diff --git a/docs/source/index.rst b/docs/source/index.rst index a4b22744525c26..08e65b7549b642 100644 --- a/docs/source/index.rst +++ b/docs/source/index.rst @@ -3,9 +3,9 @@ Transformers State-of-the-art Natural Language Processing for Pytorch and TensorFlow 2.0. -🤗 Transformers (formerly known as `pytorch-transformers` and `pytorch-pretrained-bert`) provides general-purpose -architectures (BERT, GPT-2, RoBERTa, XLM, DistilBert, XLNet...) for Natural Language Understanding (NLU) and Natural -Language Generation (NLG) with over 32+ pretrained models in 100+ languages and deep interoperability between +🤗 Transformers (formerly known as `pytorch-transformers` and `pytorch-pretrained-bert`) provides general-purpose +architectures (BERT, GPT-2, RoBERTa, XLM, DistilBert, XLNet...) for Natural Language Understanding (NLU) and Natural +Language Generation (NLG) with over 32+ pretrained models in 100+ languages and deep interoperability between TensorFlow 2.0 and PyTorch. This is the documentation of our repository `transformers `_. @@ -127,7 +127,7 @@ conversion utilities for the following models: 23. `Pegasus `_ (from Google) released with the paper `PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization `_ by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu. 24. `MBart `_ (from Facebook) released with the paper `Multilingual Denoising Pre-training for Neural Machine Translation `_ by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, - Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer. + Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer. 25. `LXMERT `_ (from UNC Chapel Hill) released with the paper `LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering `_ by Hao Tan and Mohit Bansal. @@ -223,6 +223,7 @@ conversion utilities for the following models: model_doc/dpr model_doc/pegasus model_doc/mbart + model_doc/fsmt model_doc/funnel model_doc/lxmert model_doc/bertgeneration diff --git a/docs/source/model_doc/fsmt.rst b/docs/source/model_doc/fsmt.rst new file mode 100644 index 00000000000000..ca808ef21fce41 --- /dev/null +++ b/docs/source/model_doc/fsmt.rst @@ -0,0 +1,49 @@ +FSMT +---------------------------------------------------- +**DISCLAIMER:** If you see something strange, +file a `Github Issue `__ and assign +@stas00. + +Overview +~~~~~~~~~~~~~~~~~~~~~ + +FSMT (FairSeq MachineTranslation) models were introduced in "Facebook FAIR's WMT19 News Translation Task Submission" __ by Nathan Ng, Kyra Yee, Alexei Baevski, Myle Ott, Michael Auli, Sergey Edunov. + +The abstract of the paper is the following: + + This paper describes Facebook FAIR's submission to the WMT19 shared news translation task. We participate in two language pairs and four language directions, English <-> German and English <-> Russian. Following our submission from last year, our baseline systems are large BPE-based transformer models trained with the Fairseq sequence modeling toolkit which rely on sampled back-translations. This year we experiment with different bitext data filtering schemes, as well as with adding filtered back-translated data. We also ensemble and fine-tune our models on domain-specific data, then decode using noisy channel model reranking. Our submissions are ranked first in all four directions of the human evaluation campaign. On En->De, our system significantly outperforms other systems as well as human translations. This system improves upon our WMT'18 submission by 4.5 BLEU points. + +The original code can be found here __. + +Implementation Notes +~~~~~~~~~~~~~~~~~~~~ + +- FSMT uses source and target vocab pair, that aren't combined into one. It doesn't share embed tokens either. Its tokenizer is very similar to `XLMTokenizer` and the main model is derived from `BartModel`. + + +FSMTForConditionalGeneration +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. autoclass:: transformers.FSMTForConditionalGeneration + :members: forward + + +FSMTConfig +~~~~~~~~~~~~~~~~~~~~~ + +.. autoclass:: transformers.FSMTConfig + :members: + + +FSMTTokenizer +~~~~~~~~~~~~~~~~~~~~~ + +.. autoclass:: transformers.FSMTTokenizer + :members: + + +FSMTModel +~~~~~~~~~~~~~ + +.. autoclass:: transformers.FSMTModel + :members: forward diff --git a/examples/seq2seq/test_data/fsmt/build-eval-data.py b/examples/seq2seq/test_data/fsmt/build-eval-data.py new file mode 100755 index 00000000000000..46487c07ea8432 --- /dev/null +++ b/examples/seq2seq/test_data/fsmt/build-eval-data.py @@ -0,0 +1,33 @@ +#!/usr/bin/env python + +import io +import json +import subprocess + + +pairs = [ + ["en", "ru"], + ["ru", "en"], + ["en", "de"], + ["de", "en"], +] + +n_objs = 8 + + +def get_all_data(pairs, n_objs): + text = {} + for src, tgt in pairs: + pair = f"{src}-{tgt}" + cmd = f"sacrebleu -t wmt19 -l {pair} --echo src".split() + src_lines = subprocess.run(cmd, stdout=subprocess.PIPE).stdout.decode("utf-8").splitlines() + cmd = f"sacrebleu -t wmt19 -l {pair} --echo ref".split() + tgt_lines = subprocess.run(cmd, stdout=subprocess.PIPE).stdout.decode("utf-8").splitlines() + text[pair] = {"src": src_lines[:n_objs], "tgt": tgt_lines[:n_objs]} + return text + + +text = get_all_data(pairs, n_objs) +filename = "./fsmt_val_data.json" +with io.open(filename, "w", encoding="utf-8") as f: + bleu_data = json.dump(text, f, indent=2, ensure_ascii=False) diff --git a/examples/seq2seq/test_data/fsmt/fsmt_val_data.json b/examples/seq2seq/test_data/fsmt/fsmt_val_data.json new file mode 100644 index 00000000000000..f38b305733314a --- /dev/null +++ b/examples/seq2seq/test_data/fsmt/fsmt_val_data.json @@ -0,0 +1,90 @@ +{ + "en-ru": { + "src": [ + "Welsh AMs worried about 'looking like muppets'", + "There is consternation among some AMs at a suggestion their title should change to MWPs (Member of the Welsh Parliament).", + "It has arisen because of plans to change the name of the assembly to the Welsh Parliament.", + "AMs across the political spectrum are worried it could invite ridicule.", + "One Labour AM said his group was concerned \"it rhymes with Twp and Pwp.\"", + "For readers outside of Wales: In Welsh twp means daft and pwp means poo.", + "A Plaid AM said the group as a whole was \"not happy\" and has suggested alternatives.", + "A Welsh Conservative said his group was \"open minded\" about the name change, but noted it was a short verbal hop from MWP to Muppet." + ], + "tgt": [ + "Члены Национальной ассамблеи Уэльса обеспокоены, что \"выглядят как куклы\"", + "Некоторые члены Национальной ассамблеи Уэльса в ужасе от предложения о том, что их наименование должно измениться на MPW (члены Парламента Уэльса).", + "Этот вопрос был поднят в связи с планами по переименованию ассамблеи в Парламент Уэльса.", + "Члены Национальной ассамблеи Уэльса всего политического спектра обеспокоены, что это может породить насмешки.", + "Один из лейбористских членов Национальной ассамблеи Уэльса сказал, что его партия обеспокоена тем, что \"это рифмуется с Twp и Pwp\".", + "Для читателей за предлами Уэльса: по-валлийски twp означает \"глупый\", а pwp означает \"какашка\".", + "Член Национальной ассамблеи от Плайд сказал, что эта партия в целом \"не счастлива\" и предложил альтернативы.", + "Представитель Консервативной партии Уэльса сказал, что его партия \"открыта\" к переименованию, но отметил, что между WMP и Muppet небольшая разница в произношении." + ] + }, + "ru-en": { + "src": [ + "Названо число готовящихся к отправке в Донбасс новобранцев из Украины", + "Официальный представитель Народной милиции самопровозглашенной Луганской Народной Республики (ЛНР) Андрей Марочко заявил, что зимой 2018-2019 года Украина направит в Донбасс не менее 3 тыс. новобранцев.", + "По его словам, таким образом Киев планирует \"хоть как-то доукомплектовать подразделения\".", + "\"Нежелание граждан Украины проходить службу в рядах ВС Украины, массовые увольнения привели к низкой укомплектованности подразделений\", - рассказал Марочко, которого цитирует \"РИА Новости\".", + "Он также не исключил, что реальные цифры призванных в армию украинцев могут быть увеличены в случае необходимости.", + "В 2014-2017 годах Киев начал так называемую антитеррористическую операцию (АТО), которую позже сменили на операцию объединенных сил (ООС).", + "Предполагалось, что эта мера приведет к усилению роли украинских силовиков в урегулировании ситуации.", + "В конце августа 2018 года ситуация в Донбассе обострилась из-за убийства главы ДНР Александра Захарченко." + ], + "tgt": [ + "The number of new Ukrainian recruits ready to go to Donbass has become public", + "Official representative of the peoples’ militia of the self-proclaimed Lugansk People’s Republic Andrey Marochko claimed that Ukrainian will send at least 3 thousand new recruits to Donbass in winter 2018-2019.", + "This is how Kyiv tries “at least somehow to staff the units,” he said.", + "“The unwillingness of Ukrainian citizens to serve in the Ukraine’s military forces, mass resignments lead to low understaffing,” said Marochko cited by RIA Novosti.", + "Also, he doesn’t exclude that the real numbers of conscripts in the Ukrainian army can be raised is necessary.", + "In 2014-2017, Kyiv started so-called antiterrorist operation, that ws later changed to the united forces operation.", + "This measure was supposed to strengthen the role of the Ukrainian military in settling the situation.", + "In the late August 2018, the situation in Donbass escalated as the DNR head Aleksandr Zakharchenko was killed." + ] + }, + "en-de": { + "src": [ + "Welsh AMs worried about 'looking like muppets'", + "There is consternation among some AMs at a suggestion their title should change to MWPs (Member of the Welsh Parliament).", + "It has arisen because of plans to change the name of the assembly to the Welsh Parliament.", + "AMs across the political spectrum are worried it could invite ridicule.", + "One Labour AM said his group was concerned \"it rhymes with Twp and Pwp.\"", + "For readers outside of Wales: In Welsh twp means daft and pwp means poo.", + "A Plaid AM said the group as a whole was \"not happy\" and has suggested alternatives.", + "A Welsh Conservative said his group was \"open minded\" about the name change, but noted it was a short verbal hop from MWP to Muppet." + ], + "tgt": [ + "Walisische Ageordnete sorgen sich \"wie Dödel auszusehen\"", + "Es herrscht Bestürzung unter einigen Mitgliedern der Versammlung über einen Vorschlag, der ihren Titel zu MWPs (Mitglied der walisischen Parlament) ändern soll.", + "Der Grund dafür waren Pläne, den Namen der Nationalversammlung in Walisisches Parlament zu ändern.", + "Mitglieder aller Parteien der Nationalversammlung haben Bedenken, dass sie sich dadurch Spott aussetzen könnten.", + "Ein Labour-Abgeordneter sagte, dass seine Gruppe \"sich mit Twp und Pwp reimt\".", + "Hinweis für den Leser: „twp“ im Walisischen bedeutet „bescheuert“ und „pwp“ bedeutet „Kacke“.", + "Ein Versammlungsmitglied von Plaid Cymru sagte, die Gruppe als Ganzes sei \"nicht glücklich\" und hat Alternativen vorgeschlagen.", + "Ein walisischer Konservativer sagte, seine Gruppe wäre „offen“ für eine Namensänderung, wies aber darauf hin, dass es von „MWP“ (Mitglied des Walisischen Parlaments) nur ein kurzer verbaler Sprung zu „Muppet“ ist." + ] + }, + "de-en": { + "src": [ + "Schöne Münchnerin 2018: Schöne Münchnerin 2018 in Hvar: Neun Dates", + "Von az, aktualisiert am 04.05.2018 um 11:11", + "Ja, sie will...", + "\"Schöne Münchnerin\" 2018 werden!", + "Am Nachmittag wartet erneut eine Überraschung auf unsere Kandidatinnen: sie werden das romantische Candlelight-Shooting vor der MY SOLARIS nicht alleine bestreiten, sondern an der Seite von Male-Model Fabian!", + "Hvar - Flirten, kokettieren, verführen - keine einfachen Aufgaben für unsere Mädchen.", + "Insbesondere dann, wenn in Deutschland ein Freund wartet.", + "Dennoch liefern die neun \"Schöne Münchnerin\"-Kandidatinnen beim Shooting mit People-Fotograf Tuan ab und trotzen Wind, Gischt und Regen wie echte Profis." + ], + "tgt": [ + "The Beauty of Munich 2018: the Beauty of Munich 2018 in Hvar: Nine dates", + "From A-Z, updated on 04/05/2018 at 11:11", + "Yes, she wants to...", + "to become \"The Beauty of Munich\" in 2018!", + "In the afternoon there is another surprise waiting for our contestants: they will be competing for the romantic candlelight photo shoot at MY SOLARIS not alone, but together with a male-model Fabian!", + "Hvar with its flirting, coquetting, and seduction is not an easy task for our girls.", + "Especially when there is a boyfriend waiting in Germany.", + "Despite dealing with wind, sprays and rain, the nine contestants of \"The Beauty of Munich\" behaved like real professionals at the photo shoot with People-photographer Tuan." + ] + } +} \ No newline at end of file diff --git a/examples/seq2seq/test_fsmt_bleu_score.py b/examples/seq2seq/test_fsmt_bleu_score.py new file mode 100644 index 00000000000000..95f475698f6f8f --- /dev/null +++ b/examples/seq2seq/test_fsmt_bleu_score.py @@ -0,0 +1,77 @@ +# coding=utf-8 +# Copyright 2020 Huggingface +# +# 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 io +import unittest + + +try: + from .utils import calculate_bleu +except ImportError: + from utils import calculate_bleu + +import json + +from parameterized import parameterized +from transformers import FSMTForConditionalGeneration, FSMTTokenizer +from transformers.testing_utils import get_tests_dir, require_torch, slow, torch_device + + +filename = get_tests_dir() + "/test_data/fsmt/fsmt_val_data.json" +with io.open(filename, "r", encoding="utf-8") as f: + bleu_data = json.load(f) + + +@require_torch +class ModelEvalTester(unittest.TestCase): + def get_tokenizer(self, mname): + return FSMTTokenizer.from_pretrained(mname) + + def get_model(self, mname): + model = FSMTForConditionalGeneration.from_pretrained(mname).to(torch_device) + if torch_device == "cuda": + model.half() + return model + + @parameterized.expand( + [ + ["en-ru", 26.0], + ["ru-en", 22.0], + ["en-de", 22.0], + ["de-en", 29.0], + ] + ) + @slow + def test_bleu_scores(self, pair, min_bleu_score): + # note: this test is not testing the best performance since it only evals a small batch + # but it should be enough to detect a regression in the output quality + mname = f"facebook/wmt19-{pair}" + tokenizer = self.get_tokenizer(mname) + model = self.get_model(mname) + + src_sentences = bleu_data[pair]["src"] + tgt_sentences = bleu_data[pair]["tgt"] + + batch = tokenizer(src_sentences, return_tensors="pt", truncation=True, padding="longest").to(torch_device) + outputs = model.generate( + input_ids=batch.input_ids, + num_beams=8, + ) + decoded_sentences = tokenizer.batch_decode( + outputs, skip_special_tokens=True, clean_up_tokenization_spaces=False + ) + scores = calculate_bleu(decoded_sentences, tgt_sentences) + print(scores) + self.assertGreaterEqual(scores["bleu"], min_bleu_score) diff --git a/model_cards/facebook/wmt19-de-en/README.md b/model_cards/facebook/wmt19-de-en/README.md new file mode 100644 index 00000000000000..4cbb36e2e7ac66 --- /dev/null +++ b/model_cards/facebook/wmt19-de-en/README.md @@ -0,0 +1,94 @@ + +--- + + + +language: de, en +thumbnail: +tags: +- translation +- wmt19 +license: Apache 2.0 +datasets: +- http://www.statmt.org/wmt19/ ([test-set](http://matrix.statmt.org/test_sets/newstest2019.tgz?1556572561)) +metrics: +- http://www.statmt.org/wmt19/metrics-task.html +--- + +# FSMT + +## Model description + +This is a ported version of [fairseq wmt19 transformer](https://github.com/pytorch/fairseq/blob/master/examples/wmt19/README.md) for de-en. + +For more details, please see, [Facebook FAIR's WMT19 News Translation Task Submission](https://arxiv.org/abs/1907.06616). + +The abbreviation FSMT stands for FairSeqMachineTranslation + +All four models are available: + +* [wmt19-en-ru](https://huggingface.co/facebook/wmt19-en-ru) +* [wmt19-ru-en](https://huggingface.co/facebook/wmt19-ru-en) +* [wmt19-en-de](https://huggingface.co/facebook/wmt19-en-de) +* [wmt19-de-en](https://huggingface.co/facebook/wmt19-de-en) + +## Intended uses & limitations + +#### How to use + +```python +from transformers.tokenization_fsmt import FSMTTokenizer +from transformers.modeling_fsmt import FSMTForConditionalGeneration +mname = "facebook/wmt19-de-en" +tokenizer = FSMTTokenizer.from_pretrained(mname) +model = FSMTForConditionalGeneration.from_pretrained(mname) + +input = "Maschinelles Lernen ist großartig, oder?" +input_ids = tokenizer.encode(input, return_tensors="pt") +outputs = model.generate(input_ids) +decoded = tokenizer.decode(outputs[0], skip_special_tokens=True) +print(decoded) # Machine learning is great, isn't it? + +``` + +#### Limitations and bias + +- The original (and this ported model) doesn't seem to handle well inputs with repeated sub-phrases, [content gets truncated](https://discuss.huggingface.co/t/issues-with-translating-inputs-containing-repeated-phrases/981) + +## Training data + +Pretrained weights were left identical to the original model released by fairseq. For more details, please, see the [paper](https://arxiv.org/abs/1907.06616). + +## Eval results + +pair | fairseq | transformers +-------|---------|---------- +de-en | [42.3](http://matrix.statmt.org/matrix/output/1902?run_id=6750) | 41.35 + +The score is slightly below the score reported by `fairseq`, since `transformers`` currently doesn't support: +- model ensemble, therefore the best performing checkpoint was ported (``model4.pt``). +- re-ranking + +The score was calculated using this code: + +```bash +git clone https://github.com/huggingface/transformers +cd transformers +export PAIR=de-en +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=15 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS +``` +note: fairseq reports using a beam of 50, so you should get a slightly higher score if re-run with `--num_beams 50`. + + +## TODO + +- port model ensemble (fairseq uses 4 model checkpoints) + diff --git a/model_cards/facebook/wmt19-en-de/README.md b/model_cards/facebook/wmt19-en-de/README.md new file mode 100644 index 00000000000000..426082eaac3ae5 --- /dev/null +++ b/model_cards/facebook/wmt19-en-de/README.md @@ -0,0 +1,94 @@ + +--- + + + +language: en, de +thumbnail: +tags: +- translation +- wmt19 +license: Apache 2.0 +datasets: +- http://www.statmt.org/wmt19/ ([test-set](http://matrix.statmt.org/test_sets/newstest2019.tgz?1556572561)) +metrics: +- http://www.statmt.org/wmt19/metrics-task.html +--- + +# FSMT + +## Model description + +This is a ported version of [fairseq wmt19 transformer](https://github.com/pytorch/fairseq/blob/master/examples/wmt19/README.md) for en-de. + +For more details, please see, [Facebook FAIR's WMT19 News Translation Task Submission](https://arxiv.org/abs/1907.06616). + +The abbreviation FSMT stands for FairSeqMachineTranslation + +All four models are available: + +* [wmt19-en-ru](https://huggingface.co/facebook/wmt19-en-ru) +* [wmt19-ru-en](https://huggingface.co/facebook/wmt19-ru-en) +* [wmt19-en-de](https://huggingface.co/facebook/wmt19-en-de) +* [wmt19-de-en](https://huggingface.co/facebook/wmt19-de-en) + +## Intended uses & limitations + +#### How to use + +```python +from transformers.tokenization_fsmt import FSMTTokenizer +from transformers.modeling_fsmt import FSMTForConditionalGeneration +mname = "facebook/wmt19-en-de" +tokenizer = FSMTTokenizer.from_pretrained(mname) +model = FSMTForConditionalGeneration.from_pretrained(mname) + +input = "Machine learning is great, isn't it?" +input_ids = tokenizer.encode(input, return_tensors="pt") +outputs = model.generate(input_ids) +decoded = tokenizer.decode(outputs[0], skip_special_tokens=True) +print(decoded) # Maschinelles Lernen ist großartig, oder? + +``` + +#### Limitations and bias + +- The original (and this ported model) doesn't seem to handle well inputs with repeated sub-phrases, [content gets truncated](https://discuss.huggingface.co/t/issues-with-translating-inputs-containing-repeated-phrases/981) + +## Training data + +Pretrained weights were left identical to the original model released by fairseq. For more details, please, see the [paper](https://arxiv.org/abs/1907.06616). + +## Eval results + +pair | fairseq | transformers +-------|---------|---------- +en-de | [43.1](http://matrix.statmt.org/matrix/output/1909?run_id=6862) | 42.83 + +The score is slightly below the score reported by `fairseq`, since `transformers`` currently doesn't support: +- model ensemble, therefore the best performing checkpoint was ported (``model4.pt``). +- re-ranking + +The score was calculated using this code: + +```bash +git clone https://github.com/huggingface/transformers +cd transformers +export PAIR=en-de +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=15 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS +``` +note: fairseq reports using a beam of 50, so you should get a slightly higher score if re-run with `--num_beams 50`. + + +## TODO + +- port model ensemble (fairseq uses 4 model checkpoints) + diff --git a/model_cards/facebook/wmt19-en-ru/README.md b/model_cards/facebook/wmt19-en-ru/README.md new file mode 100644 index 00000000000000..26999e652539d2 --- /dev/null +++ b/model_cards/facebook/wmt19-en-ru/README.md @@ -0,0 +1,94 @@ + +--- + + + +language: en, ru +thumbnail: +tags: +- translation +- wmt19 +license: Apache 2.0 +datasets: +- http://www.statmt.org/wmt19/ ([test-set](http://matrix.statmt.org/test_sets/newstest2019.tgz?1556572561)) +metrics: +- http://www.statmt.org/wmt19/metrics-task.html +--- + +# FSMT + +## Model description + +This is a ported version of [fairseq wmt19 transformer](https://github.com/pytorch/fairseq/blob/master/examples/wmt19/README.md) for en-ru. + +For more details, please see, [Facebook FAIR's WMT19 News Translation Task Submission](https://arxiv.org/abs/1907.06616). + +The abbreviation FSMT stands for FairSeqMachineTranslation + +All four models are available: + +* [wmt19-en-ru](https://huggingface.co/facebook/wmt19-en-ru) +* [wmt19-ru-en](https://huggingface.co/facebook/wmt19-ru-en) +* [wmt19-en-de](https://huggingface.co/facebook/wmt19-en-de) +* [wmt19-de-en](https://huggingface.co/facebook/wmt19-de-en) + +## Intended uses & limitations + +#### How to use + +```python +from transformers.tokenization_fsmt import FSMTTokenizer +from transformers.modeling_fsmt import FSMTForConditionalGeneration +mname = "facebook/wmt19-en-ru" +tokenizer = FSMTTokenizer.from_pretrained(mname) +model = FSMTForConditionalGeneration.from_pretrained(mname) + +input = "Machine learning is great, isn't it?" +input_ids = tokenizer.encode(input, return_tensors="pt") +outputs = model.generate(input_ids) +decoded = tokenizer.decode(outputs[0], skip_special_tokens=True) +print(decoded) # Машинное обучение - это здорово, не так ли? + +``` + +#### Limitations and bias + +- The original (and this ported model) doesn't seem to handle well inputs with repeated sub-phrases, [content gets truncated](https://discuss.huggingface.co/t/issues-with-translating-inputs-containing-repeated-phrases/981) + +## Training data + +Pretrained weights were left identical to the original model released by fairseq. For more details, please, see the [paper](https://arxiv.org/abs/1907.06616). + +## Eval results + +pair | fairseq | transformers +-------|---------|---------- +en-ru | [36.4](http://matrix.statmt.org/matrix/output/1914?run_id=6724) | 33.47 + +The score is slightly below the score reported by `fairseq`, since `transformers`` currently doesn't support: +- model ensemble, therefore the best performing checkpoint was ported (``model4.pt``). +- re-ranking + +The score was calculated using this code: + +```bash +git clone https://github.com/huggingface/transformers +cd transformers +export PAIR=en-ru +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=15 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS +``` +note: fairseq reports using a beam of 50, so you should get a slightly higher score if re-run with `--num_beams 50`. + + +## TODO + +- port model ensemble (fairseq uses 4 model checkpoints) + diff --git a/model_cards/facebook/wmt19-ru-en/README.md b/model_cards/facebook/wmt19-ru-en/README.md new file mode 100644 index 00000000000000..a4071e4101a53c --- /dev/null +++ b/model_cards/facebook/wmt19-ru-en/README.md @@ -0,0 +1,94 @@ + +--- + + + +language: ru, en +thumbnail: +tags: +- translation +- wmt19 +license: Apache 2.0 +datasets: +- http://www.statmt.org/wmt19/ ([test-set](http://matrix.statmt.org/test_sets/newstest2019.tgz?1556572561)) +metrics: +- http://www.statmt.org/wmt19/metrics-task.html +--- + +# FSMT + +## Model description + +This is a ported version of [fairseq wmt19 transformer](https://github.com/pytorch/fairseq/blob/master/examples/wmt19/README.md) for ru-en. + +For more details, please see, [Facebook FAIR's WMT19 News Translation Task Submission](https://arxiv.org/abs/1907.06616). + +The abbreviation FSMT stands for FairSeqMachineTranslation + +All four models are available: + +* [wmt19-en-ru](https://huggingface.co/facebook/wmt19-en-ru) +* [wmt19-ru-en](https://huggingface.co/facebook/wmt19-ru-en) +* [wmt19-en-de](https://huggingface.co/facebook/wmt19-en-de) +* [wmt19-de-en](https://huggingface.co/facebook/wmt19-de-en) + +## Intended uses & limitations + +#### How to use + +```python +from transformers.tokenization_fsmt import FSMTTokenizer +from transformers.modeling_fsmt import FSMTForConditionalGeneration +mname = "facebook/wmt19-ru-en" +tokenizer = FSMTTokenizer.from_pretrained(mname) +model = FSMTForConditionalGeneration.from_pretrained(mname) + +input = "Машинное обучение - это здорово, не так ли?" +input_ids = tokenizer.encode(input, return_tensors="pt") +outputs = model.generate(input_ids) +decoded = tokenizer.decode(outputs[0], skip_special_tokens=True) +print(decoded) # Machine learning is great, isn't it? + +``` + +#### Limitations and bias + +- The original (and this ported model) doesn't seem to handle well inputs with repeated sub-phrases, [content gets truncated](https://discuss.huggingface.co/t/issues-with-translating-inputs-containing-repeated-phrases/981) + +## Training data + +Pretrained weights were left identical to the original model released by fairseq. For more details, please, see the [paper](https://arxiv.org/abs/1907.06616). + +## Eval results + +pair | fairseq | transformers +-------|---------|---------- +ru-en | [41.3](http://matrix.statmt.org/matrix/output/1907?run_id=6937) | 39.20 + +The score is slightly below the score reported by `fairseq`, since `transformers`` currently doesn't support: +- model ensemble, therefore the best performing checkpoint was ported (``model4.pt``). +- re-ranking + +The score was calculated using this code: + +```bash +git clone https://github.com/huggingface/transformers +cd transformers +export PAIR=ru-en +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=15 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS +``` +note: fairseq reports using a beam of 50, so you should get a slightly higher score if re-run with `--num_beams 50`. + + +## TODO + +- port model ensemble (fairseq uses 4 model checkpoints) + diff --git a/src/transformers/__init__.py b/src/transformers/__init__.py index 06f3f2e153b167..652c01bb47dea5 100755 --- a/src/transformers/__init__.py +++ b/src/transformers/__init__.py @@ -30,6 +30,7 @@ from .configuration_electra import ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, ElectraConfig from .configuration_encoder_decoder import EncoderDecoderConfig from .configuration_flaubert import FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, FlaubertConfig +from .configuration_fsmt import FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP, FSMTConfig from .configuration_funnel import FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP, FunnelConfig from .configuration_gpt2 import GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP, GPT2Config from .configuration_longformer import LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, LongformerConfig @@ -158,6 +159,7 @@ ) from .tokenization_electra import ElectraTokenizer, ElectraTokenizerFast from .tokenization_flaubert import FlaubertTokenizer +from .tokenization_fsmt import FSMTTokenizer from .tokenization_funnel import FunnelTokenizer, FunnelTokenizerFast from .tokenization_gpt2 import GPT2Tokenizer, GPT2TokenizerFast from .tokenization_longformer import LongformerTokenizer, LongformerTokenizerFast @@ -338,6 +340,7 @@ FlaubertModel, FlaubertWithLMHeadModel, ) + from .modeling_fsmt import FSMTForConditionalGeneration, FSMTModel, PretrainedFSMTModel from .modeling_funnel import ( FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST, FunnelBaseModel, diff --git a/src/transformers/configuration_auto.py b/src/transformers/configuration_auto.py index a89916932edcaf..9f429e3307f365 100644 --- a/src/transformers/configuration_auto.py +++ b/src/transformers/configuration_auto.py @@ -27,6 +27,7 @@ from .configuration_electra import ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, ElectraConfig from .configuration_encoder_decoder import EncoderDecoderConfig from .configuration_flaubert import FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, FlaubertConfig +from .configuration_fsmt import FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP, FSMTConfig from .configuration_funnel import FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP, FunnelConfig from .configuration_gpt2 import GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP, GPT2Config from .configuration_longformer import LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, LongformerConfig @@ -66,6 +67,7 @@ T5_PRETRAINED_CONFIG_ARCHIVE_MAP, XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, + FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP, ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, @@ -94,6 +96,7 @@ ("longformer", LongformerConfig), ("roberta", RobertaConfig), ("flaubert", FlaubertConfig), + ("fsmt", FSMTConfig), ("bert", BertConfig), ("openai-gpt", OpenAIGPTConfig), ("gpt2", GPT2Config), @@ -126,6 +129,7 @@ ("longformer", "Longformer"), ("roberta", "RoBERTa"), ("flaubert", "FlauBERT"), + ("fsmt", "FairSeq Machine-Translation"), ("bert", "BERT"), ("openai-gpt", "OpenAI GPT"), ("gpt2", "OpenAI GPT-2"), diff --git a/src/transformers/configuration_fsmt.py b/src/transformers/configuration_fsmt.py new file mode 100644 index 00000000000000..a9dd0efe430197 --- /dev/null +++ b/src/transformers/configuration_fsmt.py @@ -0,0 +1,223 @@ +# coding=utf-8 +# Copyright 2019-present, Facebook, Inc 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. +""" FSMT configuration """ + + +import copy + +from .configuration_utils import PretrainedConfig +from .file_utils import add_start_docstrings_to_callable +from .utils import logging + + +logger = logging.get_logger(__name__) + +FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP = {} + + +FSMT_CONFIG_ARGS_DOC = r""" + Args: + langs (:obj:`List[str]`): + source language, target_language (e.g. ['en', 'ru']) + src_vocab_size (:obj:`int`): + defines the different tokens that can be represented by `inputs_ids` passed to the forward + method in the encoder. + tgt_vocab_size (:obj:`int`): + defines the different tokens that can be represented by `inputs_ids` passed to the forward + method in the decoder. + d_model (:obj:`int`, `optional`, defaults to 1024): + Dimensionality of the layers and the pooler layer. + encoder_layers (:obj:`int`, `optional`, defaults to 12): + Number of encoder layers, 16 for pegasus, 6 for bart-base and marian + decoder_layers (:obj:`int`, `optional`, defaults to 12): + Number of decoder layers, 16 for pegasus, 6 for bart-base and marian + encoder_attention_heads (:obj:`int`, `optional`, defaults to 16): + Number of attention heads for each attention layer in the Transformer encoder. + decoder_attention_heads (:obj:`int`, `optional`, defaults to 16): + Number of attention heads for each attention layer in the Transformer decoder. + decoder_ffn_dim (:obj:`int`, `optional`, defaults to 4096): + Dimensionality of the "intermediate" (i.e., feed-forward) layer in decoder. + encoder_ffn_dim (:obj:`int`, `optional`, defaults to 4096): + Dimensionality of the "intermediate" (i.e., feed-forward) layer in decoder. + activation_function (:obj:`str` or :obj:`function`, `optional`, defaults to "relu"): + The non-linear activation function (function or string) in the encoder and pooler. + If string, "gelu", "relu", "swish" and "gelu_new" are supported. + dropout (:obj:`float`, `optional`, defaults to 0.1): + The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler. + attention_dropout (:obj:`float`, `optional`, defaults to 0.0): + The dropout ratio for the attention probabilities. + activation_dropout (:obj:`float`, `optional`, defaults to 0.0): + The dropout ratio for activations inside the fully connected layer. + max_position_embeddings (:obj:`int`, `optional`, defaults to 1024): + The maximum sequence length that this model might ever be used with. + Typically set this to something large just in case (e.g., 512 or 1024 or 2048). + init_std (:obj:`float`, `optional`, defaults to 0.02): + The standard deviation of the truncated_normal_initializer for initializing all weight matrices. + scale_embedding (:obj:`bool`, `optional`, defaults to :obj:`True`): + Scale embeddings by diving by sqrt(d_model). + bos_token_id (:obj:`int`, `optional`, defaults to 0) + Beginning of stream token id. + pad_token_id (:obj:`int`, `optional`, defaults to 1) + Padding token id. + eos_token_id (:obj:`int`, `optional`, defaults to 2) + End of stream token id. + decoder_start_token_id (:obj:`int`, `optional`): + This model starts decoding with `eos_token_id` + encoder_layerdrop: (:obj:`float`, `optional`, defaults to 0.0): + Google "layerdrop arxiv", as its not explainable in one line. + decoder_layerdrop: (:obj:`float`, `optional`, defaults to 0.0): + Google "layerdrop arxiv", as its not explainable in one line. + is_encoder_decoder (:obj:`bool`, `optional`, defaults to :obj:`True`): + Whether this is an encoder/decoder model. + tie_word_embeddings (:obj:`bool`, `optional`, defaults to :obj:`False`): + Whether to tie input and output embeddings. + num_beams (:obj:`int`, `optional`, defaults to 5) + Number of beams for beam search that will be used by default in the :obj:`generate` method + of the model. 