XLNet

This model was released on 2019-06-19 and added to Hugging Face Transformers on 2020-11-16.

XLNet

Overview

The XLNet model was proposed in XLNet: Generalized Autoregressive Pretraining for Language Understanding by Zhilin Yang, Zihang Dai, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le. XLnet is an extension of the Transformer-XL model pre-trained using an autoregressive method to learn bidirectional contexts by maximizing the expected likelihood over all permutations of the input sequence factorization order.

The abstract from the paper is the following:

With the capability of modeling bidirectional contexts, denoising autoencoding based pretraining like BERT achieves better performance than pretraining approaches based on autoregressive language modeling. However, relying on corrupting the input with masks, BERT neglects dependency between the masked positions and suffers from a pretrain-finetune discrepancy. In light of these pros and cons, we propose XLNet, a generalized autoregressive pretraining method that (1) enables learning bidirectional contexts by maximizing the expected likelihood over all permutations of the factorization order and (2) overcomes the limitations of BERT thanks to its autoregressive formulation. Furthermore, XLNet integrates ideas from Transformer-XL, the state-of-the-art autoregressive model, into pretraining. Empirically, under comparable experiment settings, XLNet outperforms BERT on 20 tasks, often by a large margin, including question answering, natural language inference, sentiment analysis, and document ranking.

This model was contributed by thomwolf. The original code can be found here.

Usage tips

The specific attention pattern can be controlled at training and test time using the perm_mask input.
Due to the difficulty of training a fully auto-regressive model over various factorization order, XLNet is pretrained using only a sub-set of the output tokens as target which are selected with the target_mapping input.
To use XLNet for sequential decoding (i.e. not in fully bi-directional setting), use the perm_mask and target_mapping inputs to control the attention span and outputs (see examples in examples/pytorch/text-generation/run_generation.py)
XLNet is one of the few models that has no sequence length limit.
XLNet is not a traditional autoregressive model but uses a training strategy that builds on that. It permutes the tokens in the sentence, then allows the model to use the last n tokens to predict the token n+1. Since this is all done with a mask, the sentence is actually fed in the model in the right order, but instead of masking the first n tokens for n+1, XLNet uses a mask that hides the previous tokens in some given permutation of 1,…,sequence length.
XLNet also uses the same recurrence mechanism as Transformer-XL to build long-term dependencies.

Resources

XLNetConfig

[[autodoc]] XLNetConfig

XLNetTokenizer

[[autodoc]] XLNetTokenizer - get_special_tokens_mask - save_vocabulary

XLNetTokenizerFast

[[autodoc]] XLNetTokenizerFast

XLNet specific outputs

[[autodoc]] models.xlnet.modeling_xlnet.XLNetModelOutput

[[autodoc]] models.xlnet.modeling_xlnet.XLNetLMHeadModelOutput

[[autodoc]] models.xlnet.modeling_xlnet.XLNetForSequenceClassificationOutput

[[autodoc]] models.xlnet.modeling_xlnet.XLNetForMultipleChoiceOutput

[[autodoc]] models.xlnet.modeling_xlnet.XLNetForTokenClassificationOutput

[[autodoc]] models.xlnet.modeling_xlnet.XLNetForQuestionAnsweringSimpleOutput

[[autodoc]] models.xlnet.modeling_xlnet.XLNetForQuestionAnsweringOutput

XLNetModel

[[autodoc]] XLNetModel - forward

XLNetLMHeadModel

[[autodoc]] XLNetLMHeadModel - forward

XLNetForSequenceClassification

[[autodoc]] XLNetForSequenceClassification - forward

XLNetForMultipleChoice

[[autodoc]] XLNetForMultipleChoice - forward

XLNetForTokenClassification

[[autodoc]] XLNetForTokenClassification - forward

XLNetForQuestionAnsweringSimple

[[autodoc]] XLNetForQuestionAnsweringSimple - forward

XLNetForQuestionAnswering

[[autodoc]] XLNetForQuestionAnswering - forward