mirror of
https://github.com/huggingface/transformers.git
synced 2025-07-04 21:30:07 +06:00
Updated model card for mbart and mbart50 (#37619)
* new card for mbart and mbart50 * removed comment BADGES * Update mBart overview Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * fix typo (MBart to mBart) Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * maybe fix typo * update typo and combine notes * changed notes * changed the example sentence * fixed grammatical error and removed some lines from notes example * missed one word * removed documentation resources and added some lines of example code back in notes. --------- Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
This commit is contained in:
Vishesh-Mistry
GitHub
parent
fbfa1dd4db
commit
1d9743edc2
@ -14,154 +14,105 @@ rendered properly in your Markdown viewer.
|
|||||||
|
|
||||||
-->
|
-->
|
||||||
|
|
||||||
# MBart and MBart-50
|
<div style="float: right;">
|
||||||
|
<div class="flex flex-wrap space-x-1">
|
||||||
<div class="flex flex-wrap space-x-1">
|
<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
|
||||||
<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
|
<img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
|
||||||
<img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
|
<img alt="Flax" src="https://img.shields.io/badge/Flax-29a79b.svg?style=flat">
|
||||||
<img alt="Flax" src="https://img.shields.io/badge/Flax-29a79b.svg?style=flat&logo=data:image/png;base64,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
|
<img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
|
||||||
">
|
<img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
|
||||||
<img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
|
</div>
|
||||||
<img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
|
|
||||||
</div>
|
</div>
|
||||||
|
|
||||||
|
# mBART
|
||||||
|
|
||||||
## Overview of MBart
|
[mBART](https://huggingface.co/papers/2001.08210) is a multilingual machine translation model that pretrains the entire translation model (encoder-decoder) unlike previous methods that only focused on parts of the model. The model is trained on a denoising objective which reconstructs the corrupted text. This allows mBART to handle the source language and the target text to translate to.
|
||||||
|
|
||||||
The MBart model was presented in [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov Marjan
|
[mBART-50](https://huggingface.co/paper/2008.00401) is pretrained on an additional 25 languages.
|
||||||
Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
|
|
||||||
|
|
||||||
According to the abstract, MBART is a sequence-to-sequence denoising auto-encoder pretrained on large-scale monolingual
|
You can find all the original mBART checkpoints under the [AI at Meta](https://huggingface.co/facebook?search_models=mbart) organization.
|
||||||
corpora in many languages using the BART objective. mBART is one of the first methods for pretraining a complete
|
|
||||||
sequence-to-sequence model by denoising full texts in multiple languages, while previous approaches have focused only
|
|
||||||
on the encoder, decoder, or reconstructing parts of the text.
|
|
||||||
|
|
||||||
This model was contributed by [valhalla](https://huggingface.co/valhalla). The Authors' code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/mbart)
|
> [!TIP]
|
||||||
|
> Click on the mBART models in the right sidebar for more examples of applying mBART to different language tasks.
|
||||||
|
|
||||||
### Training of MBart
|
The example below demonstrates how to translate text with [`Pipeline`] or the [`AutoModel`] class.
|
||||||
|
|
||||||
MBart is a multilingual encoder-decoder (sequence-to-sequence) model primarily intended for translation task. As the
|
<hfoptions id="usage">
|
||||||
model is multilingual it expects the sequences in a different format. A special language id token is added in both the
|
<hfoption id="Pipeline">
|
||||||
source and target text. The source text format is `X [eos, src_lang_code]` where `X` is the source text. The
|
|
||||||
target text format is `[tgt_lang_code] X [eos]`. `bos` is never used.
|
|
||||||
|
|
||||||
The regular [`~MBartTokenizer.__call__`] will encode source text format passed as first argument or with the `text`
|
```py
|
||||||
keyword, and target text format passed with the `text_label` keyword argument.
|
import torch
|
||||||
|
from transformers import pipeline
|
||||||
|
|
||||||
- Supervised training
|
pipeline = pipeline(
|
||||||
|
task="translation",
|
||||||
```python
|
model="facebook/mbart-large-50-many-to-many-mmt",
|
||||||
>>> from transformers import MBartForConditionalGeneration, MBartTokenizer
|
device=0,
|
||||||
|
torch_dtype=torch.float16,
|
||||||
>>> tokenizer = MBartTokenizer.from_pretrained("facebook/mbart-large-en-ro", src_lang="en_XX", tgt_lang="ro_RO")
|
src_lang="en_XX",
|
||||||
>>> example_english_phrase = "UN Chief Says There Is No Military Solution in Syria"
|
tgt_lang="fr_XX",
|
||||||
>>> expected_translation_romanian = "Şeful ONU declară că nu există o soluţie militară în Siria"
|
)
|
||||||
|
print(pipeline("UN Chief Says There Is No Military Solution in Syria"))
|
||||||
>>> inputs = tokenizer(example_english_phrase, text_target=expected_translation_romanian, return_tensors="pt")
|
|
||||||
|
|
||||||
>>> model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-en-ro")
|
|
||||||
>>> # forward pass
|
|
||||||
>>> model(**inputs)
|
|
||||||
```
|
```
|
||||||
|
|
||||||
- Generation
|
</hfoption>
|
||||||
|
<hfoption id="AutoModel">
|
||||||
|
|
||||||
While generating the target text set the `decoder_start_token_id` to the target language id. The following
|
```py
|
||||||
example shows how to translate English to Romanian using the *facebook/mbart-large-en-ro* model.