1 means no beam search. + length_penalty (:obj:`float`, `optional`, defaults to 1) + Exponential penalty to the length that will be used by default in the :obj:`generate` method + of the model. + early_stopping (:obj:`bool`, `optional`, defaults to :obj:`False`) + Flag that will be used by default in the :obj:`generate` method of the model. Whether to stop + the beam search when at least ``num_beams`` sentences are finished per batch or not. +""" + + +class DecoderConfig(PretrainedConfig): + r""" + Configuration class for FSMT's decoder specific things. + note: this is a private helper class + """ + model_type = "fsmt_decoder" + + def __init__(self, vocab_size=0, bos_token_id=0): + super().__init__() + self.vocab_size = vocab_size + self.bos_token_id = bos_token_id + + +@add_start_docstrings_to_callable(FSMT_CONFIG_ARGS_DOC) +class FSMTConfig(PretrainedConfig): + r""" + Configuration class for FSMT. + """ + model_type = "fsmt" + + # update the defaults from config file + def __init__( + self, + langs, + src_vocab_size, + tgt_vocab_size, + activation_function="relu", + d_model=1024, + max_length=200, + max_position_embeddings=1024, + encoder_ffn_dim=4096, + encoder_layers=12, + encoder_attention_heads=16, + encoder_layerdrop=0.0, + decoder_ffn_dim=4096, + decoder_layers=12, + decoder_attention_heads=16, + decoder_layerdrop=0.0, + attention_dropout=0.0, + dropout=0.1, + activation_dropout=0.0, + init_std=0.02, + pad_token_id=1, + bos_token_id=0, + eos_token_id=2, + decoder_start_token_id=2, + is_encoder_decoder=True, + scale_embedding=True, + tie_word_embeddings=False, + num_beams=5, + length_penalty=1.0, + early_stopping=False, + **common_kwargs + ): + r""" + :class:`~transformers.FSMTConfig` is the configuration class for `FSMTModel`. + + Examples:: + + >>> from transformers import FSMTConfig, FSMTModel + + >>> config = FSMTConfig.from_pretrained('facebook/wmt19-en-ru') + >>> model = FSMTModel(config) + + """ + if "hidden_size" in common_kwargs: + raise ValueError("hidden size is called d_model") + super().__init__( + pad_token_id=pad_token_id, + bos_token_id=bos_token_id, + eos_token_id=eos_token_id, + decoder_start_token_id=decoder_start_token_id, + is_encoder_decoder=is_encoder_decoder, + tie_word_embeddings=tie_word_embeddings, + **common_kwargs, + ) + self.langs = langs + self.src_vocab_size = src_vocab_size + self.tgt_vocab_size = tgt_vocab_size + self.d_model = d_model # encoder_embed_dim and decoder_embed_dim + self.max_length = max_length + + self.encoder_ffn_dim = encoder_ffn_dim + self.encoder_layers = self.num_hidden_layers = encoder_layers + self.encoder_attention_heads = encoder_attention_heads + self.encoder_layerdrop = encoder_layerdrop + self.decoder_layerdrop = decoder_layerdrop + self.decoder_ffn_dim = decoder_ffn_dim + self.decoder_layers = decoder_layers + self.decoder_attention_heads = decoder_attention_heads + self.max_position_embeddings = max_position_embeddings + self.init_std = init_std # Normal(0, this parameter) + self.activation_function = activation_function + + self.num_beams = num_beams + self.length_penalty = length_penalty + self.early_stopping = early_stopping + + self.decoder = DecoderConfig(vocab_size=tgt_vocab_size, bos_token_id=eos_token_id) + + self.scale_embedding = scale_embedding # scale factor will be sqrt(d_model) if True + + # 3 Types of Dropout + self.attention_dropout = attention_dropout + self.activation_dropout = activation_dropout + self.dropout = dropout + + @property + def num_attention_heads(self) -> int: + return self.encoder_attention_heads + + @property + def hidden_size(self) -> int: + return self.d_model + + def to_dict(self): + """ + Serializes this instance to a Python dictionary. Override the default `to_dict()` from `PretrainedConfig`. + + Returns: + :obj:`Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance, + """ + output = copy.deepcopy(self.__dict__) + output["decoder"] = self.decoder.to_dict() + output["model_type"] = self.__class__.model_type + return output diff --git a/src/transformers/convert_fsmt_original_pytorch_checkpoint_to_pytorch.py b/src/transformers/convert_fsmt_original_pytorch_checkpoint_to_pytorch.py new file mode 100755 index 00000000000000..c9d29a8f054b46 --- /dev/null +++ b/src/transformers/convert_fsmt_original_pytorch_checkpoint_to_pytorch.py @@ -0,0 +1,271 @@ +# coding=utf-8 +# Copyright 2018 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. + +# Note: if you intend to run this script make sure you look under scripts/fsmt/ +# to locate the appropriate script to do the work correctly. There is a set of scripts to: +# - download and prepare data and run the conversion script +# - perform eval to get the best hparam into the config +# - generate model_cards - useful if you have multiple models from the same paper + +import argparse +import json +import os +import re +from collections import OrderedDict +from os.path import basename, dirname + +import fairseq +import torch +from fairseq import hub_utils +from fairseq.data.dictionary import Dictionary + +from transformers import WEIGHTS_NAME +from transformers.configuration_fsmt import FSMTConfig +from transformers.modeling_fsmt import FSMTForConditionalGeneration +from transformers.tokenization_fsmt import VOCAB_FILES_NAMES +from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE + +from .utils import logging + + +logging.set_verbosity_info() + +json_indent = 2 + +# based on the results of a search on a range of `num_beams`, `length_penalty` and `early_stopping` +# values against wmt19 test data to obtain the best BLEU scores, we will use the following defaults: +# +# * `num_beams`: 5 (higher scores better, but requires more memory/is slower, can be adjusted by users) +# * `early_stopping`: `False` consistently scored better +# * `length_penalty` varied, so will assign the best one depending on the model +best_score_hparams = { + # fairseq: + "wmt19-ru-en": {"length_penalty": 1.1}, + "wmt19-en-ru": {"length_penalty": 1.15}, + "wmt19-en-de": {"length_penalty": 1.0}, + "wmt19-de-en": {"length_penalty": 1.1}, + # allenai: + "wmt16-en-de-dist-12-1": {"length_penalty": 0.6}, + "wmt16-en-de-dist-6-1": {"length_penalty": 0.6}, + "wmt16-en-de-12-1": {"length_penalty": 0.8}, + "wmt19-de-en-6-6-base": {"length_penalty": 0.6}, + "wmt19-de-en-6-6-big": {"length_penalty": 0.6}, +} + +# this remaps the different models to their organization names +org_names = {} +for m in ["wmt19-ru-en", "wmt19-en-ru", "wmt19-en-de", "wmt19-de-en"]: + org_names[m] = "facebook" +for m in [ + "wmt16-en-de-dist-12-1", + "wmt16-en-de-dist-6-1", + "wmt16-en-de-12-1", + "wmt19-de-en-6-6-base", + "wmt19-de-en-6-6-big", +]: + org_names[m] = "allenai" + + +def rewrite_dict_keys(d): + # (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up, + # e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er': 7} + d2 = dict((re.sub(r"@@$", "", k), v) if k.endswith("@@") else (re.sub(r"$", "", k), v) for k, v in d.items()) + keep_keys = " ".split() + # restore the special tokens + for k in keep_keys: + del d2[f"{k}"] + d2[k] = d[k] # restore + return d2 + + +def convert_fsmt_checkpoint_to_pytorch(fsmt_checkpoint_path, pytorch_dump_folder_path): + + # prep + assert os.path.exists(fsmt_checkpoint_path) + os.makedirs(pytorch_dump_folder_path, exist_ok=True) + print(f"Writing results to {pytorch_dump_folder_path}") + + # handle various types of models + + checkpoint_file = basename(fsmt_checkpoint_path) + fsmt_folder_path = dirname(fsmt_checkpoint_path) + + cls = fairseq.model_parallel.models.transformer.ModelParallelTransformerModel + models = cls.hub_models() + kwargs = {"bpe": "fastbpe", "tokenizer": "moses"} + data_name_or_path = "." + # note: since the model dump is old, fairseq has upgraded its model some + # time later, and it does a whole lot of rewrites and splits on the saved + # weights, therefore we can't use torch.load() directly on the model file. + # see: upgrade_state_dict(state_dict) in fairseq_model.py + print(f"using checkpoint {checkpoint_file}") + chkpt = hub_utils.from_pretrained( + fsmt_folder_path, checkpoint_file, data_name_or_path, archive_map=models, **kwargs + ) + + args = dict(vars(chkpt["args"])) + + src_lang = args["source_lang"] + tgt_lang = args["target_lang"] + + data_root = dirname(pytorch_dump_folder_path) + model_dir = basename(pytorch_dump_folder_path) + + # dicts + src_dict_file = os.path.join(fsmt_folder_path, f"dict.{src_lang}.txt") + tgt_dict_file = os.path.join(fsmt_folder_path, f"dict.{tgt_lang}.txt") + + src_dict = Dictionary.load(src_dict_file) + src_vocab = rewrite_dict_keys(src_dict.indices) + src_vocab_size = len(src_vocab) + src_vocab_file = os.path.join(pytorch_dump_folder_path, "vocab-src.json") + print(f"Generating {src_vocab_file}") + with open(src_vocab_file, "w", encoding="utf-8") as f: + f.write(json.dumps(src_vocab, ensure_ascii=False, indent=json_indent)) + + tgt_dict = Dictionary.load(tgt_dict_file) + tgt_vocab = rewrite_dict_keys(tgt_dict.indices) + tgt_vocab_size = len(tgt_vocab) + tgt_vocab_file = os.path.join(pytorch_dump_folder_path, "vocab-tgt.json") + print(f"Generating {tgt_vocab_file}") + with open(tgt_vocab_file, "w", encoding="utf-8") as f: + f.write(json.dumps(tgt_vocab, ensure_ascii=False, indent=json_indent)) + + # merges_file (bpecodes) + merges_file = os.path.join(pytorch_dump_folder_path, VOCAB_FILES_NAMES["merges_file"]) + fsmt_merges_file = os.path.join(fsmt_folder_path, "bpecodes") + with open(fsmt_merges_file, encoding="utf-8") as fin: + merges = fin.read() + merges = re.sub(r" \d+$", "", merges, 0, re.M) # remove frequency number + print(f"Generating {merges_file}") + with open(merges_file, "w", encoding="utf-8") as fout: + fout.write(merges) + + # model config + fsmt_model_config_file = os.path.join(pytorch_dump_folder_path, "config.json") + + # validate bpe/tokenizer config, as currently it's hardcoded to moses+fastbpe - + # may have to modify the tokenizer if a different type is used by a future model + assert args["bpe"] == "fastbpe", f"need to extend tokenizer to support bpe={args['bpe']}" + assert args["tokenizer"] == "moses", f"need to extend tokenizer to support bpe={args['tokenizer']}" + + model_conf = { + "architectures": ["FSMTForConditionalGeneration"], + "model_type": "fsmt", + "activation_dropout": args["activation_dropout"], + "activation_function": "relu", + "attention_dropout": args["attention_dropout"], + "d_model": args["decoder_embed_dim"], + "dropout": args["dropout"], + "init_std": 0.02, + "max_position_embeddings": args["max_source_positions"], + "num_hidden_layers": args["encoder_layers"], + "src_vocab_size": src_vocab_size, + "tgt_vocab_size": tgt_vocab_size, + "langs": [src_lang, tgt_lang], + "encoder_attention_heads": args["encoder_attention_heads"], + "encoder_ffn_dim": args["encoder_ffn_embed_dim"], + "encoder_layerdrop": args["encoder_layerdrop"], + "encoder_layers": args["encoder_layers"], + "decoder_attention_heads": args["decoder_attention_heads"], + "decoder_ffn_dim": args["decoder_ffn_embed_dim"], + "decoder_layerdrop": args["decoder_layerdrop"], + "decoder_layers": args["decoder_layers"], + "bos_token_id": 0, + "pad_token_id": 1, + "eos_token_id": 2, + "is_encoder_decoder": True, + "scale_embedding": not args["no_scale_embedding"], + "tie_word_embeddings": args["share_all_embeddings"], + } + + # good hparam defaults to start with + model_conf["num_beams"] = 5 + model_conf["early_stopping"] = False + if model_dir in best_score_hparams and "length_penalty" in best_score_hparams[model_dir]: + model_conf["length_penalty"] = best_score_hparams[model_dir]["length_penalty"] + else: + model_conf["length_penalty"] = 1.0 + + print(f"Generating {fsmt_model_config_file}") + with open(fsmt_model_config_file, "w", encoding="utf-8") as f: + f.write(json.dumps(model_conf, ensure_ascii=False, indent=json_indent)) + + # tokenizer config + fsmt_tokenizer_config_file = os.path.join(pytorch_dump_folder_path, TOKENIZER_CONFIG_FILE) + + tokenizer_conf = { + "langs": [src_lang, tgt_lang], + "model_max_length": 1024, + } + + print(f"Generating {fsmt_tokenizer_config_file}") + with open(fsmt_tokenizer_config_file, "w", encoding="utf-8") as f: + f.write(json.dumps(tokenizer_conf, ensure_ascii=False, indent=json_indent)) + + # model + model = chkpt["models"][0] + model_state_dict = model.state_dict() + + # rename keys to start with 'model.' + model_state_dict = OrderedDict(("model." + k, v) for k, v in model_state_dict.items()) + + # remove unneeded keys + ignore_keys = [ + "model.model", + "model.encoder.version", + "model.decoder.version", + "model.encoder_embed_tokens.weight", + "model.decoder_embed_tokens.weight", + ] + for k in ignore_keys: + model_state_dict.pop(k, None) + + config = FSMTConfig.from_pretrained(pytorch_dump_folder_path) + model_new = FSMTForConditionalGeneration(config) + + # check that it loads ok + model_new.load_state_dict(model_state_dict, strict=False) + + # save + pytorch_weights_dump_path = os.path.join(pytorch_dump_folder_path, WEIGHTS_NAME) + print(f"Generating {pytorch_weights_dump_path}") + torch.save(model_state_dict, pytorch_weights_dump_path) + + print("Conversion is done!") + print("\nLast step is to upload the files to s3") + print(f"cd {data_root}") + print(f"transformers-cli upload {model_dir}") + print( + "Note: CDN caches files for up to 24h, so either use a local model path " + "or use `from_pretrained(mname, use_cdn=False)` to use the non-cached version." + ) + + +if __name__ == "__main__": + parser = argparse.ArgumentParser() + # Required parameters + parser.add_argument( + "--fsmt_checkpoint_path", + default=None, + type=str, + required=True, + help="Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts, bpecodes, etc.", + ) + parser.add_argument( + "--pytorch_dump_folder_path", default=None, type=str, required=True, help="Path to the output PyTorch model." + ) + args = parser.parse_args() + convert_fsmt_checkpoint_to_pytorch(args.fsmt_checkpoint_path, args.pytorch_dump_folder_path) diff --git a/src/transformers/modeling_auto.py b/src/transformers/modeling_auto.py index fd2371cc8781b6..0ac47a4e9e3aed 100644 --- a/src/transformers/modeling_auto.py +++ b/src/transformers/modeling_auto.py @@ -30,6 +30,7 @@ ElectraConfig, EncoderDecoderConfig, FlaubertConfig, + FSMTConfig, FunnelConfig, GPT2Config, LongformerConfig, @@ -113,6 +114,7 @@ FlaubertModel, FlaubertWithLMHeadModel, ) +from .modeling_fsmt import FSMTForConditionalGeneration, FSMTModel from .modeling_funnel import ( FunnelForMaskedLM, FunnelForMultipleChoice, @@ -211,6 +213,7 @@ (TransfoXLConfig, TransfoXLModel), (XLNetConfig, XLNetModel), (FlaubertConfig, FlaubertModel), + (FSMTConfig, FSMTModel), (XLMConfig, XLMModel), (CTRLConfig, CTRLModel), (ElectraConfig, ElectraModel), @@ -230,6 +233,7 @@ (CamembertConfig, CamembertForMaskedLM), (XLMRobertaConfig, XLMRobertaForMaskedLM), (BartConfig, BartForConditionalGeneration), + (FSMTConfig, FSMTForConditionalGeneration), (LongformerConfig, LongformerForMaskedLM), (RobertaConfig, RobertaForMaskedLM), (BertConfig, BertForPreTraining), @@ -319,6 +323,7 @@ (MarianConfig, MarianMTModel), (MBartConfig, MBartForConditionalGeneration), (BartConfig, BartForConditionalGeneration), + (FSMTConfig, FSMTForConditionalGeneration), (EncoderDecoderConfig, EncoderDecoderModel), ] ) diff --git a/src/transformers/modeling_fsmt.py b/src/transformers/modeling_fsmt.py new file mode 100644 index 00000000000000..a0c914cedef954 --- /dev/null +++ b/src/transformers/modeling_fsmt.py @@ -0,0 +1,1212 @@ +# coding=utf-8 +# Copyright 2020 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. +# +# Original implementation: https://github.com/pytorch/fairseq/tree/master/examples/wmt19 +# Authors: +# - @alexeib Alexei Baevski +# - @edunov Sergey Edunov +# - @michaelauli Michael Auli +# - @myleott Myle Ott +# - @nng555 Nathan Ng +# - David Grangier +# - Kyra Yee +# +# Paper: Facebook FAIR's WMT19 News Translation Task Submission https://arxiv.org/abs/1907.06616 +# +"""PyTorch Fairseq model, ported from https://github.com/pytorch/fairseq/tree/master/examples/wmt19""" + +import math +import random +import warnings +from typing import Any, Dict, List, Optional, Tuple + +import torch +import torch.nn.functional as F +from torch import Tensor, nn +from torch.nn import CrossEntropyLoss + +from .activations import ACT2FN +from .configuration_fsmt import FSMTConfig +from .file_utils import ( + add_code_sample_docstrings, + add_end_docstrings, + add_start_docstrings, + add_start_docstrings_to_callable, + replace_return_docstrings, +) +from .modeling_outputs import BaseModelOutput, BaseModelOutputWithPast, Seq2SeqLMOutput, Seq2SeqModelOutput +from .modeling_utils import PreTrainedModel +from .utils import logging + + +logger = logging.get_logger(__name__) + +_CONFIG_FOR_DOC = "FSMTConfig" +_TOKENIZER_FOR_DOC = "FSMTTokenizer" + +# See all FSMT models at https://huggingface.co/models?search=fsmt + + +# Porting notes: +# this one is modeled after BartModel* +# +# Currently only translation (fairseq also has weights for LM) +# +# fairseq provides weights for ru-en, en-ru and de-en, en-de pairs. All have been ported. +# - ru-en, en-ru use asymmetric vocab +# - de-en, en-de use a merged single vocab (but the code works as if they are separate) +# +# Differences with Bart: +# - not using bos token +# - 2 separate vocabs (src and target) +# - embed weights aren't tied +# - uses a model Ensemble (but that part isn't ported/implemented yet) - so we +# aren't getting as good of a BLEU score +# - uses a projection layer at the end of the decoder +# - doesn't use final_logits_bias +# - beam search: stops as soon as num_beams == len(hypos) (whereas transformers +# is not satisfied there and will continue searching until the next cycles +# aren't promising something better), comparing BLEU scores - the transformers +# algorithm is slightly superior, therefore using the latter. But if you want +# to match fairseq outputs, you need to pass ``early_stopping=True`` to ``generate()``. +# +# SinusoidalPositionalEmbedding is slightly different from Bart's - generates +# different embeddings. This implementation is copied verbatim from fairseq with +# some small changes to make it work here. +# +# Other changes: +# - doesn't support use_cache as Bart's version does +# +# +# FSMTConfig changes with BartConfig +# +# Differences with BART: +# - src/tgt vocabs aren't shared +# - token embeddings aren't shared +# - needs a language pair +# - scale_embedding are True +# +# some unused args were removed too +# +# +# TODO: +# - port model ensemble (fs uses 4 model checkpoints) +# - solve beam search discrepancies + +""" + +Here is how to compare BLEU scores against fairseq implementation: + +# en-ru + +export PAIR=en-ru +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=50 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS + +# (fairseq BLEU: 36.4 http://matrix.statmt.org/matrix/output/1914?score_id=37605) + + +# ru-en + +export PAIR=ru-en +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=50 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS + + +# (fairseq BLEU: 41.3 http://matrix.statmt.org/matrix/output/1907?run_id=6937) + + +# de-en + +export PAIR=de-en +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +export NUM_BEAMS=50 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS + +# (fairseq BLEU: 42.3 http://matrix.statmt.org/matrix/output/1902?run_id=6750) + + + +# en-de + +export PAIR=en-de +export DATA_DIR=data/$PAIR +export SAVE_DIR=data/$PAIR +export BS=8 +mkdir -p $DATA_DIR +sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source +sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target +echo $PAIR +PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py facebook/wmt19-$PAIR $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS + +# (fairseq BLEU: 43.1 http://matrix.statmt.org/matrix/output/1909?run_id=6862) + +""" + + +FSMT_START_DOCSTRING = r""" + + This model is a PyTorch `torch.nn.Module `_ sub-class. Use it as a regular PyTorch Module and + refer to the PyTorch documentation for all matters related to general usage and behavior. + + Parameters: + config (:class:`~transformers.FSMTConfig`): Model configuration class with all the parameters of the model. + Initializing with a config file does not load the weights associated with the model, only the configuration. + Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model weights. + +""" +FSMT_GENERATION_EXAMPLE = r""" + Translation example:: + + from transformers import FSMTTokenizer, FSMTForConditionalGeneration + + mname = "facebook/wmt19-ru-en" + model = FSMTForConditionalGeneration.from_pretrained(mname) + tokenizer = FSMTTokenizer.from_pretrained(mname) + + src_text = "Машинное обучение - это здорово, не так ли?" + input_ids = tokenizer.encode(src_text, return_tensors='pt') + outputs = model.generate(input_ids, num_beams=5, num_return_sequences=3) + for i, output in enumerate(outputs): + decoded = tokenizer.decode(output, skip_special_tokens=True) + print(f"{i}: {decoded}) + # 1: Machine learning is great, isn't it? ... + +""" + +FSMT_INPUTS_DOCSTRING = r""" + Args: + input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`): + Indices of input sequence tokens in the vocabulary. Use FSMTTokenizer.encode to produce them. + Padding will be ignored by default should you provide it. + Indices can be obtained using :class:`transformers.FSMTTokenizer.encode(text)`. + attention_mask (:obj:`torch.Tensor` of shape :obj:`(batch_size, sequence_length)`, `optional`): + Mask to avoid performing attention on padding token indices in input_ids. + Mask values selected in ``[0, 1]``: + ``1`` for tokens that are NOT MASKED, ``0`` for MASKED tokens. + encoder_outputs (:obj:`tuple(tuple(torch.FloatTensor)`, `optional`): + Tuple consists of (`last_hidden_state`, `optional`: `hidden_states`, `optional`: `attentions`) + `last_hidden_state` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`) is a sequence of hidden-states at the output of the last layer of the encoder. + Used in the cross-attention of the decoder. + decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`): + Provide for translation and summarization training. By default, the model will create this tensor by shifting the input_ids right, following the paper. + decoder_attention_mask (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, tgt_seq_len)`, `optional`): + Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default. + If you want to change padding behavior, you should read :func:`~transformers.modeling_fairseqtranslator._prepare_decoder_inputs` and modify. + See diagram 1 in the paper for more info on the default strategy + past_key_values (:obj:`tuple(tuple(torch.FloatTensor))` of length :obj:`config.n_layers` with each tuple having 4 tensors of shape :obj:`(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`): + Contains pre-computed key and value hidden-states of the attention blocks. + Can be used to speed up decoding. + If ``past_key_values`` are used, the user can optionally input only the last + ``decoder_input_ids`` (those that don't have their past key value states given to this model) of shape + :obj:`(batch_size, 1)` instead of all ``decoder_input_ids`` of shape :obj:`(batch_size, sequence_length)`. + use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`): + If `use_cache` is True, ``past_key_values`` are returned and can be used to speed up decoding (see + ``past_key_values``). + output_attentions (:obj:`bool`, `optional`): + If set to ``True``, the attentions tensors of all attention layers are returned. See ``attentions`` under returned tensors for more detail. + output_hidden_states (:obj:`bool`, `optional`): + If set to ``True``, the hidden states of all layers are returned. See ``hidden_states`` under returned tensors for more detail. + return_dict (:obj:`bool`, `optional`): + If set to ``True``, the model will return a :class:`~transformers.file_utils.ModelOutput` instead of a + plain tuple. +""" + + +have_fused_layer_norm = False +if torch.cuda.is_available(): + try: + from apex.normalization import FusedLayerNorm + + have_fused_layer_norm = True + except ImportError: + pass + +LayerNorm = FusedLayerNorm if have_fused_layer_norm else torch.nn.LayerNorm + + +def invert_mask(attention_mask): + """Turns 1->0, 0->1, False->True, True-> False""" + assert attention_mask.dim() == 2 + return attention_mask.eq(0) + + +def _prepare_fsmt_decoder_inputs( + config, input_ids, decoder_input_ids=None, decoder_padding_mask=None, causal_mask_dtype=torch.float32 +): + """Prepare masks that ignore padding tokens in the decoder and a causal mask for the decoder if + none are provided. This mimics the default behavior in fairseq. To override it pass in masks. + Note: this is not called during generation + """ + pad_token_id = config.pad_token_id + if decoder_input_ids is None: + decoder_input_ids = shift_tokens_right(input_ids, pad_token_id) + bsz, tgt_len = decoder_input_ids.size() + if decoder_padding_mask is None: + decoder_padding_mask = make_padding_mask(decoder_input_ids, pad_token_id) + else: + decoder_padding_mask = invert_mask(decoder_padding_mask) + causal_mask = torch.triu(fill_with_neg_inf(torch.zeros(tgt_len, tgt_len)), 1).to( + dtype=causal_mask_dtype, device=decoder_input_ids.device + ) + return decoder_input_ids, decoder_padding_mask, causal_mask + + +class PretrainedFSMTModel(PreTrainedModel): + config_class = FSMTConfig + base_model_prefix = "model" + + def _init_weights(self, module): + std = self.config.init_std + if isinstance(module, nn.Linear): + module.weight.data.normal_(mean=0.0, std=std) + if module.bias is not None: + module.bias.data.zero_() + elif isinstance(module, SinusoidalPositionalEmbedding): + pass + elif isinstance(module, nn.Embedding): + module.weight.data.normal_(mean=0.0, std=std) + if module.padding_idx is not None: + module.weight.data[module.padding_idx].zero_() + + @property + def dummy_inputs(self): + pad_token = self.config.pad_token_id + input_ids = torch.tensor([[0, 6, 10, 4, 2], [0, 8, 12, 2, pad_token]], device=self.device) + dummy_inputs = { + "attention_mask": input_ids.ne(pad_token), + "input_ids": input_ids, + } + return dummy_inputs + + +def _make_linear_from_emb(emb): + vocab_size, emb_size = emb.weight.shape + lin_layer = nn.Linear(vocab_size, emb_size, bias=False) + lin_layer.weight.data = emb.weight.data + return lin_layer + + +# Helper Functions, mostly for making masks +def _check_shapes(shape_1, shape2): + if shape_1 != shape2: + raise AssertionError("shape mismatch: {} != {}".format(shape_1, shape2)) + + +def shift_tokens_right(input_ids, pad_token_id): + """Shift input ids one token to the right, and wrap the last non pad token (usually ).""" + prev_output_tokens = input_ids.clone() + index_of_eos = (input_ids.ne(pad_token_id).sum(dim=1) - 1).unsqueeze(-1) + prev_output_tokens[:, 0] = input_ids.gather(1, index_of_eos).squeeze() + prev_output_tokens[:, 1:] = input_ids[:, :-1] + return prev_output_tokens + + +def make_padding_mask(input_ids, padding_idx=1): + """True for pad tokens""" + padding_mask = input_ids.eq(padding_idx) + if not padding_mask.any(): + padding_mask = None + return padding_mask + + +# Helper Modules + + +class EncoderLayer(nn.Module): + def __init__(self, config: FSMTConfig): + super().__init__() + self.embed_dim = config.d_model + self.self_attn = Attention(self.embed_dim, config.encoder_attention_heads, dropout=config.attention_dropout) + self.self_attn_layer_norm = LayerNorm(self.embed_dim) + self.dropout = config.dropout + self.activation_fn = ACT2FN[config.activation_function] + self.activation_dropout = config.activation_dropout + self.fc1 = nn.Linear(self.embed_dim, config.encoder_ffn_dim) + self.fc2 = nn.Linear(config.encoder_ffn_dim, self.embed_dim) + self.final_layer_norm = LayerNorm(self.embed_dim) + + def forward(self, x, encoder_padding_mask, output_attentions=False): + """ + Args: + x (Tensor): input to the layer of shape `(seq_len, batch, embed_dim)` + encoder_padding_mask (ByteTensor): binary ByteTensor of shape + `(batch, src_len)` where padding elements are indicated by ``1``. + for t_tgt, t_src is excluded (or masked out), =0 means it is + included in attention + + Returns: + encoded output of shape `(seq_len, batch, embed_dim)` + """ + residual = x + x, attn_weights = self.self_attn( + query=x, key=x, key_padding_mask=encoder_padding_mask, output_attentions=output_attentions + ) + x = F.dropout(x, p=self.dropout, training=self.training) + x = residual + x + x = self.self_attn_layer_norm(x) + + residual = x + x = self.activation_fn(self.fc1(x)) + x = F.dropout(x, p=self.activation_dropout, training=self.training) + x = self.fc2(x) + x = F.dropout(x, p=self.dropout, training=self.training) + x = residual + x + x = self.final_layer_norm(x) + return x, attn_weights + + +class FSMTEncoder(nn.Module): + """ + Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer + is a :class:`EncoderLayer`. + + Args: + config: FSMTConfig + """ + + def __init__(self, config: FSMTConfig, embed_tokens): + super().