Understanding How Positional Encodings Work in Transformer Model (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing studies have reported superiority of relative PEs in translation tasks. |
| Approach: | They analyze in which part of a transformer model PEs work and compare them using experiments . they find that relative PEs should be added only to query and key of attention mechanism . |
| Outcome: | The results show that relative and absolute PEs work in a transformer model, and should be added to the query and key of an attention mechanism, not to the value. |
Similar Papers
Improve Transformer Models with Better Relative Position Embeddings (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for generating position embeddings are not fully utilized in NLP tasks. |
| Approach: | They propose to generalize the absolute position embedding to a generalized relative position embedded method . they also propose to use the relative embeddable method to improve the accuracy of large models . |
| Outcome: | The proposed method improves accuracy on the SQuAD1.1 dataset compared to previous methods . it can be easily adopted as a drop-in replacement for improving accuracy of large models . |
The Locality and Symmetry of Positional Encodings (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing work uses fixed or learned positional encodings to infuse word order information into transformer-based language models. |
| Approach: | They conduct a systematic study of positional encodings in Bidirectional Masked Language Models. |
| Outcome: | The proposed model improves performance across a wide range of natural language understanding tasks. |
Self-Attention with Relative Position Representations (N18-2)
Copied to clipboard
| Challenge: | Recent approaches to sequence to sequence learning leverage recurrence, convolution, attention or combination of recurrent and convolutional neural networks. |
| Approach: | They propose an approach that extends the self-attention mechanism to consider representations of relative positions, or distances between sequence elements. |
| Outcome: | The proposed approach yields 1.3 BLEU and 0.3 BLUE on translation tasks . it is based on a relation-aware self-attention mechanism that can generalize to arbitrary graph-labeled inputs. |
The Case for Translation-Invariant Self-Attention in Transformer-Based Language Models (2021.acl-short)
Copied to clipboard
| Challenge: | Existing approaches for positional dependencies do not satisfy all criteria for optimal position encoding. |
| Approach: | They propose a translation-invariant self-attention approach that accounts for relative position between tokens in an interpretable fashion without conventional embeddings. |
| Outcome: | The proposed model improves on regular ALBERT on GLUE tasks while adding orders of magnitude less positional parameters. |
Absolute Position Embedding Learns Sinusoid-like Waves for Attention Based on Relative Position (2023.emnlp-main)
Copied to clipboard
| Challenge: | Attention weight is a clue to interpret how a Transformer-based model makes an inference. |
| Approach: | They analyze the mechanism behind the concentration of attention on nearby tokens . they find that attention in some heads is largely determined by relative positions . |
| Outcome: | The attention weights of the self-attention in a Transformer-based model are analyzed . the model can learn relationships between tokens while allowing parallelization, they show . |
Why Are Positional Encodings Nonessential for Deep Autoregressive Transformers? A Petroglyph Revisited (2025.findings-acl)
Copied to clipboard
| Challenge: | Autoregressive Transformer language models do not require explicit positional encodings (PEs) this is because a cascade of (permutation invariant) set processors collectively exhibit sequence-sensitive behavior in the autoregressively setting. |
| Approach: | They propose to explain why autoregressive Transformers require explicit positional encodings (PEs) this property has been known since early efforts adopting the Transformer for language modeling . |
| Outcome: | The proposed model can distinguish sequences with permuted tokens without the need for explicit PEs. |
Length Extrapolation of Transformers: A Survey from the Perspective of Positional Encoding (2024.findings-emnlp)
Copied to clipboard
Liang Zhao, Xiachong Feng, Xiaocheng Feng, Weihong Zhong, Dongliang Xu, Qing Yang, Hongtao Liu, Bing Qin, Ting Liu
| Challenge: | Existing methods to enhance length extrapolation of large language models have been developed, but a systematic survey is lacking. |
| Approach: | They propose to examine the effects of positional encoding on length extrapolation. |
| Outcome: | The proposed methods improve the extrapolation of large language models, but they are still lacking a systematic survey. |
What Do Position Embeddings Learn? An Empirical Study of Pre-Trained Language Model Positional Encoding (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing work on pre-trained Transformers has focused on learning the meaning of positions . Embedding the position information in the self-attention mechanism is also an indispensable factor in NLP . |
| Approach: | They propose to use feature-level analysis to examine pre-trained Transformers' position embeddings . they also use empirical experiments to determine the appropriate positional encoding function . |
| Outcome: | The results of the empirical study can guide future work to choose the appropriate positional encoding function for specific tasks. |
PermuteFormer: Efficient Relative Position Encoding for Long Sequences (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing Transformers that scale to long sequences are not compatible with relative position encoding. |
| Approach: | They propose a Performer-based model with relative position encoding that scales linearly on long sequences. |
| Outcome: | The proposed model outperforms performer on long sequences with no computational overhead and outperformed vanilla Transformer on most of the tasks. |
Transformer Language Models without Positional Encodings Still Learn Positional Information (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Using positional embeddings, Causal transformer language models learn an implicit notion of absolute positions. |
| Approach: | They propose to use positional embeddings to encode positional information in transformer language models. |
| Outcome: | The proposed model learns an implicit notion of absolute positions across datasets, model sizes, and sequence lengths. |