors=return_tensors, **extra_kwargs)
+        tgt_lang_id = self.convert_tokens_to_ids(tgt_lang)
+        inputs["forced_bos_token_id"] = tgt_lang_id
+        return inputs
+
+    def prepare_seq2seq_batch(
+        self,
+        src_texts: List[str],
+        src_lang: str = "eng_Latn",
+        tgt_texts: Optional[List[str]] = None,
+        tgt_lang: str = "fra_Latn",
+        **kwargs,
+    ) -> BatchEncoding:
+        self.src_lang = src_lang
+        self.tgt_lang = tgt_lang
+        return super().prepare_seq2seq_batch(src_texts, tgt_texts, **kwargs)
+
+    @contextmanager
+    def as_target_tokenizer(self):
+        """
+        Temporarily sets the tokenizer for encoding the targets. Useful for tokenizer associated to
+        sequence-to-sequence models that need a slightly different processing for the labels.
+        """
+        self.set_tgt_lang_special_tokens(self.tgt_lang)
+        yield
+        self.set_src_lang_special_tokens(self.src_lang)
+
+    def set_src_lang_special_tokens(self, src_lang) -> None:
+        """Reset the special tokens to the source lang setting. No prefix and suffix=[eos, src_lang_code]."""
+        self.cur_lang_code = self.convert_tokens_to_ids(src_lang)
+        self.prefix_tokens = []
+        self.suffix_tokens = [self.eos_token_id, self.cur_lang_code]
+
+        prefix_tokens_str = self.convert_ids_to_tokens(self.prefix_tokens)
+        suffix_tokens_str = self.convert_ids_to_tokens(self.suffix_tokens)
+
+        self._tokenizer.post_processor = processors.TemplateProcessing(
+            single=prefix_tokens_str + ["$A"] + suffix_tokens_str,
+            pair=prefix_tokens_str + ["$A", "$B"] + suffix_tokens_str,
+            special_tokens=list(zip(prefix_tokens_str + suffix_tokens_str, self.prefix_tokens + self.suffix_tokens)),
+        )
+
+    def set_tgt_lang_special_tokens(self, lang: str) -> None:
+        """Reset the special tokens to the target language setting. No prefix and suffix=[eos, tgt_lang_code]."""
+        self.cur_lang_code = self.convert_tokens_to_ids(lang)
+        self.prefix_tokens = []
+        self.suffix_tokens = [self.eos_token_id, self.cur_lang_code]
+
+        prefix_tokens_str = self.convert_ids_to_tokens(self.prefix_tokens)
+        suffix_tokens_str = self.convert_ids_to_tokens(self.suffix_tokens)
+
+        self._tokenizer.post_processor = processors.TemplateProcessing(
+            single=prefix_tokens_str + ["$A"] + suffix_tokens_str,
+            pair=prefix_tokens_str + ["$A", "$B"] + suffix_tokens_str,
+            special_tokens=list(zip(prefix_tokens_str + suffix_tokens_str, self.prefix_tokens + self.suffix_tokens)),
+        )
+
+    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
+        if not self.can_save_slow_tokenizer:
+            raise ValueError(
+                "Your fast tokenizer does not have the necessary information to save the vocabulary for a slow "
+                "tokenizer."
+            )
+
+        if not os.path.isdir(save_directory):
+            logger.error(f"Vocabulary path ({save_directory}) should be a directory.")
+            return
+        out_vocab_file = os.path.join(
+            save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
+        )
+
+        if os.path.abspath(self.vocab_file) != os.path.abspath(out_vocab_file):
+            copyfile(self.vocab_file, out_vocab_file)
+
+        return (out_vocab_file,)

src/transformers/utils/dummy_sentencepiece_objects.py

@@ -115,6 +115,13 @@ class MT5Tokenizer(metaclass=DummyObject):
         requires_backends(self, ["sentencepiece"])
 
 
+class NllbTokenizer(metaclass=DummyObject):
+    _backends = ["sentencepiece"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["sentencepiece"])
+
+
 class PegasusTokenizer(metaclass=DummyObject):
     _backends = ["sentencepiece"]
 

src/transformers/utils/dummy_tokenizers_objects.py

@@ -269,6 +269,13 @@ class MvpTokenizerFast(metaclass=DummyObject):
         requires_backends(self, ["tokenizers"])
 
 
+class NllbTokenizerFast(metaclass=DummyObject):
+    _backends = ["tokenizers"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["tokenizers"])
+
+
 class OpenAIGPTTokenizerFast(metaclass=DummyObject):
     _backends = ["tokenizers"]
 

tests/models/nllb/test_tokenization_nllb.py

+# Copyright 2022 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.