__init__() + + self.dropout = config.dropout + self.layerdrop = config.encoder_layerdrop + + embed_dim = embed_tokens.embedding_dim + self.embed_scale = math.sqrt(embed_dim) if config.scale_embedding else 1.0 + self.padding_idx = embed_tokens.padding_idx + self.max_source_positions = config.max_position_embeddings + + self.embed_tokens = embed_tokens + # print(config.max_position_embeddings, embed_dim, self.padding_idx) + num_embeddings = config.src_vocab_size + self.embed_positions = SinusoidalPositionalEmbedding( + embed_dim, + self.padding_idx, + init_size=num_embeddings + self.padding_idx + 1, # removed: config.max_position_embeddings + ) + self.layers = nn.ModuleList([EncoderLayer(config) for _ in range(config.encoder_layers)]) + + def forward( + self, input_ids, attention_mask=None, output_attentions=False, output_hidden_states=False, return_dict=False + ): + """ + Args: + input_ids (LongTensor): tokens in the source language of shape + `(batch, src_len)` + attention_mask (torch.LongTensor): indicating which indices are padding tokens. + Returns: + BaseModelOutput or Tuple comprised of: + - **x** (Tensor): the last encoder layer's output of + shape `(src_len, batch, embed_dim)` + - **encoder_states** (tuple(torch.FloatTensor)): all intermediate + hidden states of shape `(src_len, batch, embed_dim)`. + Only populated if *output_hidden_states:* is True. + - **all_attentions** (tuple(torch.FloatTensor)): Attention weights for each layer. + During training might not be of length n_layers because of layer dropout. + """ + # check attention mask and invert + if attention_mask is not None: + attention_mask = invert_mask(attention_mask) + + inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale + embed_pos = self.embed_positions(input_ids) + x = inputs_embeds + embed_pos + x = F.dropout(x, p=self.dropout, training=self.training) + + # B x T x C -> T x B x C + x = x.transpose(0, 1) + + encoder_states = [] if output_hidden_states else None + all_attentions = () if output_attentions else None + for encoder_layer in self.layers: + if output_hidden_states: + encoder_states.append(x) + # add LayerDrop (see https://arxiv.org/abs/1909.11556 for description) + dropout_probability = random.uniform(0, 1) + if self.training and (dropout_probability < self.layerdrop): # skip the layer + attn = None + else: + x, attn = encoder_layer(x, attention_mask, output_attentions=output_attentions) + + if output_attentions: + all_attentions = all_attentions + (attn,) + + if output_hidden_states: + encoder_states.append(x) + # T x B x C -> B x T x C + encoder_states = tuple(hidden_state.transpose(0, 1) for hidden_state in encoder_states) + + # T x B x C -> B x T x C + x = x.transpose(0, 1) + + if not return_dict: + return tuple(v for v in [x, encoder_states, all_attentions] if v is not None) + return BaseModelOutput(last_hidden_state=x, hidden_states=encoder_states, attentions=all_attentions) + + +class DecoderLayer(nn.Module): + def __init__(self, config: FSMTConfig): + super().__init__() + self.embed_dim = config.d_model + + self.self_attn = Attention( + embed_dim=self.embed_dim, + num_heads=config.decoder_attention_heads, + dropout=config.attention_dropout, + ) + self.dropout = config.dropout + self.activation_fn = ACT2FN[config.activation_function] + self.activation_dropout = config.activation_dropout + + self.self_attn_layer_norm = LayerNorm(self.embed_dim) + self.encoder_attn = Attention( + self.embed_dim, + config.decoder_attention_heads, + dropout=config.attention_dropout, + encoder_decoder_attention=True, + ) + self.encoder_attn_layer_norm = LayerNorm(self.embed_dim) + self.fc1 = nn.Linear(self.embed_dim, config.decoder_ffn_dim) + self.fc2 = nn.Linear(config.decoder_ffn_dim, self.embed_dim) + self.final_layer_norm = LayerNorm(self.embed_dim) + + def forward( + self, + x, + encoder_hidden_states, + encoder_attn_mask=None, + layer_state=None, + causal_mask=None, + decoder_padding_mask=None, + output_attentions=False, + ): + residual = x + + if layer_state is None: + layer_state = {} + + # Self Attention + x, self_attn_weights = self.self_attn( + query=x, + key=x, + layer_state=layer_state, # adds keys to layer state + key_padding_mask=decoder_padding_mask, + attn_mask=causal_mask, + output_attentions=output_attentions, + ) + x = F.dropout(x, p=self.dropout, training=self.training) + x = residual + x + x = self.self_attn_layer_norm(x) + + # Cross attention + residual = x + assert self.encoder_attn.cache_key != self.self_attn.cache_key + x, _ = self.encoder_attn( + query=x, + key=encoder_hidden_states, + key_padding_mask=encoder_attn_mask, + layer_state=layer_state, # mutates layer state + ) + x = F.dropout(x, p=self.dropout, training=self.training) + x = residual + x + x = self.encoder_attn_layer_norm(x) + + # Fully Connected + residual = x + x = self.activation_fn(self.fc1(x)) + x = F.dropout(x, p=self.activation_dropout, training=self.training) + x = self.fc2(x) + x = F.dropout(x, p=self.dropout, training=self.training) + x = residual + x + x = self.final_layer_norm(x) + return ( + x, + self_attn_weights, + layer_state, + ) # just self_attn weights for now, following t5, layer_state = cache for decoding + + +class FSMTDecoder(nn.Module): + """ + Transformer decoder consisting of *config.decoder_layers* layers. Each layer + is a :class:`DecoderLayer`. + Args: + config: FSMTConfig + embed_tokens (torch.nn.Embedding): output embedding + """ + + def __init__(self, config: FSMTConfig, embed_tokens: nn.Embedding): + super().__init__() + self.dropout = config.dropout + self.layerdrop = config.decoder_layerdrop + self.padding_idx = embed_tokens.padding_idx + self.max_target_positions = config.max_position_embeddings + self.embed_scale = math.sqrt(config.d_model) if config.scale_embedding else 1.0 + self.embed_tokens = embed_tokens + embed_dim = embed_tokens.embedding_dim + num_embeddings = config.tgt_vocab_size + self.embed_positions = SinusoidalPositionalEmbedding( + embed_dim, + self.padding_idx, + init_size=num_embeddings + self.padding_idx + 1, + ) + self.layers = nn.ModuleList( + [DecoderLayer(config) for _ in range(config.decoder_layers)] + ) # type: List[DecoderLayer] + + self.output_projection = nn.Linear( + self.embed_tokens.weight.shape[1], + self.embed_tokens.weight.shape[0], + bias=False, + ) + self.output_projection.weight = self.embed_tokens.weight + + def forward( + self, + input_ids, + encoder_hidden_states, + encoder_padding_mask, + decoder_padding_mask, + decoder_causal_mask, + past_key_values=None, + use_cache=False, + output_attentions=False, + output_hidden_states=False, + return_dict=False, + **unused, + ): + """ + Includes several features from "Jointly Learning to Align and + Translate with Transformer Models" (Garg et al., EMNLP 2019). + + Args: + input_ids (LongTensor): previous decoder outputs of shape + `(batch, tgt_len)`, for teacher forcing + encoder_hidden_states: output from the encoder, used for + encoder-side attention + encoder_padding_mask: for ignoring pad tokens + past_key_values (dict or None): dictionary used for storing state during generation + + Returns: + BaseModelOutputWithPast or tuple: + - the decoder's features of shape `(batch, tgt_len, embed_dim)` + - the cache + - hidden states + - attentions + """ + if "decoder_cached_states" in unused: + warnings.warn( + "The `decoder_cached_states` argument is deprecated and will be removed in a future version, use `past_key_values` instead.", + FutureWarning, + ) + past_key_values = unused.pop("decoder_cached_states") + if "decoder_past_key_values" in unused: + warnings.warn( + "The `decoder_past_key_values` argument is deprecated and will be removed in a future version, use `past_key_values` instead.", + FutureWarning, + ) + past_key_values = unused.pop("decoder_past_key_values") + + # check attention mask and invert + if encoder_padding_mask is not None: + encoder_padding_mask = invert_mask(encoder_padding_mask) + + # embed positions + positions = self.embed_positions(input_ids) # , use_cache=use_cache) + + if use_cache: + input_ids = input_ids[:, -1:] + positions = positions[:, -1:] # happens after we embed them + # assert input_ids.ne(self.padding_idx).any() + + x = self.embed_tokens(input_ids) * self.embed_scale + x += positions + x = F.dropout(x, p=self.dropout, training=self.training) + + # Convert to FSMT output format: (seq_len, BS, model_dim) -> (BS, seq_len, model_dim) + x = x.transpose(0, 1) + encoder_hidden_states = encoder_hidden_states.transpose(0, 1) + + # decoder layers + all_hidden_states = () if output_hidden_states else None + all_self_attns = () if output_attentions else None + next_decoder_cache = [] + for idx, decoder_layer in enumerate(self.layers): + # add LayerDrop (see https://arxiv.org/abs/1909.11556 for description) + if output_hidden_states: + all_hidden_states += (x,) + dropout_probability = random.uniform(0, 1) + if self.training and (dropout_probability < self.layerdrop): + continue + + layer_state = past_key_values[idx] if past_key_values is not None else None + + x, layer_self_attn, layer_past = decoder_layer( + x, + encoder_hidden_states, + encoder_attn_mask=encoder_padding_mask, + decoder_padding_mask=decoder_padding_mask, + layer_state=layer_state, + causal_mask=decoder_causal_mask, + output_attentions=output_attentions, + ) + + if use_cache: + next_decoder_cache.append(layer_past.copy()) + + if output_attentions: + all_self_attns += (layer_self_attn,) + + # Convert to standard output format: (seq_len, BS, model_dim) -> (BS, seq_len, model_dim) + if output_hidden_states: + all_hidden_states = tuple(hidden_state.transpose(0, 1) for hidden_state in all_hidden_states) + x = x.transpose(0, 1) + encoder_hidden_states = encoder_hidden_states.transpose(0, 1) + + x = self.output_projection(x) + + next_cache = next_decoder_cache if use_cache else None + + if not return_dict: + return tuple(v for v in [x, next_cache, all_hidden_states, all_self_attns] if v is not None) + return BaseModelOutputWithPast( + last_hidden_state=x, past_key_values=next_cache, hidden_states=all_hidden_states, attentions=all_self_attns + ) + + +def _reorder_buffer(attn_cache, new_order): + for k, input_buffer_k in attn_cache.items(): + if input_buffer_k is not None: + attn_cache[k] = input_buffer_k.index_select(0, new_order) + return attn_cache + + +class Attention(nn.Module): + """Multi-headed attention from 'Attention Is All You Need' paper""" + + def __init__( + self, + embed_dim, + num_heads, + dropout=0.0, + bias=True, + encoder_decoder_attention=False, # otherwise self_attention + ): + super().__init__() + self.embed_dim = embed_dim + self.num_heads = num_heads + self.dropout = dropout + self.head_dim = embed_dim // num_heads + assert self.head_dim * num_heads == self.embed_dim, "embed_dim must be divisible by num_heads" + self.scaling = self.head_dim ** -0.5 + + self.encoder_decoder_attention = encoder_decoder_attention + self.k_proj = nn.Linear(embed_dim, embed_dim, bias=bias) + self.v_proj = nn.Linear(embed_dim, embed_dim, bias=bias) + self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias) + self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias) + self.cache_key = "encoder_decoder" if self.encoder_decoder_attention else "self" + + def _shape(self, tensor, seq_len, bsz): + return tensor.contiguous().view(seq_len, bsz * self.num_heads, self.head_dim).transpose(0, 1) + + def forward( + self, + query, + key: Optional[Tensor], + key_padding_mask: Optional[Tensor] = None, + layer_state: Optional[Dict[str, Optional[Tensor]]] = None, + attn_mask: Optional[Tensor] = None, + output_attentions=False, + ) -> Tuple[Tensor, Optional[Tensor]]: + """Input shape: Time(SeqLen) x Batch x Channel""" + static_kv: bool = self.encoder_decoder_attention + tgt_len, bsz, embed_dim = query.size() + assert embed_dim == self.embed_dim + assert list(query.size()) == [tgt_len, bsz, embed_dim] + # get here for encoder decoder cause of static_kv + if layer_state is not None: # reuse k,v and encoder_padding_mask + saved_state = layer_state.get(self.cache_key, {}) + if "prev_key" in saved_state and static_kv: + # previous time steps are cached - no need to recompute key and value if they are static + key = None + else: + saved_state = None + layer_state = {} + + q = self.q_proj(query) * self.scaling + if static_kv: + if key is None: + k = v = None + else: + k = self.k_proj(key) + v = self.v_proj(key) + else: + k = self.k_proj(query) + v = self.v_proj(query) + + q = self._shape(q, tgt_len, bsz) + if k is not None: + k = self._shape(k, -1, bsz) + if v is not None: + v = self._shape(v, -1, bsz) + + if saved_state is not None: + k, v, key_padding_mask = self._use_saved_state(k, v, saved_state, key_padding_mask, static_kv, bsz) + + # Update cache + layer_state[self.cache_key] = { + "prev_key": k.view(bsz, self.num_heads, -1, self.head_dim), + "prev_value": v.view(bsz, self.num_heads, -1, self.head_dim), + "prev_key_padding_mask": key_padding_mask if not static_kv else None, + } + + assert k is not None + src_len = k.size(1) + attn_weights = torch.bmm(q, k.transpose(1, 2)) + assert attn_weights.size() == (bsz * self.num_heads, tgt_len, src_len) + + if attn_mask is not None: + attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + attn_mask + attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len) + + # This is part of a workaround to get around fork/join parallelism not supporting Optional types. + if key_padding_mask is not None and key_padding_mask.dim() == 0: + key_padding_mask = None + assert key_padding_mask is None or key_padding_mask.size()[:2] == ( + bsz, + src_len, + ) + + if key_padding_mask is not None: # don't attend to padding symbols + attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + reshaped = key_padding_mask.unsqueeze(1).unsqueeze(2) + attn_weights = attn_weights.masked_fill(reshaped, float("-inf")) + attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len) + attn_weights = F.softmax(attn_weights, dim=-1) + attn_probs = F.dropout( + attn_weights, + p=self.dropout, + training=self.training, + ) + + assert v is not None + attn_output = torch.bmm(attn_probs, v) + assert attn_output.size() == (bsz * self.num_heads, tgt_len, self.head_dim) + attn_output = attn_output.transpose(0, 1).contiguous().view(tgt_len, bsz, embed_dim) + attn_output = self.out_proj(attn_output) + if output_attentions: + attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + else: + attn_weights = None + return attn_output, attn_weights + + def _use_saved_state(self, k, v, saved_state, key_padding_mask, static_kv, bsz): + # saved states are stored with shape (bsz, num_heads, seq_len, head_dim) + if "prev_key" in saved_state: + _prev_key = saved_state["prev_key"] + assert _prev_key is not None + prev_key = _prev_key.view(bsz * self.num_heads, -1, self.head_dim) + if static_kv: + k = prev_key + else: + assert k is not None + k = torch.cat([prev_key, k], dim=1) + if "prev_value" in saved_state: + _prev_value = saved_state["prev_value"] + assert _prev_value is not None + prev_value = _prev_value.view(bsz * self.num_heads, -1, self.head_dim) + if static_kv: + v = prev_value + else: + assert v is not None + v = torch.cat([prev_value, v], dim=1) + assert k is not None and v is not None + prev_key_padding_mask: Optional[Tensor] = saved_state.get("prev_key_padding_mask", None) + if prev_key_padding_mask is not None: + if static_kv: + new_key_padding_mask = prev_key_padding_mask + else: + new_key_padding_mask = torch.cat([prev_key_padding_mask, key_padding_mask], dim=1) + else: + new_key_padding_mask = key_padding_mask + return k, v, new_key_padding_mask + + +def fill_with_neg_inf(t): + """FP16-compatible function that fills a input_ids with -inf.""" + return t.float().fill_(float("-inf")).type_as(t) + + +# Public API +def _get_shape(t): + return getattr(t, "shape", None) + + +@add_start_docstrings( + "The bare FSMT Model outputting raw hidden-states without any specific head on top.", + FSMT_START_DOCSTRING, +) +class FSMTModel(PretrainedFSMTModel): + def __init__(self, config: FSMTConfig): + super().