|
import torch
|
||||||
|
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
||||||
|
|
||||||
```python
|
article_en = "UN Chief Says There Is No Military Solution in Syria"
|
||||||
>>> from transformers import MBartForConditionalGeneration, MBartTokenizer
|
|
||||||
|
|
||||||
>>> tokenizer = MBartTokenizer.from_pretrained("facebook/mbart-large-en-ro", src_lang="en_XX")
|
model = AutoModelForSeq2SeqLM.from_pretrained("facebook/mbart-large-50-many-to-many-mmt", torch_dtype=torch.bfloat16, attn_implementation="sdpa", device_map="auto")
|
||||||
>>> article = "UN Chief Says There Is No Military Solution in Syria"
|
tokenizer = AutoTokenizer.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
|
||||||
>>> inputs = tokenizer(article, return_tensors="pt")
|
|
||||||
>>> translated_tokens = model.generate(**inputs, decoder_start_token_id=tokenizer.lang_code_to_id["ro_RO"])
|
tokenizer.src_lang = "en_XX"
|
||||||
>>> tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
|
encoded_hi = tokenizer(article_en, return_tensors="pt").to("cuda")
|
||||||
"Şeful ONU declară că nu există o soluţie militară în Siria"
|
generated_tokens = model.generate(**encoded_hi, forced_bos_token_id=tokenizer.lang_code_to_id["fr_XX"], cache_implementation="static")
|
||||||
|
print(tokenizer.batch_decode(generated_tokens, skip_special_tokens=True))
|
||||||
```
|
```
|
||||||
|
|
||||||
## Overview of MBart-50
|
</hfoption>
|
||||||
|
</hfoptions>
|
||||||
|
|
||||||
MBart-50 was introduced in the [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) paper by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav
|
## Notes
|
||||||
Chaudhary, Jiatao Gu, Angela Fan. MBart-50 is created using the original *mbart-large-cc25* checkpoint by extending
|
|
||||||
its embedding layers with randomly initialized vectors for an extra set of 25 language tokens and then pretrained on 50
|
|
||||||
languages.
|
|
||||||
|
|
||||||
According to the abstract
|
- You can check the full list of language codes via `tokenizer.lang_code_to_id.keys()`.
|
||||||
|
- mBART requires a special language id token in the source and target text during training. The source text format is `X [eos, src_lang_code]` where `X` is the source text. The target text format is `[tgt_lang_code] X [eos]`. The `bos` token is never used. The [`~PreTrainedTokenizerBase._call_`] encodes the source text format passed as the first argument or with the `text` keyword. The target text format is passed with the `text_label` keyword.
|
||||||
|
- Set the `decoder_start_token_id` to the target language id for mBART.
|
||||||
|
|
||||||
*Multilingual translation models can be created through multilingual finetuning. Instead of finetuning on one
|
```py
|
||||||
direction, a pretrained model is finetuned on many directions at the same time. It demonstrates that pretrained models
|
import torch
|
||||||
can be extended to incorporate additional languages without loss of performance. Multilingual finetuning improves on
|
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
||||||
average 1 BLEU over the strongest baselines (being either multilingual from scratch or bilingual finetuning) while
|
|
||||||
improving 9.3 BLEU on average over bilingual baselines from scratch.*
|
|
||||||
|
|
||||||
|
model = AutoModelForSeq2SeqLM.from_pretrained("facebook/mbart-large-en-ro", torch_dtype=torch.bfloat16, attn_implementation="sdpa", device_map="auto")
|
||||||
|
tokenizer = MBartTokenizer.from_pretrained("facebook/mbart-large-en-ro", src_lang="en_XX")
|
||||||
|
|
||||||
### Training of MBart-50
|
article = "UN Chief Says There Is No Military Solution in Syria"
|
||||||
|
inputs = tokenizer(article, return_tensors="pt")
|
||||||
|
|
||||||
The text format for MBart-50 is slightly different from mBART. For MBart-50 the language id token is used as a prefix
|
translated_tokens = model.generate(**inputs, decoder_start_token_id=tokenizer.lang_code_to_id["ro_RO"])
|
||||||
for both source and target text i.e the text format is `[lang_code] X [eos]`, where `lang_code` is source
|
tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
|
||||||
language id for source text and target language id for target text, with `X` being the source or target text
|
```
|
||||||
respectively.