+
+import shutil
+import tempfile
+import unittest
+
+from transformers import (
+    SPIECE_UNDERLINE,
+    AddedToken,
+    BatchEncoding,
+    NllbTokenizer,
+    NllbTokenizerFast,
+    is_torch_available,
+)
+from transformers.testing_utils import (
+    get_tests_dir,
+    nested_simplify,
+    require_sentencepiece,
+    require_tokenizers,
+    require_torch,
+)
+
+from ...test_tokenization_common import TokenizerTesterMixin
+
+
+SAMPLE_VOCAB = get_tests_dir("fixtures/test_sentencepiece.model")
+
+
+if is_torch_available():
+    from transformers.models.m2m_100.modeling_m2m_100 import shift_tokens_right
+
+EN_CODE = 256047
+RO_CODE = 256145
+
+
+@require_sentencepiece
+@require_tokenizers
+class NllbTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
+    tokenizer_class = NllbTokenizer
+    rust_tokenizer_class = NllbTokenizerFast
+    test_rust_tokenizer = True
+    test_sentencepiece = True
+    from_pretrained_kwargs = {}
+
+    def setUp(self):
+        super().setUp()
+
+        # We have a SentencePiece fixture for testing
+        tokenizer = NllbTokenizer(SAMPLE_VOCAB, keep_accents=True)
+        tokenizer.save_pretrained(self.tmpdirname)
+
+    def test_full_tokenizer(self):
+        tokenizer = NllbTokenizer(SAMPLE_VOCAB, keep_accents=True)
+
+        tokens = tokenizer.tokenize("This is a test")
+        self.assertListEqual(tokens, ["▁This", "▁is", "▁a", "▁t", "est"])
+
+        self.assertListEqual(
+            tokenizer.convert_tokens_to_ids(tokens),
+            [value + tokenizer.fairseq_offset for value in [285, 46, 10, 170, 382]],
+        )
+
+        tokens = tokenizer.tokenize("I was born in 92000, and this is falsé.")
+        self.assertListEqual(
+            tokens,
+            [
+                SPIECE_UNDERLINE + "I",
+                SPIECE_UNDERLINE + "was",
+                SPIECE_UNDERLINE + "b",
+                "or",
+                "n",
+                SPIECE_UNDERLINE + "in",
+                SPIECE_UNDERLINE + "",
+                "9",
+                "2",
+                "0",
+                "0",
+                "0",
+                ",",
+                SPIECE_UNDERLINE + "and",
+                SPIECE_UNDERLINE + "this",
+                SPIECE_UNDERLINE + "is",
+                SPIECE_UNDERLINE + "f",
+                "al",
+                "s",
+                "é",
+                ".",
+            ],
+        )
+        ids = tokenizer.convert_tokens_to_ids(tokens)
+        self.assertListEqual(
+            ids,
+            [
+                value + tokenizer.fairseq_offset
+                for value in [8, 21, 84, 55, 24, 19, 7, 2, 602, 347, 347, 347, 3, 12, 66, 46, 72, 80, 6, 2, 4]
+            ],
+        )
+
+        back_tokens = tokenizer.convert_ids_to_tokens(ids)
+        self.assertListEqual(
+            back_tokens,
+            [
+                SPIECE_UNDERLINE + "I",
+                SPIECE_UNDERLINE + "was",
+                SPIECE_UNDERLINE + "b",
+                "or",
+                "n",
+                SPIECE_UNDERLINE + "in",
+                SPIECE_UNDERLINE + "",
+                "<unk>",
+                "2",
+                "0",
+                "0",
+                "0",
+                ",",
+                SPIECE_UNDERLINE + "and",
+                SPIECE_UNDERLINE + "this",
+                SPIECE_UNDERLINE + "is",
+                SPIECE_UNDERLINE + "f",
+                "al",
+                "s",
+                "<unk>",
+                ".",
+            ],
+        )
+
+    # overwrite from test_tokenization_common to speed up test
+    def test_save_pretrained(self):
+        self.tokenizers_list[0] = (self.rust_tokenizer_class, "hf-internal-testing/tiny-random-nllb", {})
+        for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
+            with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
+                tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
+                tokenizer_p = self.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
+
+                tmpdirname2 = tempfile.mkdtemp()
+
+                tokenizer_r_files = tokenizer_r.save_pretrained(tmpdirname2)
+                tokenizer_p_files = tokenizer_p.save_pretrained(tmpdirname2)
+
+                # Checks it save with the same files + the tokenizer.json file for the fast one
+                self.assertTrue(any("tokenizer.json" in f for f in tokenizer_r_files))
+                tokenizer_r_files = tuple(f for f in tokenizer_r_files if "tokenizer.json" not in f)
+                self.assertSequenceEqual(tokenizer_r_files, tokenizer_p_files)
+
+                # Checks everything loads correctly in the same way
+                tokenizer_rp = tokenizer_r.from_pretrained(tmpdirname2)
+                tokenizer_pp = tokenizer_p.from_pretrained(tmpdirname2)