__init__(config) + + padding_idx = config.pad_token_id + encoder_embed_tokens = nn.Embedding(config.src_vocab_size, config.d_model, padding_idx) + decoder_embed_tokens = nn.Embedding(config.tgt_vocab_size, config.d_model, padding_idx) + + self.encoder = FSMTEncoder(config, encoder_embed_tokens) + self.decoder = FSMTDecoder(config, decoder_embed_tokens) + + self.init_weights() + + @add_start_docstrings_to_callable(FSMT_INPUTS_DOCSTRING) + @add_code_sample_docstrings( + tokenizer_class=_TOKENIZER_FOR_DOC, + checkpoint="facebook/wmt19-ru-en", + output_type=BaseModelOutputWithPast, + config_class=_CONFIG_FOR_DOC, + ) + def forward( + self, + input_ids, + attention_mask=None, + decoder_input_ids=None, + encoder_outputs: Optional[Tuple] = None, + decoder_attention_mask=None, + past_key_values=None, + use_cache=None, + output_attentions=None, + output_hidden_states=None, + return_dict=None, + **kwargs, + ): + if "decoder_past_key_values" in kwargs: + warnings.warn( + "The `decoder_past_key_values` argument is deprecated and will be removed in a future version, use `past_key_values` instead.", + FutureWarning, + ) + past_key_values = kwargs.pop("decoder_past_key_values") + + if decoder_input_ids is None: + use_cache = False + + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) + use_cache = use_cache if use_cache is not None else self.config.use_cache + return_dict = return_dict if return_dict is not None else self.config.use_return_dict + + # make masks if user doesn't supply + if not use_cache: + decoder_input_ids, decoder_padding_mask, causal_mask = _prepare_fsmt_decoder_inputs( + self.config, + input_ids, + decoder_input_ids=decoder_input_ids, + decoder_padding_mask=decoder_attention_mask, + causal_mask_dtype=self.decoder.embed_tokens.weight.dtype, + ) + else: + decoder_padding_mask, causal_mask = None, None + + assert decoder_input_ids is not None + + if encoder_outputs is None: + encoder_outputs = self.encoder( + input_ids=input_ids, + attention_mask=attention_mask, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) + # If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOuput when return_dict=False + elif return_dict and not isinstance(encoder_outputs, BaseModelOutput): + encoder_outputs = BaseModelOutput( + last_hidden_state=encoder_outputs[0], + hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None, + attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None, + ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) + decoder_outputs = self.decoder( + decoder_input_ids, + encoder_outputs[0], + attention_mask, + decoder_padding_mask, + decoder_causal_mask=causal_mask, + past_key_values=past_key_values, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) + + if not return_dict: + return decoder_outputs + encoder_outputs + + return Seq2SeqModelOutput( + last_hidden_state=decoder_outputs.last_hidden_state, + past_key_values=decoder_outputs.past_key_values, + decoder_hidden_states=decoder_outputs.hidden_states, + decoder_attentions=decoder_outputs.attentions, + encoder_last_hidden_state=encoder_outputs.last_hidden_state, + encoder_hidden_states=encoder_outputs.hidden_states, + encoder_attentions=encoder_outputs.attentions, + ) + + def get_input_embeddings(self): + return self.encoder.embed_tokens + + def set_input_embeddings(self, value): + self.encoder.embed_tokens = value + + def get_output_embeddings(self): + return self.decoder.embed_tokens + + def set_output_embeddings(self, value): + self.decoder.embed_tokens = value + + +@add_start_docstrings( + "The FSMT Model with a language modeling head. Can be used for summarization.", FSMT_START_DOCSTRING +) +class FSMTForConditionalGeneration(PretrainedFSMTModel): + base_model_prefix = "model" + + def __init__(self, config: FSMTConfig): + super().__init__(config) + base_model = FSMTModel(config) + self.model = base_model + + def resize_token_embeddings(self, new_num_tokens: int) -> nn.Embedding: + new_embeddings = super().resize_token_embeddings(new_num_tokens) + self.model.encoder.embed_tokens = new_embeddings + + new_embeddings = super().resize_token_embeddings(new_num_tokens) + self.model.decoder.embed_tokens = new_embeddings + + # XXX: this is not quite correct, as we have 2 different `new_embeddings`, and + # only one return value is expected. Needs to be redesigned in the core to support dual dicts + raise NotImplementedError("this method needs re-thinking for models with 2 separate dictionaries") + + return new_embeddings + + @add_start_docstrings_to_callable(FSMT_INPUTS_DOCSTRING) + @replace_return_docstrings(output_type=Seq2SeqLMOutput, config_class=_CONFIG_FOR_DOC) + @add_end_docstrings(FSMT_GENERATION_EXAMPLE) + def forward( + self, + input_ids, + attention_mask=None, + encoder_outputs=None, + decoder_input_ids=None, + decoder_attention_mask=None, + past_key_values=None, + labels=None, + use_cache=None, + output_attentions=None, + output_hidden_states=None, + return_dict=None, + **unused, + ): + r""" + labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`): + Labels for computing the masked language modeling loss. + Indices should either be in ``[0, ..., config.vocab_size]`` or -100 (see ``input_ids`` docstring). + Tokens with indices set to ``-100`` are ignored (masked), the loss is only computed for the tokens + with labels in ``[0, ..., config.vocab_size]``. + + Returns: + + """ + return_dict = return_dict if return_dict is not None else self.config.use_return_dict + + if labels is not None: + use_cache = False + + outputs = self.model( + input_ids, + attention_mask=attention_mask, + decoder_input_ids=decoder_input_ids, + encoder_outputs=encoder_outputs, + decoder_attention_mask=decoder_attention_mask, + past_key_values=past_key_values, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) + lm_logits = outputs[0] + + masked_lm_loss = None + if labels is not None: + loss_fct = CrossEntropyLoss() + # TODO(SS): do we need to ignore pad tokens in labels? + masked_lm_loss = loss_fct(lm_logits.view(-1, self.config.tgt_vocab_size), labels.view(-1)) + + if not return_dict: + output = (lm_logits,) + outputs[1:] + return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output + + return Seq2SeqLMOutput( + loss=masked_lm_loss, + logits=lm_logits, + past_key_values=outputs.past_key_values, + decoder_hidden_states=outputs.decoder_hidden_states, + decoder_attentions=outputs.decoder_attentions, + encoder_last_hidden_state=outputs.encoder_last_hidden_state, + encoder_hidden_states=outputs.encoder_hidden_states, + encoder_attentions=outputs.encoder_attentions, + ) + + def prepare_inputs_for_generation( + self, decoder_input_ids, past, attention_mask, use_cache, encoder_outputs, **kwargs + ): + return { + "input_ids": None, # encoder_outputs is defined. input_ids not needed + "encoder_outputs": encoder_outputs, + "past_key_values": past, + "decoder_input_ids": decoder_input_ids, + "attention_mask": attention_mask, + "use_cache": use_cache, # change this to avoid caching (presumably for debugging) + } + + def adjust_logits_during_generation(self, logits, cur_len, max_length): + if cur_len == max_length - 1 and self.config.eos_token_id is not None: + self._force_token_ids_generation(logits, self.config.eos_token_id) + return logits + + def _force_token_ids_generation(self, scores, token_ids) -> None: + """force one of token_ids to be generated by setting prob of all other tokens to 0""" + if isinstance(token_ids, int): + token_ids = [token_ids] + all_but_token_ids_mask = torch.tensor( + [x for x in range(self.config.tgt_vocab_size) if x not in token_ids], + dtype=torch.long, + device=next(self.parameters()).device, + ) + assert len(scores.shape) == 2, "scores should be of rank 2 with shape: [batch_size, vocab_size]" + scores[:, all_but_token_ids_mask] = -float("inf") + + @staticmethod + def _reorder_cache(past, beam_idx): + reordered_past = [] + for layer_past in past: + # get the correct batch idx from decoder layer's batch dim for cross and self-attn + layer_past_new = { + attn_key: _reorder_buffer(attn_cache, beam_idx) for attn_key, attn_cache in layer_past.items() + } + reordered_past.append(layer_past_new) + return reordered_past + + def get_encoder(self): + return self.model.encoder + + def get_output_embeddings(self): + return self.model.decoder.embed_tokens + + +def make_positions(tensor, padding_idx: int): + """Replace non-padding symbols with their position numbers. + + Position numbers begin at padding_idx+1. Padding symbols are ignored. + """ + # The series of casts and type-conversions here are carefully + # balanced to both work with ONNX export and XLA. In particular XLA + # prefers ints, cumsum defaults to output longs, and ONNX doesn't know + # how to handle the dtype kwarg in cumsum. + mask = tensor.ne(padding_idx).int() + return (torch.cumsum(mask, dim=1).type_as(mask) * mask).long() + padding_idx + + +class SinusoidalPositionalEmbedding(nn.Module): + """This module produces sinusoidal positional embeddings of any length. + + Padding symbols are ignored. + """ + + def __init__(self, embedding_dim, padding_idx, init_size=1024): + super().__init__() + self.embedding_dim = embedding_dim + self.padding_idx = padding_idx + self.weights = SinusoidalPositionalEmbedding.get_embedding(init_size, embedding_dim, padding_idx) + self.register_buffer("_float_tensor", torch.zeros(1)) # used for getting the right device + self.max_positions = int(1e5) + + # XXX: bart uses s/num_embeddings/num_positions/, s/weights/weight/ - could make those match + @staticmethod + def get_embedding(num_embeddings: int, embedding_dim: int, padding_idx: Optional[int] = None): + """Build sinusoidal embeddings. + + This matches the implementation in tensor2tensor, but differs slightly + from the description in Section 3.5 of "Attention Is All You Need". + """ + half_dim = embedding_dim // 2 + emb = math.log(10000) / (half_dim - 1) + emb = torch.exp(torch.arange(half_dim, dtype=torch.float) * -emb) + emb = torch.arange(num_embeddings, dtype=torch.float).unsqueeze(1) * emb.unsqueeze(0) + emb = torch.cat([torch.sin(emb), torch.cos(emb)], dim=1).view(num_embeddings, -1) + if embedding_dim % 2 == 1: + # zero pad + emb = torch.cat([emb, torch.zeros(num_embeddings, 1)], dim=1) + if padding_idx is not None: + emb[padding_idx, :] = 0 + return emb + + def forward( + self, + input, + incremental_state: Optional[Any] = None, + timestep: Optional[Tensor] = None, + positions: Optional[Any] = None, + ): + """Input is expected to be of size [bsz x seqlen].""" + # bspair = torch.onnx.operators.shape_as_tensor(input) + # bsz, seq_len = bspair[0], bspair[1] + bsz, seq_len = input.shape[:2] + max_pos = self.padding_idx + 1 + seq_len + if self.weights is None or max_pos > self.weights.size(0): + # recompute/expand embeddings if needed + self.weights = SinusoidalPositionalEmbedding.get_embedding(max_pos, self.embedding_dim, self.padding_idx) + self.weights = self.weights.to(self._float_tensor) + + if incremental_state is not None: + # positions is the same for every token when decoding a single step + pos = timestep.view(-1)[0] + 1 if timestep is not None else seq_len + return self.weights[self.padding_idx + pos, :].expand(bsz, 1, -1) + + positions = make_positions(input, self.padding_idx) + + return self.weights.index_select(0, positions.view(-1)).view(bsz, seq_len, -1).detach() diff --git a/src/transformers/tokenization_auto.py b/src/transformers/tokenization_auto.py index d450c76d04f940..2e7f38193f8494 100644 --- a/src/transformers/tokenization_auto.py +++ b/src/transformers/tokenization_auto.py @@ -29,6 +29,7 @@ ElectraConfig, EncoderDecoderConfig, FlaubertConfig, + FSMTConfig, FunnelConfig, GPT2Config, LongformerConfig, @@ -59,6 +60,7 @@ from .tokenization_distilbert import DistilBertTokenizer, DistilBertTokenizerFast from .tokenization_electra import ElectraTokenizer, ElectraTokenizerFast from .tokenization_flaubert import FlaubertTokenizer +from .tokenization_fsmt import FSMTTokenizer from .tokenization_funnel import FunnelTokenizer, FunnelTokenizerFast from .tokenization_gpt2 import GPT2Tokenizer, GPT2TokenizerFast from .tokenization_longformer import LongformerTokenizer, LongformerTokenizerFast @@ -109,6 +111,7 @@ (FlaubertConfig, (FlaubertTokenizer, None)), (XLMConfig, (XLMTokenizer, None)), (CTRLConfig, (CTRLTokenizer, None)), + (FSMTConfig, (FSMTTokenizer, None)), (BertGenerationConfig, (BertGenerationTokenizer, None)), ] ) diff --git a/src/transformers/tokenization_fsmt.py b/src/transformers/tokenization_fsmt.py new file mode 100644 index 00000000000000..63f720c70a51f8 --- /dev/null +++ b/src/transformers/tokenization_fsmt.py @@ -0,0 +1,535 @@ +# coding=utf-8 +# Copyright 2019 The Open AI 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. +"""Tokenization classes for FSMT.""" + + +import json +import logging +import os +import re +import unicodedata +from typing import Dict, List, Optional + +import sacremoses as sm + +from .file_utils import add_start_docstrings +from .tokenization_utils import BatchEncoding, PreTrainedTokenizer +from .tokenization_utils_base import PREPARE_SEQ2SEQ_BATCH_DOCSTRING + + +logger = logging.getLogger(__name__) + +VOCAB_FILES_NAMES = { + "src_vocab_file": "vocab-src.json", + "tgt_vocab_file": "vocab-tgt.json", + "merges_file": "merges.txt", +} + +PRETRAINED_VOCAB_FILES_MAP = {} +PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {} +PRETRAINED_INIT_CONFIGURATION = {} + + +def get_pairs(word): + """ + Return set of symbol pairs in a word. + word is represented as tuple of symbols (symbols being variable-length strings) + """ + pairs = set() + prev_char = word[0] + for char in word[1:]: + pairs.add((prev_char, char)) + prev_char = char + return pairs + + +def replace_unicode_punct(text): + """ + Port of https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/replace-unicode-punctuation.perl + """ + text = text.replace("，", ",") + text = re.sub(r"。\s*", ". ", text) + text = text.replace("、", ",") + text = text.replace("”", '"') + text = text.replace("“", '"') + text = text.replace("∶", ":") + text = text.replace("：", ":") + text = text.replace("？", "?") + text = text.replace("《", '"') + text = text.replace("》", '"') + text = text.replace("）", ")") + text = text.replace("！", "!") + text = text.replace("（", "(") + text = text.replace("；", ";") + text = text.replace("１", "1") + text = text.replace("」", '"') + text = text.replace("「", '"') + text = text.replace("０", "0") + text = text.replace("３", "3") + text = text.replace("２", "2") + text = text.replace("５", "5") + text = text.replace("６", "6") + text = text.replace("９", "9") + text = text.replace("７", "7") + text = text.replace("８", "8") + text = text.replace("４", "4") + text = re.sub(r"．\s*", ". ", text) + text = text.replace("～", "~") + text = text.replace("’", "'") + text = text.replace("…", "...") + text = text.replace("━", "-") + text = text.replace("〈", "<") + text = text.replace("〉", ">") + text = text.replace("【", "[") + text = text.replace("】", "]") + text = text.replace("％", "%") + return text + + +def remove_non_printing_char(text): + """ + Port of https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/remove-non-printing-char.perl + """ + output = [] + for char in text: + cat = unicodedata.category(char) + if cat.startswith("C"): + continue + output.append(char) + return "".join(output) + + +# Porting notes: +# this one is modeled after XLMTokenizer +# +# added: +# - src_vocab_file, +# - tgt_vocab_file, +# - langs, + + +class FSMTTokenizer(PreTrainedTokenizer): + """ + BPE tokenizer for FSMT (fairseq transformer) + See: https://github.com/pytorch/fairseq/tree/master/examples/wmt19 + + - Moses preprocessing & tokenization for most supported languages + - (optionally) lower case & normalize all inputs text + - argument ``special_tokens`` and function ``set_special_tokens``, can be used to add additional symbols \ + (ex: "__classify__") to a vocabulary + - `langs` defines a pair of languages + + This tokenizer inherits from :class:`~transformers.PreTrainedTokenizer` which contains most of the methods. Users + should refer to the superclass for more information regarding methods. + + Args: + langs (:obj:`List[str]`): + a list of two languages to translate from and to, e.g. ``["en", "ru"]``. + src_vocab_file (:obj:`string`): + Source language vocabulary file. + tgt_vocab_file (:obj:`string`): + Target language vocabulary file. + merges_file (:obj:`string`): + Merges file. + do_lower_case (:obj:`bool`, `optional`, defaults to :obj:`True`): + Whether to lowercase the input when tokenizing. + unk_token (:obj:`string`, `optional`, defaults to ""): + 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. + bos_token (:obj:`string`, `optional`, defaults to ""): + The beginning of sequence token that was used during pre-training. Can be used a sequence classifier token. + + .. note:: + + 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 :obj:`cls_token`. + sep_token (:obj:`string`, `optional`, defaults to ""): + 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. + pad_token (:obj:`string`, `optional`, defaults to ""): + The token used for padding, for example when batching sequences of different lengths. + + """ + + vocab_files_names = VOCAB_FILES_NAMES + pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP + pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION + max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES + + def __init__( + self, + langs=None, + src_vocab_file=None, + tgt_vocab_file=None, + merges_file=None, + unk_token="", + bos_token="", + sep_token="", + pad_token="", + **kwargs + ): + super().