This model was released on 2021-05-02 and added to Hugging Face Transformers on 2022-01-29.

</div>

XLM-RoBERTa-XL

XLM-RoBERTa-XL is a 3.5B parameter multilingual masked language model pretrained on 100 languages. It shows that by scaling model capacity, multilingual models demonstrates strong performance on high-resource languages and can even zero-shot low-resource languages.

You can find all the original XLM-RoBERTa-XL checkpoints under the AI at Meta organization.

[!TIP] Click on the XLM-RoBERTa-XL models in the right sidebar for more examples of how to apply XLM-RoBERTa-XL to different cross-lingual tasks like classification, translation, and question answering.

The example below demonstrates how to predict the <mask> token with [Pipeline], [AutoModel], and from the command line.

from transformers import pipeline


pipeline = pipeline(
    task="fill-mask",
    model="facebook/xlm-roberta-xl",
    device=0
)
pipeline("Bonjour, je suis un modèle <mask>.")

import torch

from transformers import AutoModelForMaskedLM, AutoTokenizer


tokenizer = AutoTokenizer.from_pretrained(
    "facebook/xlm-roberta-xl",
)
model = AutoModelForMaskedLM.from_pretrained(
    "facebook/xlm-roberta-xl",
    device_map="auto",
    attn_implementation="sdpa"
)
inputs = tokenizer("Bonjour, je suis un modèle <mask>.", return_tensors="pt").to(model.device)

with torch.no_grad():
    outputs = model(**inputs)
    predictions = outputs.logits

masked_index = torch.where(inputs['input_ids'] == tokenizer.mask_token_id)[1]
predicted_token_id = predictions[0, masked_index].argmax(dim=-1)
predicted_token = tokenizer.decode(predicted_token_id)

print(f"The predicted token is: {predicted_token}")

Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the Quantization overview for more available quantization backends.

The example below uses torchao to only quantize the weights to int4.

import torch

from transformers import AutoModelForMaskedLM, AutoTokenizer, TorchAoConfig


quantization_config = TorchAoConfig("int4_weight_only", group_size=128)
tokenizer = AutoTokenizer.from_pretrained(
    "facebook/xlm-roberta-xl",
)
model = AutoModelForMaskedLM.from_pretrained(
    "facebook/xlm-roberta-xl",
    device_map="auto",
    attn_implementation="sdpa",
    quantization_config=quantization_config
)
inputs = tokenizer("Bonjour, je suis un modèle <mask>.", return_tensors="pt").to(model.device)

with torch.no_grad():
    outputs = model(**inputs)
    predictions = outputs.logits

masked_index = torch.where(inputs['input_ids'] == tokenizer.mask_token_id)[1]
predicted_token_id = predictions[0, masked_index].argmax(dim=-1)
predicted_token = tokenizer.decode(predicted_token_id)

print(f"The predicted token is: {predicted_token}")

Notes

• Unlike some XLM models, XLM-RoBERTa-XL doesn't require lang tensors to understand which language is used. It automatically determines the language from the input ids.

XLMRobertaXLConfig

[[autodoc]] XLMRobertaXLConfig

XLMRobertaXLModel

[[autodoc]] XLMRobertaXLModel - forward

XLMRobertaXLForCausalLM

[[autodoc]] XLMRobertaXLForCausalLM - forward

XLMRobertaXLForMaskedLM

[[autodoc]] XLMRobertaXLForMaskedLM - forward

XLMRobertaXLForSequenceClassification

[[autodoc]] XLMRobertaXLForSequenceClassification - forward

XLMRobertaXLForMultipleChoice

[[autodoc]] XLMRobertaXLForMultipleChoice - forward

XLMRobertaXLForTokenClassification

[[autodoc]] XLMRobertaXLForTokenClassification - forward

XLMRobertaXLForQuestionAnswering

[[autodoc]] XLMRobertaXLForQuestionAnswering - forward