+
+                # Check special tokens are set accordingly on Rust and Python
+                for key in tokenizer_pp.special_tokens_map:
+                    self.assertTrue(hasattr(tokenizer_rp, key))
+
+                shutil.rmtree(tmpdirname2)
+
+                # Save tokenizer rust, legacy_format=True
+                tmpdirname2 = tempfile.mkdtemp()
+
+                tokenizer_r_files = tokenizer_r.save_pretrained(tmpdirname2, legacy_format=True)
+                tokenizer_p_files = tokenizer_p.save_pretrained(tmpdirname2)
+
+                # Checks it save with the same files
+                self.assertSequenceEqual(tokenizer_r_files, tokenizer_p_files)
+
+                # Checks everything loads correctly in the same way
+                tokenizer_rp = tokenizer_r.from_pretrained(tmpdirname2)
+                tokenizer_pp = tokenizer_p.from_pretrained(tmpdirname2)
+
+                # Check special tokens are set accordingly on Rust and Python
+                for key in tokenizer_pp.special_tokens_map:
+                    self.assertTrue(hasattr(tokenizer_rp, key))
+
+                shutil.rmtree(tmpdirname2)
+
+                # Save tokenizer rust, legacy_format=False
+                tmpdirname2 = tempfile.mkdtemp()
+
+                tokenizer_r_files = tokenizer_r.save_pretrained(tmpdirname2, legacy_format=False)
+                tokenizer_p_files = tokenizer_p.save_pretrained(tmpdirname2)
+
+                # Checks it saved the tokenizer.json file
+                self.assertTrue(any("tokenizer.json" in f for f in tokenizer_r_files))
+
+                # Checks everything loads correctly in the same way
+                tokenizer_rp = tokenizer_r.from_pretrained(tmpdirname2)
+                tokenizer_pp = tokenizer_p.from_pretrained(tmpdirname2)
+
+                # Check special tokens are set accordingly on Rust and Python
+                for key in tokenizer_pp.special_tokens_map:
+                    self.assertTrue(hasattr(tokenizer_rp, key))
+
+                shutil.rmtree(tmpdirname2)
+
+    @require_torch
+    def test_prepare_seq2seq_batch(self):
+        if not self.test_seq2seq:
+            return
+
+        tokenizers = self.get_tokenizers()
+        for tokenizer in tokenizers:
+            with self.subTest(f"{tokenizer.__class__.__name__}"):
+                # Longer text that will definitely require truncation.
+                src_text = [
+                    " UN Chief Says There Is No Military Solution in Syria",
+                    " Secretary-General Ban Ki-moon says his response to Russia's stepped up military support for"
+                    " Syria is that 'there is no military solution' to the nearly five-year conflict and more weapons"
+                    " will only worsen the violence and misery for millions of people.",
+                ]
+                tgt_text = [
+                    "Şeful ONU declară că nu există o soluţie militară în Siria",
+                    "Secretarul General Ban Ki-moon declară că răspunsul său la intensificarea sprijinului militar al"
+                    ' Rusiei pentru Siria este că "nu există o soluţie militară" la conflictul de aproape cinci ani şi'
+                    " că noi arme nu vor face decât să înrăutăţească violenţele şi mizeria pentru milioane de oameni.",
+                ]
+                try:
+                    batch = tokenizer.prepare_seq2seq_batch(
+                        src_texts=src_text,
+                        tgt_texts=tgt_text,
+                        max_length=3,
+                        max_target_length=10,
+                        return_tensors="pt",
+                        src_lang="eng_Latn",
+                        tgt_lang="ron_Latn",
+                    )
+                except NotImplementedError:
+                    return
+                self.assertEqual(batch.input_ids.shape[1], 3)
+                self.assertEqual(batch.labels.shape[1], 10)
+                # max_target_length will default to max_length if not specified
+                batch = tokenizer.prepare_seq2seq_batch(
+                    src_text, tgt_texts=tgt_text, max_length=3, return_tensors="pt"
+                )
+                self.assertEqual(batch.input_ids.shape[1], 3)
+                self.assertEqual(batch.labels.shape[1], 3)
+
+                batch_encoder_only = tokenizer.prepare_seq2seq_batch(
+                    src_texts=src_text, max_length=3, max_target_length=10, return_tensors="pt"
+                )
+                self.assertEqual(batch_encoder_only.input_ids.shape[1], 3)
+                self.assertEqual(batch_encoder_only.attention_mask.shape[1], 3)
+                self.assertNotIn("decoder_input_ids", batch_encoder_only)
+
+    @unittest.skip("Unfortunately way too slow to build a BPE with SentencePiece.")