__init__( + unk_token=unk_token, + bos_token=bos_token, + sep_token=sep_token, + pad_token=pad_token, + **kwargs, + ) + + self.src_vocab_file = src_vocab_file + self.tgt_vocab_file = tgt_vocab_file + self.merges_file = merges_file + + # cache of sm.MosesPunctNormalizer instance + self.cache_moses_punct_normalizer = dict() + # cache of sm.MosesTokenizer instance + self.cache_moses_tokenizer = dict() + self.cache_moses_detokenizer = dict() + + if langs and len(langs) == 2: + self.src_lang, self.tgt_lang = langs + else: + raise ValueError( + f"arg `langs` needs to be a list of 2 langs, e.g. ['en', 'ru'], but got {langs}. " + "Usually that means that tokenizer can't find a mapping for the given model path " + "in PRETRAINED_VOCAB_FILES_MAP, and other maps of this tokenizer." + ) + + with open(src_vocab_file, encoding="utf-8") as src_vocab_handle: + self.encoder = json.load(src_vocab_handle) + with open(tgt_vocab_file, encoding="utf-8") as tgt_vocab_handle: + tgt_vocab = json.load(tgt_vocab_handle) + self.decoder = {v: k for k, v in tgt_vocab.items()} + with open(merges_file, encoding="utf-8") as merges_handle: + merges = merges_handle.read().split("\n")[:-1] + merges = [tuple(merge.split()[:2]) for merge in merges] + self.bpe_ranks = dict(zip(merges, range(len(merges)))) + self.cache = {} + + # hack override + def get_vocab(self) -> Dict[str, int]: + return self.get_src_vocab() + + # hack override + @property + def vocab_size(self) -> int: + return self.src_vocab_size + + def moses_punct_norm(self, text, lang): + if lang not in self.cache_moses_punct_normalizer: + punct_normalizer = sm.MosesPunctNormalizer(lang=lang) + self.cache_moses_punct_normalizer[lang] = punct_normalizer + return self.cache_moses_punct_normalizer[lang].normalize(text) + + def moses_tokenize(self, text, lang): + if lang not in self.cache_moses_tokenizer: + moses_tokenizer = sm.MosesTokenizer(lang=lang) + self.cache_moses_tokenizer[lang] = moses_tokenizer + return self.cache_moses_tokenizer[lang].tokenize( + text, aggressive_dash_splits=True, return_str=False, escape=True + ) + + def moses_detokenize(self, tokens, lang): + if lang not in self.cache_moses_tokenizer: + moses_detokenizer = sm.MosesDetokenizer(lang=self.tgt_lang) + self.cache_moses_detokenizer[lang] = moses_detokenizer + return self.cache_moses_detokenizer[lang].detokenize(tokens) + + def moses_pipeline(self, text, lang): + text = replace_unicode_punct(text) + text = self.moses_punct_norm(text, lang) + text = remove_non_printing_char(text) + return text + + @property + def src_vocab_size(self): + return len(self.encoder) + + @property + def tgt_vocab_size(self): + return len(self.decoder) + + def get_src_vocab(self): + return dict(self.encoder, **self.added_tokens_encoder) + + def get_tgt_vocab(self): + return dict(self.decoder, **self.added_tokens_decoder) + + def bpe(self, token): + word = tuple(token[:-1]) + (token[-1] + "",) + if token in self.cache: + return self.cache[token] + pairs = get_pairs(word) + + if not pairs: + return token + "" + + while True: + bigram = min(pairs, key=lambda pair: self.bpe_ranks.get(pair, float("inf"))) + if bigram not in self.bpe_ranks: + break + first, second = bigram + new_word = [] + i = 0 + while i < len(word): + try: + j = word.index(first, i) + except ValueError: + new_word.extend(word[i:]) + break + else: + new_word.extend(word[i:j]) + i = j + + if word[i] == first and i < len(word) - 1 and word[i + 1] == second: + new_word.append(first + second) + i += 2 + else: + new_word.append(word[i]) + i += 1 + new_word = tuple(new_word) + word = new_word + if len(word) == 1: + break + else: + pairs = get_pairs(word) + word = " ".join(word) + if word == "\n ": + word = "\n" + self.cache[token] = word + return word + + def _tokenize(self, text, lang="en", bypass_tokenizer=False): + """ + Tokenize a string given language code using Moses. + + Details of tokenization: + - [sacremoses](https://github.com/alvations/sacremoses): port of Moses + - Install with `pip install sacremoses` + + Args: + - lang: ISO language code (default = 'en') (string). Languages should belong of the model supported languages. However, we don't enforce it. + - bypass_tokenizer: Allow users to preprocess and tokenize the sentences externally (default = False) (bool). If True, we only apply BPE. + + Returns: + List of tokens. + """ + # ignore `lang` which is currently isn't explicitly passed in tokenization_utils.py and always results in lang=en + # if lang != self.src_lang: + # raise ValueError(f"Expected lang={self.src_lang}, but got {lang}") + lang = self.src_lang + + if bypass_tokenizer: + text = text.split() + else: + text = self.moses_pipeline(text, lang=lang) + text = self.moses_tokenize(text, lang=lang) + + split_tokens = [] + for token in text: + if token: + split_tokens.extend([t for t in self.bpe(token).split(" ")]) + + return split_tokens + + def _convert_token_to_id(self, token): + """ Converts a token (str) in an id using the vocab. """ + return self.encoder.get(token, self.encoder.get(self.unk_token)) + + def _convert_id_to_token(self, index): + """Converts an index (integer) in a token (str) using the vocab.""" + return self.decoder.get(index, self.unk_token) + + def convert_tokens_to_string(self, tokens): + """ Converts a sequence of tokens (string) in a single string. """ + + # remove BPE + tokens = [t.replace(" ", "").replace("", " ") for t in tokens] + tokens = "".join(tokens).split() + # detokenize + text = self.moses_detokenize(tokens, self.tgt_lang) + return text + + 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. + A FAIRSEQ_TRANSFORMER sequence has the following format: + + - single sequence: `` X `` + - pair of sequences: `` A B `` + + Args: + token_ids_0 (:obj:`List[int]`): + List of IDs to which the special tokens will be added + token_ids_1 (:obj:`List[int]`, `optional`): + Optional second list of IDs for sequence pairs. + + Returns: + :obj:`List[int]`: list of `input IDs <../glossary.html#input-ids>`__ with the appropriate special tokens. + + """ + sep = [self.sep_token_id] + + # no bos used in fairseq + if token_ids_1 is None: + return token_ids_0 + sep + return token_ids_0 + sep + token_ids_1 + sep + + 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]: + """ + Retrieves 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`` methods. + + Args: + token_ids_0 (:obj:`List[int]`): + List of ids. + token_ids_1 (:obj:`List[int]`, `optional`): + Optional second list of IDs for sequence pairs. + already_has_special_tokens (:obj:`bool`, `optional`, defaults to :obj:`False`): + Set to True if the token list is already formatted with special tokens for the model + + Returns: + :obj:`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: + if token_ids_1 is not None: + raise ValueError( + "You should not supply a second sequence if the provided sequence of " + "ids is already formated with special tokens for the model." + ) + return list( + map( + lambda x: 1 if x in [self.sep_token_id, self.cls_token_id] else 0, + token_ids_0, + ) + ) + # no bos used in fairseq + if token_ids_1 is not None: + return ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1)) + [1] + return ([0] * len(token_ids_0)) + [1] + + def create_token_type_ids_from_sequences( + self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None + ) -> List[int]: + """ + Creates a mask from the two sequences passed to be used in a sequence-pair classification task. + An FAIRSEQ_TRANSFORMER sequence pair mask has the following format: + + :: + + 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 + | first sequence | second sequence | + + if token_ids_1 is None, only returns the first portion of the mask (0s). + + Args: + token_ids_0 (:obj:`List[int]`): + List of ids. + token_ids_1 (:obj:`List[int]`, `optional`): + Optional second list of IDs for sequence pairs. + + Returns: + :obj:`List[int]`: List of `token type IDs <../glossary.html#token-type-ids>`_ according to the given + sequence(s). + """ + sep = [self.sep_token_id] + + # no bos used in fairseq + if token_ids_1 is None: + return len(token_ids_0 + sep) * [0] + return len(token_ids_0 + sep) * [0] + len(token_ids_1 + sep) * [1] + + @add_start_docstrings(PREPARE_SEQ2SEQ_BATCH_DOCSTRING) + def prepare_seq2seq_batch( + self, + src_texts: List[str], + tgt_texts: Optional[List[str]] = None, + max_length: Optional[int] = None, + max_target_length: Optional[int] = None, + return_tensors: str = "pt", + truncation=True, + padding="longest", + **unused, + ) -> BatchEncoding: + """Prepare model inputs for translation. For best performance, translate one sentence at a time.""" + + if type(src_texts) is not list: + raise ValueError("src_texts is expected to be a list") + if "" in src_texts: + raise ValueError(f"found empty string in src_texts: {src_texts}") + + tokenizer_kwargs = dict( + add_special_tokens=True, + return_tensors=return_tensors, + max_length=max_length, + truncation=truncation, + padding=padding, + ) + model_inputs: BatchEncoding = self(src_texts, **tokenizer_kwargs) + + if tgt_texts is None: + return model_inputs + if max_target_length is not None: + tokenizer_kwargs["max_length"] = max_target_length + + model_inputs["labels"] = self(tgt_texts, **tokenizer_kwargs)["input_ids"] + return model_inputs + + def save_vocabulary(self, save_directory): + """ + Save the vocabulary and special tokens file to a directory. + + Args: + save_directory (:obj:`str`): + The directory in which to save the vocabulary. + + Returns: + :obj:`Tuple(str)`: Paths to the files saved. + """ + if not os.path.isdir(save_directory): + logger.error("Vocabulary path ({}) should be a directory".format(save_directory)) + return + + src_vocab_file = os.path.join(save_directory, VOCAB_FILES_NAMES["src_vocab_file"]) + tgt_vocab_file = os.path.join(save_directory, VOCAB_FILES_NAMES["tgt_vocab_file"]) + merges_file = os.path.join(save_directory, VOCAB_FILES_NAMES["merges_file"]) + + with open(src_vocab_file, "w", encoding="utf-8") as f: + f.write(json.dumps(self.encoder, ensure_ascii=False)) + + with open(tgt_vocab_file, "w", encoding="utf-8") as f: + tgt_vocab = {v: k for k, v in self.decoder.items()} + f.write(json.dumps(tgt_vocab, ensure_ascii=False)) + + index = 0 + with open(merges_file, "w", encoding="utf-8") as writer: + for bpe_tokens, token_index in sorted(self.bpe_ranks.items(), key=lambda kv: kv[1]): + if index != token_index: + logger.warning( + "Saving vocabulary to {}: BPE merge indices are not consecutive." + " Please check that the tokenizer is not corrupted!".format(merges_file) + ) + index = token_index + writer.write(" ".join(bpe_tokens) + "\n") + index += 1 + + return src_vocab_file, tgt_vocab_file, merges_file diff --git a/tests/test_modeling_fsmt.py b/tests/test_modeling_fsmt.py new file mode 100644 index 00000000000000..4c56d76a7aace8 --- /dev/null +++ b/tests/test_modeling_fsmt.py @@ -0,0 +1,501 @@ +# coding=utf-8 +# Copyright 2020 Huggingface +# +# 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 tempfile +import unittest + +import timeout_decorator # noqa + +from parameterized import parameterized +from transformers import is_torch_available +from transformers.file_utils import cached_property +from transformers.testing_utils import require_torch, slow, torch_device + +from .test_configuration_common import ConfigTester +from .test_modeling_common import ModelTesterMixin, ids_tensor + + +if is_torch_available(): + import torch + + from transformers import FSMTConfig, FSMTForConditionalGeneration, FSMTModel, FSMTTokenizer + from transformers.modeling_fsmt import ( + SinusoidalPositionalEmbedding, + _prepare_fsmt_decoder_inputs, + invert_mask, + shift_tokens_right, + ) + + +@require_torch +class ModelTester: + def __init__( + self, + parent, + ): + self.parent = parent + self.src_vocab_size = 99 + self.tgt_vocab_size = 99 + self.langs = ["ru", "en"] + self.batch_size = 13 + self.seq_length = 7 + self.is_training = False + self.use_labels = False + self.hidden_size = 16 + self.num_hidden_layers = 2 + self.num_attention_heads = 4 + self.intermediate_size = 4 + self.hidden_act = "relu" + self.hidden_dropout_prob = 0.1 + self.attention_probs_dropout_prob = 0.1 + self.max_position_embeddings = 20 + self.bos_token_id = 0 + self.pad_token_id = 1 + self.eos_token_id = 2 + torch.manual_seed(0) + + # hack needed for modeling_common tests - despite not really having this attribute in this model + self.vocab_size = self.src_vocab_size + + def prepare_config_and_inputs_for_common(self): + input_ids = ids_tensor([self.batch_size, self.seq_length], self.src_vocab_size).clamp( + 3, + ) + input_ids[:, -1] = 2 # Eos Token + + config = FSMTConfig( + vocab_size=self.src_vocab_size, # hack needed for common tests + src_vocab_size=self.src_vocab_size, + tgt_vocab_size=self.tgt_vocab_size, + langs=self.langs, + d_model=self.hidden_size, + encoder_layers=self.num_hidden_layers, + decoder_layers=self.num_hidden_layers, + encoder_attention_heads=self.num_attention_heads, + decoder_attention_heads=self.num_attention_heads, + encoder_ffn_dim=self.intermediate_size, + decoder_ffn_dim=self.intermediate_size, + dropout=self.hidden_dropout_prob, + attention_dropout=self.attention_probs_dropout_prob, + max_position_embeddings=self.max_position_embeddings, + eos_token_id=self.eos_token_id, + bos_token_id=self.bos_token_id, + pad_token_id=self.pad_token_id, + ) + inputs_dict = prepare_fsmt_inputs_dict(config, input_ids) + return config, inputs_dict + + +def prepare_fsmt_inputs_dict( + config, + input_ids, + attention_mask=None, +): + if attention_mask is None: + attention_mask = input_ids.ne(config.pad_token_id) + return { + "input_ids": input_ids, + "attention_mask": attention_mask, + } + + +@require_torch +class FSMTModelTest(ModelTesterMixin, unittest.TestCase): + all_model_classes = (FSMTModel, FSMTForConditionalGeneration) if is_torch_available() else () + all_generative_model_classes = (FSMTForConditionalGeneration,) if is_torch_available() else () + is_encoder_decoder = True + # TODO(SS): fix the below in a separate PR + test_pruning = False + test_torchscript = True + test_head_masking = False + test_resize_embeddings = True # This requires inputs_dict['input_ids'] + test_missing_keys = False # because FSMTForConditionalGeneration and FSMTModel now have identical state_dict + + def setUp(self): + self.model_tester = ModelTester(self) + self.langs = ["en", "ru"] + config = { + "langs": self.langs, + "src_vocab_size": 10, + "tgt_vocab_size": 20, + } + # XXX: hack to appease to all other models requiring `vocab_size` + config["vocab_size"] = 99 # no such thing in FSMT + self.config_tester = ConfigTester(self, config_class=FSMTConfig, **config) + + def test_config(self): + self.config_tester.run_common_tests() + + # XXX: override test_model_common_attributes / different Embedding type + def test_model_common_attributes(self): + config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common() + + for model_class in self.all_model_classes: + model = model_class(config) + self.assertIsInstance(model.get_input_embeddings(), (torch.nn.Embedding)) + model.set_input_embeddings(torch.nn.Embedding(10, 10)) + x = model.get_output_embeddings() + self.assertTrue(x is None or isinstance(x, torch.nn.modules.sparse.Embedding)) + + def test_initialization_more(self): + config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common() + model = FSMTModel(config) + model.to(torch_device) + model.eval() + # test init + # self.assertTrue((model.encoder.embed_tokens.weight == model.shared.weight).all().item()) + + def _check_var(module): + """Check that we initialized