|
|
||||||
|
|
||||||
|
- mBART-50 has a different text format. The language id token is used as the prefix for the source and target text. The text format is `[lang_code] X [eos]` where `lang_code` is the source language id for the source text and target language id for the target text. `X` is the source or target text respectively.
|
||||||
|
- Set the `eos_token_id` as the `decoder_start_token_id` for mBART-50. The target language id is used as the first generated token by passing `forced_bos_token_id` to [`~GenerationMixin.generate`].
|
||||||
|
|
||||||
MBart-50 has its own tokenizer [`MBart50Tokenizer`].
|
```py
|
||||||
|
import torch
|
||||||
|
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
||||||
|
|
||||||
- Supervised training
|
model = AutoModelForSeq2SeqLM.from_pretrained("facebook/mbart-large-50-many-to-many-mmt", torch_dtype=torch.bfloat16, attn_implementation="sdpa", device_map="auto")
|
||||||
|
tokenizer = MBartTokenizer.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
|
||||||
|
|
||||||
```python
|
article_ar = "الأمين العام للأمم المتحدة يقول إنه لا يوجد حل عسكري في سوريا."
|
||||||
from transformers import MBartForConditionalGeneration, MBart50TokenizerFast
|
tokenizer.src_lang = "ar_AR"
|
||||||
|
|
||||||
model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-50")
|
encoded_ar = tokenizer(article_ar, return_tensors="pt")
|
||||||
tokenizer = MBart50TokenizerFast.from_pretrained("facebook/mbart-large-50", src_lang="en_XX", tgt_lang="ro_RO")
|
generated_tokens = model.generate(**encoded_ar, forced_bos_token_id=tokenizer.lang_code_to_id["en_XX"])
|
||||||
|
tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
|
||||||
src_text = " UN Chief Says There Is No Military Solution in Syria"
|
```
|
||||||
tgt_text = "Şeful ONU declară că nu există o soluţie militară în Siria"
|
|
||||||
|
|
||||||
model_inputs = tokenizer(src_text, text_target=tgt_text, return_tensors="pt")
|
|
||||||
|
|
||||||
model(**model_inputs) # forward pass
|
|
||||||
```
|
|
||||||
|
|
||||||
- Generation
|
|
||||||
|
|
||||||
To generate using the mBART-50 multilingual translation models, `eos_token_id` is used as the
|
|
||||||
`decoder_start_token_id` and the target language id is forced as the first generated token. To force the
|
|
||||||
target language id as the first generated token, pass the *forced_bos_token_id* parameter to the *generate* method.
|
|
||||||
The following example shows how to translate between Hindi to French and Arabic to English using the
|
|
||||||
*facebook/mbart-50-large-many-to-many* checkpoint.
|
|
||||||
|
|
||||||
```python
|
|
||||||
from transformers import MBartForConditionalGeneration, MBart50TokenizerFast
|
|
||||||
|
|
||||||
article_hi = "संयुक्त राष्ट्र के प्रमुख का कहना है कि सीरिया में कोई सैन्य समाधान नहीं है"
|
|
||||||
article_ar = "الأمين العام للأمم المتحدة يقول إنه لا يوجد حل عسكري في سوريا."
|
|
||||||
|
|
||||||
model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
|
|
||||||
tokenizer = MBart50TokenizerFast.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
|
|
||||||
|
|
||||||
# translate Hindi to French
|
|
||||||
tokenizer.src_lang = "hi_IN"
|
|
||||||
encoded_hi = tokenizer(article_hi, return_tensors="pt")
|
|
||||||
generated_tokens = model.generate(**encoded_hi, forced_bos_token_id=tokenizer.lang_code_to_id["fr_XX"])
|
|
||||||
tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
|
|
||||||
# => "Le chef de l 'ONU affirme qu 'il n 'y a pas de solution militaire en Syria."
|
|
||||||
|
|
||||||
# translate Arabic to English
|
|
||||||
tokenizer.src_lang = "ar_AR"
|
|
||||||
encoded_ar = tokenizer(article_ar, return_tensors="pt")
|
|
||||||
generated_tokens = model.generate(**encoded_ar, forced_bos_token_id=tokenizer.lang_code_to_id["en_XX"])
|
|
||||||
tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
|
|
||||||
# => "The Secretary-General of the United Nations says there is no military solution in Syria."
|
|
||||||
```
|
|
||||||
|
|
||||||
## Documentation resources
|
|
||||||
|
|
||||||
- [Text classification task guide](../tasks/sequence_classification)
|
|
||||||
- [Question answering task guide](../tasks/question_answering)
|
|
||||||
- [Causal language modeling task guide](../tasks/language_modeling)
|
|
||||||
- [Masked language modeling task guide](../tasks/masked_language_modeling)
|
|
||||||
- [Translation task guide](../tasks/translation)
|
|
||||||
- [Summarization task guide](../tasks/summarization)
|
|
||||||
|
|
||||||
## MBartConfig
|
## MBartConfig
|
||||||
|
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user