BERT adopts masked language modeling (MLM) for pre-training and is one of the most successful pre-training models. Since BERT neglects dependency among predicted tokens, XLNet introduces permuted language modeling (PLM) for pre-training to address this problem. However, XLNet does not leverage the full position information of a sentence and thus suffers from position discrepancy between pre-training and fine-tuning. In this paper, we propose MPNet, a novel pre-training method that inherits the advantages of BERT and XLNet and avoids their limitations. MPNet leverages the dependency among predicted tokens through permuted language modeling (vs. MLM in BERT), and takes auxiliary position information as input to make the model see a full sentence and thus reducing the position discrepancy (vs.

Summary and Contributions: In this paper, the authors propose a new pre-trained language model called MPNet, which combines the advantages of both BERT (MLM) and XLNet (PLM). The proposed MPNet leverages the dependency among predicted tokens through PLM and takes auxiliary position information as input to reduce the position discrepancy. In practice, to combine MLM and PLM poses great challenge, and the authors propose several delicate designs to overcome the issues. The experiments are mainly carried out on a base model (i.e., 110M params), and the results show that the proposed MPNet could give consistent and significant improvements over similar baseline models. Overall, the idea of the paper is straightforward and easy to understand, which is a natural extension to combine the MLM and PLM.

This paper proposes a new approach to self-supervised pretraining on text, building very closely on prior work in BERT and XLNet. The paper demonstrates that the approach yields models that are noticably better than comparable models from prior work, and isolates the reason for this in a reasonably thorough ablation. While one reviewer raises concerns about the quality of comparisons, I'm convinced that they are already pretty much up to the standards of the field, and will be fully satisfactory after the promised revisions/additions. While this work is perhaps somewhat incremental, the method is effective and relevant to a highly prominent open problem in ML for text.

BERT adopts masked language modeling (MLM) for pre-training and is one of the most successful pre-training models. Since BERT neglects dependency among predicted tokens, XLNet introduces permuted language modeling (PLM) for pre-training to address this problem. However, XLNet does not leverage the full position information of a sentence and thus suffers from position discrepancy between pre-training and fine-tuning. In this paper, we propose MPNet, a novel pre-training method that inherits the advantages of BERT and XLNet and avoids their limitations. MPNet leverages the dependency among predicted tokens through permuted language modeling (vs.