various parameters from N(0, config.init_std).""" + self.assertAlmostEqual(torch.std(module.weight).item(), config.init_std, 2) + + _check_var(model.encoder.embed_tokens) + _check_var(model.encoder.layers[0].self_attn.k_proj) + _check_var(model.encoder.layers[0].fc1) + # XXX: different std for fairseq version of SinusoidalPositionalEmbedding + # self.assertAlmostEqual(torch.std(model.encoder.embed_positions.weights).item(), config.init_std, 2) + + def test_advanced_inputs(self): + config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common() + config.use_cache = False + inputs_dict["input_ids"][:, -2:] = config.pad_token_id + decoder_input_ids, decoder_attn_mask, causal_mask = _prepare_fsmt_decoder_inputs( + config, inputs_dict["input_ids"] + ) + model = FSMTModel(config).to(torch_device).eval() + + decoder_features_with_created_mask = model(**inputs_dict)[0] + decoder_features_with_passed_mask = model( + decoder_attention_mask=invert_mask(decoder_attn_mask), decoder_input_ids=decoder_input_ids, **inputs_dict + )[0] + _assert_tensors_equal(decoder_features_with_passed_mask, decoder_features_with_created_mask) + useless_mask = torch.zeros_like(decoder_attn_mask) + decoder_features = model(decoder_attention_mask=useless_mask, **inputs_dict)[0] + self.assertTrue(isinstance(decoder_features, torch.Tensor)) # no hidden states or attentions + self.assertEqual( + decoder_features.size(), + (self.model_tester.batch_size, self.model_tester.seq_length, config.tgt_vocab_size), + ) + if decoder_attn_mask.min().item() < -1e3: # some tokens were masked + self.assertFalse((decoder_features_with_created_mask == decoder_features).all().item()) + + # Test different encoder attention masks + decoder_features_with_long_encoder_mask = model( + inputs_dict["input_ids"], attention_mask=inputs_dict["attention_mask"].long() + )[0] + _assert_tensors_equal(decoder_features_with_long_encoder_mask, decoder_features_with_created_mask) + + def test_save_load_strict(self): + config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common() + for model_class in self.all_model_classes: + model = model_class(config) + + with tempfile.TemporaryDirectory() as tmpdirname: + model.save_pretrained(tmpdirname) + model2, info = model_class.from_pretrained(tmpdirname, output_loading_info=True) + self.assertEqual(info["missing_keys"], []) + + @unittest.skip("can't be implemented for FSMT due to dual vocab.") + def test_resize_tokens_embeddings(self): + pass + + @unittest.skip("Passing inputs_embeds not implemented for FSMT.") + def test_inputs_embeds(self): + pass + + @unittest.skip("model weights aren't tied in FSMT.") + def test_tie_model_weights(self): + pass + + # def test_auto_model(self): + # # XXX: add a tiny model to s3? + # model_name = "facebook/wmt19-ru-en-tiny" + # tiny = AutoModel.from_pretrained(model_name) # same vocab size + # tok = AutoTokenizer.from_pretrained(model_name) # same tokenizer + # inputs_dict = tok.batch_encode_plus(["Hello my friends"], return_tensors="pt") + + # with torch.no_grad(): + # tiny(**inputs_dict) + + +@require_torch +class FSMTHeadTests(unittest.TestCase): + src_vocab_size = 99 + tgt_vocab_size = 99 + langs = ["ru", "en"] + + def _get_config(self): + return FSMTConfig( + src_vocab_size=self.src_vocab_size, + tgt_vocab_size=self.tgt_vocab_size, + langs=self.langs, + d_model=24, + encoder_layers=2, + decoder_layers=2, + encoder_attention_heads=2, + decoder_attention_heads=2, + encoder_ffn_dim=32, + decoder_ffn_dim=32, + max_position_embeddings=48, + eos_token_id=2, + pad_token_id=1, + bos_token_id=0, + return_dict=True, + ) + + def _get_config_and_data(self): + input_ids = torch.tensor( + [ + [71, 82, 18, 33, 46, 91, 2], + [68, 34, 26, 58, 30, 82, 2], + [5, 97, 17, 39, 94, 40, 2], + [76, 83, 94, 25, 70, 78, 2], + [87, 59, 41, 35, 48, 66, 2], + [55, 13, 16, 58, 5, 2, 1], # note padding + [64, 27, 31, 51, 12, 75, 2], + [52, 64, 86, 17, 83, 39, 2], + [48, 61, 9, 24, 71, 82, 2], + [26, 1, 60, 48, 22, 13, 2], + [21, 5, 62, 28, 14, 76, 2], + [45, 98, 37, 86, 59, 48, 2], + [70, 70, 50, 9, 28, 0, 2], + ], + dtype=torch.long, + device=torch_device, + ) + + batch_size = input_ids.shape[0] + config = self._get_config() + return config, input_ids, batch_size + + def test_generate_beam_search(self): + input_ids = torch.Tensor([[71, 82, 2], [68, 34, 2]]).long().to(torch_device) + config = self._get_config() + lm_model = FSMTForConditionalGeneration(config).to(torch_device) + lm_model.eval() + + max_length = 5 + new_input_ids = lm_model.generate( + input_ids.clone(), + do_sample=True, + num_return_sequences=1, + num_beams=2, + no_repeat_ngram_size=3, + max_length=max_length, + ) + self.assertEqual(new_input_ids.shape, (input_ids.shape[0], max_length)) + # TODO(SS): uneven length batches, empty inputs + + def test_shift_tokens_right(self): + input_ids = torch.Tensor([[71, 82, 18, 33, 2, 1, 1], [68, 34, 26, 58, 30, 82, 2]]).long() + shifted = shift_tokens_right(input_ids, 1) + n_pad_before = input_ids.eq(1).float().sum() + n_pad_after = shifted.eq(1).float().sum() + self.assertEqual(shifted.shape, input_ids.shape) + self.assertEqual(n_pad_after, n_pad_before - 1) + self.assertTrue(torch.eq(shifted[:, 0], 2).all()) + + def test_generate_fp16(self): + config, input_ids, batch_size = self._get_config_and_data() + attention_mask = input_ids.ne(1).to(torch_device) + model = FSMTForConditionalGeneration(config).eval().to(torch_device) + if torch_device == "cuda": + model.half() + model.generate(input_ids, attention_mask=attention_mask) + model.generate(num_beams=4, do_sample=True, early_stopping=False, num_return_sequences=3) + + def test_dummy_inputs(self): + config, *_ = self._get_config_and_data() + model = FSMTForConditionalGeneration(config).eval().to(torch_device) + model(**model.dummy_inputs) + + def test_prepare_fsmt_decoder_inputs(self): + config, *_ = self._get_config_and_data() + input_ids = _long_tensor(([4, 4, 2])) + decoder_input_ids = _long_tensor([[26388, 2, config.pad_token_id]]) + ignore = float("-inf") + decoder_input_ids, decoder_attn_mask, causal_mask = _prepare_fsmt_decoder_inputs( + config, input_ids, decoder_input_ids + ) + expected_causal_mask = torch.tensor( + [[0, ignore, ignore], [0, 0, ignore], [0, 0, 0]] # never attend to the final token, because its pad + ).to(input_ids.device) + self.assertEqual(decoder_attn_mask.size(), decoder_input_ids.size()) + self.assertTrue(torch.eq(expected_causal_mask, causal_mask).all()) + + +def _assert_tensors_equal(a, b, atol=1e-12, prefix=""): + """If tensors not close, or a and b arent both tensors, raise a nice Assertion error.""" + if a is None and b is None: + return True + try: + if torch.allclose(a, b, atol=atol): + return True + raise + except Exception: + msg = "{} != {}".format(a, b) + if prefix: + msg = prefix + ": " + msg + raise AssertionError(msg) + + +def _long_tensor(tok_lst): + return torch.tensor(tok_lst, dtype=torch.long, device=torch_device) + + +TOLERANCE = 1e-4 + + +@require_torch +class FSMTModelIntegrationTests(unittest.TestCase): + tokenizers_cache = {} + models_cache = {} + default_mname = "facebook/wmt19-en-ru" + + @cached_property + def default_tokenizer(self): + return self.get_tokenizer(self.default_mname) + + @cached_property + def default_model(self): + return self.get_model(self.default_mname) + + def get_tokenizer(self, mname): + if mname not in self.tokenizers_cache: + self.tokenizers_cache[mname] = FSMTTokenizer.from_pretrained(mname) + return self.tokenizers_cache[mname] + + def get_model(self, mname): + if mname not in self.models_cache: + self.models_cache[mname] = FSMTForConditionalGeneration.from_pretrained(mname).to(torch_device) + if torch_device == "cuda": + self.models_cache[mname].half() + return self.models_cache[mname] + + @slow + def test_inference_no_head(self): + tokenizer = self.default_tokenizer + model = FSMTModel.from_pretrained(self.default_mname).to(torch_device) + + src_text = "My friend computer will translate this for me" + input_ids = tokenizer([src_text], return_tensors="pt")["input_ids"] + input_ids = _long_tensor(input_ids) + inputs_dict = prepare_fsmt_inputs_dict(model.config, input_ids) + with torch.no_grad(): + output = model(**inputs_dict)[0] + expected_shape = torch.Size((1, 10, model.config.tgt_vocab_size)) + self.assertEqual(output.shape, expected_shape) + # expected numbers were generated when en-ru model, using just fairseq's model4.pt + # may have to adjust if switched to a different checkpoint + expected_slice = torch.tensor( + [[-1.5753, -1.5753, 2.8975], [-0.9540, -0.9540, 1.0299], [-3.3131, -3.3131, 0.5219]] + ) + self.assertTrue(torch.allclose(output[:, :3, :3], expected_slice, atol=TOLERANCE)) + + @parameterized.expand( + [ + ["en-ru"], + ["ru-en"], + ["en-de"], + ["de-en"], + ] + ) + @slow + def test_translation(self, pair): + text = { + "en": "Machine learning is great, isn't it?", + "ru": "Машинное обучение - это здорово, не так ли?", + "de": "Maschinelles Lernen ist großartig, oder?", + } + + src, tgt = pair.split("-") + print(f"Testing {src} -> {tgt}") + mname = f"facebook/wmt19-{pair}" + + src_sentence = text[src] + tgt_sentence = text[tgt] + + tokenizer = self.get_tokenizer(mname) + model = self.get_model(mname) + + input_ids = tokenizer.encode(src_sentence, return_tensors="pt") + outputs = model.generate(input_ids) + decoded = tokenizer.decode(outputs[0], skip_special_tokens=True) + assert decoded == tgt_sentence, f"\n\ngot: {decoded}\nexp: {tgt_sentence}\n" + + +@require_torch +class TestSinusoidalPositionalEmbeddings(unittest.TestCase): + padding_idx = 1 + tolerance = 1e-4 + + def test_basic(self): + input_ids = torch.tensor([[4, 10]], dtype=torch.long, device=torch_device) + emb1 = SinusoidalPositionalEmbedding(embedding_dim=6, padding_idx=self.padding_idx, init_size=6).to( + torch_device + ) + emb = emb1(input_ids) + desired_weights = torch.tensor( + [ + [9.0930e-01, 1.9999e-02, 2.0000e-04, -4.1615e-01, 9.9980e-01, 1.0000e00], + [1.4112e-01, 2.9995e-02, 3.0000e-04, -9.8999e-01, 9.9955e-01, 1.0000e00], + ] + ) + self.assertTrue( + torch.allclose(emb[0], desired_weights, atol=self.tolerance), + msg=f"\nexp:\n{desired_weights}\ngot:\n{emb[0]}\n", + ) + + def test_odd_embed_dim(self): + # odd embedding_dim is allowed + SinusoidalPositionalEmbedding.get_embedding( + num_embeddings=4, embedding_dim=5, padding_idx=self.padding_idx + ).to(torch_device) + + # odd num_embeddings is allowed + SinusoidalPositionalEmbedding.get_embedding( + num_embeddings=5, embedding_dim=4, padding_idx=self.padding_idx + ).to(torch_device) + + @unittest.skip("different from marian (needs more research)") + def test_positional_emb_weights_against_marian(self): + + desired_weights = torch.tensor( + [ + [0, 0, 0, 0, 0], + [0.84147096, 0.82177866, 0.80180490, 0.78165019, 0.76140374], + [0.90929741, 0.93651021, 0.95829457, 0.97505713, 0.98720258], + ] + ) + emb1 = SinusoidalPositionalEmbedding(init_size=512, embedding_dim=512, padding_idx=self.padding_idx).to( + torch_device + ) + weights = emb1.weights.data[:3, :5] + # XXX: only the 1st and 3rd lines match - this is testing against + # verbatim copy of SinusoidalPositionalEmbedding from fairseq + self.assertTrue( + torch.allclose(weights, desired_weights, atol=self.tolerance), + msg=f"\nexp:\n{desired_weights}\ngot:\n{weights}\n", + ) + + # test that forward pass is just a lookup, there is no ignore padding logic + input_ids = torch.tensor( + [[4, 10, self.padding_idx, self.padding_idx, self.padding_idx]], dtype=torch.long, device=torch_device + ) + no_cache_pad_zero = emb1(input_ids)[0] + # XXX: only the 1st line matches the 3rd + self.assertTrue( + torch.allclose(torch.tensor(desired_weights, device=torch_device), no_cache_pad_zero[:3, :5], atol=1e-3) + ) diff --git a/tests/test_tokenization_fsmt.py b/tests/test_tokenization_fsmt.py new file mode 100644 index 00000000000000..c3e08d566ad478 --- /dev/null +++ b/tests/test_tokenization_fsmt.py @@ -0,0 +1,147 @@ +# coding=utf-8 +# Copyright 2018 The Google AI Language Team Authors. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# 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 json +import os +import unittest + +from transformers.file_utils import cached_property +from transformers.testing_utils import slow +from transformers.tokenization_fsmt import VOCAB_FILES_NAMES, FSMTTokenizer + +from .test_tokenization_common import TokenizerTesterMixin + + +class FSMTTokenizationTest(TokenizerTesterMixin, unittest.TestCase): + tokenizer_class = FSMTTokenizer + + def setUp(self): + super().setUp() + + # Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt + vocab = [ + "l", + "o", + "w", + "e", + "r", + "s", + "t", + "i", + "d", + "n", + "w", + "r", + "t", + "lo", + "low", + "er", + "low", + "lowest", + "newer", + "wider", + "", + ] + vocab_tokens = dict(zip(vocab, range(len(vocab)))) + merges = ["l o 123", "lo w 1456", "e r 1789", ""] + + self.langs = ["en", "ru"] + config = { + "langs": self.langs, + "src_vocab_size": 10, + "tgt_vocab_size": 20, + } + + self.src_vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["src_vocab_file"]) + self.tgt_vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["tgt_vocab_file"]) + config_file = os.path.join(self.tmpdirname, "tokenizer_config.json") + self.merges_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["merges_file"]) + with open(self.src_vocab_file, "w") as fp: + fp.write(json.dumps(vocab_tokens)) + with open(self.tgt_vocab_file, "w") as fp: + fp.write(json.dumps(vocab_tokens)) + with open(self.merges_file, "w") as fp: + fp.write("\n".join(merges)) + with open(config_file, "w") as fp: + fp.write(json.dumps(config)) + + @cached_property + def tokenizer_ru_en(self): + return FSMTTokenizer.from_pretrained("facebook/wmt19-ru-en") + + @cached_property + def tokenizer_en_ru(self): + return FSMTTokenizer.from_pretrained("facebook/wmt19-en-ru") + + def test_full_tokenizer(self): + """ Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt """ + tokenizer = FSMTTokenizer(self.langs, self.src_vocab_file, self.tgt_vocab_file, self.merges_file) + + text = "lower" + bpe_tokens = ["low", "er"] + tokens = tokenizer.tokenize(text) + self.assertListEqual(tokens, bpe_tokens) + + input_tokens = tokens + [""] + input_bpe_tokens = [14, 15, 20] + self.assertListEqual(tokenizer.convert_tokens_to_ids(input_tokens), input_bpe_tokens) + + @slow + def test_sequence_builders(self): + tokenizer = self.tokenizer_ru_en + + text = tokenizer.encode("sequence builders", add_special_tokens=False) + text_2 = tokenizer.encode("multi-sequence build", add_special_tokens=False) + + encoded_sentence = tokenizer.build_inputs_with_special_tokens(text) + encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2) + + assert encoded_sentence == text + [2] + assert encoded_pair == text + [2] + text_2 + [2] + + @slow + def test_match_encode_decode(self): + tokenizer_enc = self.tokenizer_en_ru + tokenizer_dec = self.tokenizer_ru_en + + targets = [ + [ + "Here's a little song I wrote. Don't worry, be happy.", + [2470, 39, 11, 2349, 7222, 70, 5979, 7, 8450, 1050, 13160, 5, 26, 6445, 7, 2], + ], + ["This is it. No more. I'm done!", [132, 21, 37, 7, 1434, 86, 7, 70, 6476, 1305, 427, 2]], + ] + + # if data needs to be recreated or added, run: + # import torch + # model = torch.hub.load("pytorch/fairseq", "transformer.wmt19.en-ru", checkpoint_file="model4.pt", tokenizer="moses", bpe="fastbpe") + # for src_text, _ in targets: print(f"""[\n"{src_text}",\n {model.encode(src_text).tolist()}\n],""") + + for src_text, tgt_input_ids in targets: + input_ids = tokenizer_enc.encode(src_text, return_tensors="pt")[0].tolist() + self.assertListEqual(input_ids, tgt_input_ids) + + # and decode backward, using the reversed languages model + decoded_text = tokenizer_dec.decode(input_ids, skip_special_tokens=True) + self.assertEqual(decoded_text, src_text) + + @unittest.skip("FSMTConfig.__init__ requires non-optional args") + def test_torch_encode_plus_sent_to_model(self): + pass + + @unittest.skip("FSMTConfig.__init__ requires non-optional args") + def test_np_encode_plus_sent_to_model(self): + pass