+    def test_save_slow_from_fast_and_reload_fast(self):
+        pass
+
+    def test_special_tokens_initialization(self):
+        for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
+            with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
+
+                added_tokens = [AddedToken("<special>", lstrip=True)]
+
+                tokenizer_r = self.rust_tokenizer_class.from_pretrained(
+                    pretrained_name, additional_special_tokens=added_tokens, **kwargs
+                )
+                r_output = tokenizer_r.encode("Hey this is a <special> token")
+
+                special_token_id = tokenizer_r.encode("<special>", add_special_tokens=False)[0]
+
+                self.assertTrue(special_token_id in r_output)
+
+                if self.test_slow_tokenizer:
+                    tokenizer_cr = self.rust_tokenizer_class.from_pretrained(
+                        pretrained_name,
+                        additional_special_tokens=added_tokens,
+                        **kwargs,  # , from_slow=True <- unfortunately too slow to convert
+                    )
+                    tokenizer_p = self.tokenizer_class.from_pretrained(
+                        pretrained_name, additional_special_tokens=added_tokens, **kwargs
+                    )
+
+                    p_output = tokenizer_p.encode("Hey this is a <special> token")
+
+                    cr_output = tokenizer_cr.encode("Hey this is a <special> token")
+
+                    self.assertEqual(p_output, r_output)
+                    self.assertEqual(cr_output, r_output)
+                    self.assertTrue(special_token_id in p_output)
+                    self.assertTrue(special_token_id in cr_output)
+
+
+@require_torch
+@require_sentencepiece
+@require_tokenizers
+class NllbDistilledIntegrationTest(unittest.TestCase):
+    checkpoint_name = "facebook/nllb-200-distilled-600M"
+    src_text = [
+        " UN Chief Says There Is No Military Solution in Syria",
+        """ Secretary-General Ban Ki-moon says his response to Russia's stepped up military support for Syria is that "there is no military solution" to the nearly five-year conflict and more weapons will only worsen the violence and misery for millions of people.""",
+    ]
+    tgt_text = [
+        "Şeful ONU declară că nu există o soluţie militară în Siria",
+        "Secretarul General Ban Ki-moon declară că răspunsul său la intensificarea sprijinului militar al Rusiei"
+        ' pentru Siria este că "nu există o soluţie militară" la conflictul de aproape cinci ani şi că noi arme nu vor'
+        " face decât să înrăutăţească violenţele şi mizeria pentru milioane de oameni.",
+    ]
+    expected_src_tokens = [
+        16297,
+        134408,
+        8165,
+        248066,
+        14734,
+        950,
+        1135,
+        105721,
+        3573,
+        83,
+        27352,
+        108,
+        49486,
+        2,
+        256047,
+    ]
+
+    @classmethod
+    def setUpClass(cls):
+        cls.tokenizer: NllbTokenizer = NllbTokenizer.from_pretrained(
+            cls.checkpoint_name, src_lang="eng_Latn", tgt_lang="ron_Latn"
+        )
+        cls.pad_token_id = 1
+        return cls
+
+    def test_language_codes(self):
+        self.assertEqual(self.tokenizer.fairseq_tokens_to_ids["ace_Arab"], 256001)
+        self.assertEqual(self.tokenizer.fairseq_tokens_to_ids["ace_Latn"], 256002)
+        self.assertEqual(self.tokenizer.fairseq_tokens_to_ids["fra_Latn"], 256057)
+
+    def test_enro_tokenizer_batch_encode_plus(self):
+        ids = self.tokenizer.batch_encode_plus(self.src_text).input_ids[0]
+        self.assertListEqual(self.expected_src_tokens, ids)
+
+    def test_enro_tokenizer_decode_ignores_language_codes(self):
+        self.assertIn(RO_CODE, self.tokenizer.all_special_ids)
+        # fmt: off
