Can I read DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing on EtoBox?

DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing by He, Pengcheng; Gao, Jianfeng; Chen, Weizhu is a scholarly article available to read on EtoBox.

What is DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing about?

This paper presents a new pre-trained language model, DeBERTaV3, which improves the original DeBERTa model by replacing mask language modeling (MLM) with replaced token detection (RTD), a more sample-efficient pre-training task. Our analysis shows that vanilla embedding sharing in ELECTRA hurts training efficiency and model performance. This is because the training losses of the discriminator and the generator pull token embeddings in different directions, creating the "tug-of-war" dynamics. We thus propose a new gradient-disentangled embedding sharing method that avoids the tug-of-war dynamics, improving both training efficiency and the quality of the pre-trained model. We have pre-trained DeBERTaV3 using the same settings as DeBERTa to demonstrate its exceptional performance on a wide range of downstream natural language understanding (NLU) tasks. Taking the GLUE benchmark with eight tasks as an example, the DeBERTaV3 Large model achieves a 91.37% average score, which is 1.37% over DeBERTa and 1.91% over ELECTRA, setting a new state-of-the-art (SOTA) among the models with a similar structure. Furthermore, we have pre-trained a multi-lingual model mDeBERTa and observed a larger im

Author

He, Pengcheng; Gao, Jianfeng; Chen, Weizhu

Published

2021

Language

EN