Improve Transformer Pre-Training with Decoupled Directional Relative Position Encoding and Representation Differentiations

• URL: http://arxiv.org/abs/2210.04246v1
• Date: Sun, 9 Oct 2022 12:35:04 GMT
• Title: Improve Transformer Pre-Training with Decoupled Directional Relative Position Encoding and Representation Differentiations
• Authors: Haojie Zhang, Mingfei Liang, Ruobing Xie, Zhenlong Sun, Bo Zhang, Leyu Lin
• Abstract summary: We revisit the Transformer-based pre-trained language models and identify two problems that may limit the expressiveness of the model. Existing relative position encoding models confuse two heterogeneous information: relative distance and direction. We propose two novel techniques to improve pre-trained language models.
• Score: 23.2969212998404
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In this work, we revisit the Transformer-based pre-trained language models and identify two problems that may limit the expressiveness of the model. Firstly, existing relative position encoding models (e.g., T5 and DEBERTA) confuse two heterogeneous information: relative distance and direction. It may make the model unable to capture the associative semantics of the same direction or the same distance, which in turn affects the performance of downstream tasks. Secondly, we notice the pre-trained BERT with Mask Language Modeling (MLM) pre-training objective outputs similar token representations and attention weights of different heads, which may impose difficulties in capturing discriminative semantic representations. Motivated by the above investigation, we propose two novel techniques to improve pre-trained language models: Decoupled Directional Relative Position (DDRP) encoding and MTH pre-training objective. DDRP decouples the relative distance features and the directional features in classical relative position encoding for better position information understanding. MTH designs two novel auxiliary losses besides MLM to enlarge the dissimilarities between (a) last hidden states of different tokens, and (b) attention weights of different heads, alleviating homogenization and anisotropic problem in representation learning for better optimization. Extensive experiments and ablation studies on GLUE benchmark demonstrate the effectiveness of our proposed methods.

Related papers

• Zero-Shot Embeddings Inform Learning and Forgetting with Vision-Language Encoders [6.7181844004432385]
  The Inter-Intra Modal Measure (IIMM) functions as a strong predictor of performance changes with fine-tuning. Fine-tuning on tasks with higher IIMM scores produces greater in-domain performance gains but also induces more severe out-of-domain performance degradation. With only a single forward pass of the target data, practitioners can leverage this key insight to evaluate the degree to which a model can be expected to improve following fine-tuning.
  arXiv  Detail & Related papers  (2024-07-22T15:35:09Z)
• Dude: Dual Distribution-Aware Context Prompt Learning For Large Vision-Language Model [27.56988000960972]
  We introduce a new framework based on a dual context of both domain-shared and class-specific contexts. Such dual prompt methods enhance the model's feature representation by joining implicit and explicit factors encoded in Large Language Models. We also formulate the Unbalanced Optimal Transport (UOT) theory to quantify the relationships between constructed prompts and visual tokens.
  arXiv  Detail & Related papers  (2024-07-05T13:15:29Z)
• Morphing Tokens Draw Strong Masked Image Models [28.356863521946607]
  Masked image modeling (MIM) is a promising option for training Vision Transformers. We introduce a novel self-supervision signal called Dynamic Token Morphing (DTM) DTM dynamically aggregates contextually related tokens to yield contextualized targets.
  arXiv  Detail & Related papers  (2023-12-30T14:53:09Z)
• Location Sensitive Embedding for Knowledge Graph Reasoning [0.0]
  Key challenge in translational distance models is their inability to effectively differentiate between 'head' and 'tail' entities in graphs. To address this problem, a novel location-sensitive embedding (LSE) method has been developed. LSE innovatively modifies the head entity using relation-specific mappings, conceptualizing relations as linear transformations rather than mere translations. Experiments conducted on four large-scale KG datasets for link prediction show LSEd either outperforms or is competitive with state-of-the-art related works.
  arXiv  Detail & Related papers  (2023-12-01T22:35:19Z)
• VECO 2.0: Cross-lingual Language Model Pre-training with Multi-granularity Contrastive Learning [56.47303426167584]
  We propose a cross-lingual pre-trained model VECO2.0 based on contrastive learning with multi-granularity alignments. Specifically, the sequence-to-sequence alignment is induced to maximize the similarity of the parallel pairs and minimize the non-parallel pairs. token-to-token alignment is integrated to bridge the gap between synonymous tokens excavated via the thesaurus dictionary from the other unpaired tokens in a bilingual instance.
  arXiv  Detail & Related papers  (2023-04-17T12:23:41Z)
• On the Role of Bidirectionality in Language Model Pre-Training [85.14614350372004]
  We study the role of bidirectionality in next token prediction, text infilling, zero-shot priming and fine-tuning. We train models with up to 6.7B parameters, and find differences to remain consistent at scale.
  arXiv  Detail & Related papers  (2022-05-24T02:25:05Z)
• Exploring Unsupervised Pretraining Objectives for Machine Translation [99.5441395624651]
  Unsupervised cross-lingual pretraining has achieved strong results in neural machine translation (NMT) Most approaches adapt masked-language modeling (MLM) to sequence-to-sequence architectures, by masking parts of the input and reconstructing them in the decoder. We compare masking with alternative objectives that produce inputs resembling real (full) sentences, by reordering and replacing words based on their context.
  arXiv  Detail & Related papers  (2021-06-10T10:18:23Z)
• SLUA: A Super Lightweight Unsupervised Word Alignment Model via Cross-Lingual Contrastive Learning [79.91678610678885]
  We propose a super lightweight unsupervised word alignment model (SLUA) Experimental results on several public benchmarks demonstrate that our model achieves competitive, if not better, performance. Notably, we recognize our model as a pioneer attempt to unify bilingual word embedding and word alignments.
  arXiv  Detail & Related papers  (2021-02-08T05:54:11Z)
• Unsupervised Paraphrasing with Pretrained Language Models [85.03373221588707]
  We propose a training pipeline that enables pre-trained language models to generate high-quality paraphrases in an unsupervised setting. Our recipe consists of task-adaptation, self-supervision, and a novel decoding algorithm named Dynamic Blocking. We show with automatic and human evaluations that our approach achieves state-of-the-art performance on both the Quora Question Pair and the ParaNMT datasets.
  arXiv  Detail & Related papers  (2020-10-24T11:55:28Z)
• Towards Accurate Knowledge Transfer via Target-awareness Representation Disentanglement [56.40587594647692]
  We propose a novel transfer learning algorithm, introducing the idea of Target-awareness REpresentation Disentanglement (TRED) TRED disentangles the relevant knowledge with respect to the target task from the original source model and used as a regularizer during fine-tuning the target model. Experiments on various real world datasets show that our method stably improves the standard fine-tuning by more than 2% in average.
  arXiv  Detail & Related papers  (2020-10-16T17:45:08Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.