+        generated_ids = [RO_CODE, 4254, 98068, 112923, 39072, 3909, 713, 102767, 26, 17314, 35642, 14683, 33118, 2022, 66987, 2, 256047]
+        # fmt: on
+
+        result = self.tokenizer.decode(generated_ids, skip_special_tokens=True)
+        expected_romanian = self.tokenizer.decode(generated_ids[1:], skip_special_tokens=True)
+        self.assertEqual(result, expected_romanian)
+        self.assertNotIn(self.tokenizer.eos_token, result)
+
+    def test_enro_tokenizer_truncation(self):
+        src_text = ["this is gunna be a long sentence " * 20]
+        assert isinstance(src_text[0], str)
+        desired_max_length = 10
+        ids = self.tokenizer(src_text, max_length=desired_max_length, truncation=True).input_ids[0]
+        self.assertEqual(ids[-2], 2)
+        self.assertEqual(ids[-1], EN_CODE)
+        self.assertEqual(len(ids), desired_max_length)
+
+    def test_mask_token(self):
+        self.assertListEqual(self.tokenizer.convert_tokens_to_ids(["<mask>", "ar_AR"]), [256203, 3])
+
+    def test_special_tokens_unaffacted_by_save_load(self):
+        tmpdirname = tempfile.mkdtemp()
+        original_special_tokens = self.tokenizer.fairseq_tokens_to_ids
+        self.tokenizer.save_pretrained(tmpdirname)
+        new_tok = NllbTokenizer.from_pretrained(tmpdirname)
+        self.assertDictEqual(new_tok.fairseq_tokens_to_ids, original_special_tokens)
+
+    @require_torch
+    def test_enro_tokenizer_prepare_batch(self):
+        batch = self.tokenizer(
+            self.src_text, padding=True, truncation=True, max_length=len(self.expected_src_tokens), return_tensors="pt"
+        )
+        with self.tokenizer.as_target_tokenizer():
+            targets = self.tokenizer(
+                self.tgt_text,
+                padding=True,
+                truncation=True,
+                max_length=len(self.expected_src_tokens),
+                return_tensors="pt",
+            )
+        labels = targets["input_ids"]
+        batch["decoder_input_ids"] = shift_tokens_right(
+            labels, self.tokenizer.pad_token_id, self.tokenizer.lang_code_to_id["ron_Latn"]
+        )
+
+        self.assertIsInstance(batch, BatchEncoding)
+
+        self.assertEqual((2, 15), batch.input_ids.shape)
+        self.assertEqual((2, 15), batch.attention_mask.shape)
+        result = batch.input_ids.tolist()[0]
+        self.assertListEqual(self.expected_src_tokens, result)
+        self.assertEqual(2, batch.decoder_input_ids[0, -1])  # EOS
+        # Test that special tokens are reset
+        self.assertEqual(self.tokenizer.prefix_tokens, [])
+        self.assertEqual(self.tokenizer.suffix_tokens, [self.tokenizer.eos_token_id, EN_CODE])
+
+    def test_seq2seq_max_length(self):
+        batch = self.tokenizer(self.src_text, padding=True, truncation=True, max_length=3, return_tensors="pt")
+        with self.tokenizer.as_target_tokenizer():
+            targets = self.tokenizer(self.tgt_text, padding=True, truncation=True, max_length=10, return_tensors="pt")
+        labels = targets["input_ids"]
+        batch["decoder_input_ids"] = shift_tokens_right(
+            labels,
+            self.tokenizer.pad_token_id,
+            decoder_start_token_id=self.tokenizer.lang_code_to_id[self.tokenizer.tgt_lang],
+        )
+
+        self.assertEqual(batch.input_ids.shape[1], 3)
+        self.assertEqual(batch.decoder_input_ids.shape[1], 10)
+
+    @require_torch
+    def test_tokenizer_translation(self):
+        inputs = self.tokenizer._build_translation_inputs(
+            "A test", return_tensors="pt", src_lang="eng_Latn", tgt_lang="fra_Latn"
+        )
+
+        self.assertEqual(
+            nested_simplify(inputs),
+            {
+                # A, test, EOS, en_XX
+                "input_ids": [[70, 7356, 2, 256047]],
+                "attention_mask": [[1, 1, 1, 1]],
+                # ar_AR
+                "forced_bos_token_id": 256057,
+            },
+        )