paper-with-me

홈 › Papers

3D-RPE: Enhancing Long-Context Modeling Through 3D Rotary Position Encoding

2024-06-14 · Xindian Ma, Wenyuan Liu, Peng Zhang, Nan Xu

Inspired by the Bloch Sphere representation, we propose a novel rotary position encoding on a three-dimensional sphere, named 3D Rotary Position Encoding (3D-RPE). 3D-RPE is an advanced version of the widely used 2D Rotary Position Encoding (RoPE), with two major advantages for modeling long contexts: controllable long-term decay and improved position resolution. For controllable long-term decay, 3D-RPE allows for the regulation of long-term decay within the chunk size, ensuring the modeling of relative positional information between tokens at a distant relative position. For enhanced position resolution, 3D-RPE can mitigate the degradation of position resolution caused by position interpolation on RoPE. We have conducted experiments on long-context Natural Language Understanding (NLU) and long-sequence Language Modeling (LM) tasks. From the experimental results, 3D-RPE achieved performance improvements over RoPE, especially in long-context NLU tasks.

📄 PDF Abstract BibTeX arXiv:2406.09897

Code (0)

등록된 구현이 없습니다.

Tasks

Language ModelingLanguage ModellingNatural Language UnderstandingPosition

Similar Papers 제목 키워드 기반

Breaking the Stage Barrier: A Novel Single-Stage Approach to Long Context Extension for Large Language Models

2024-12-10 · Haoran Lian, Junmin Chen, Wei Huang, Yizhe Xiong 외

Recently, Large language models (LLMs) have revolutionized Natural Language Processing (NLP). Pretrained LLMs, due to limited training context size, struggle with handling long token sequences, limiting their performance…

Continual PretrainingLanguage ModelingLanguage ModellingPosition

Extending Context Window of Large Language Models from a Distributional Perspective

2024-10-02 · Yingsheng Wu, Yuxuan Gu, Xiaocheng Feng, Weihong Zhong 외

Scaling the rotary position embedding (RoPE) has become a common method for extending the context window of RoPE-based large language models (LLMs). However, existing scaling methods often rely on empirical approaches an…

16k8k

Beyond Real: Imaginary Extension of Rotary Position Embeddings for Long-Context LLMs

2025-12-08 · Xiaoran Liu, Yuerong Song, Zhigeng Liu, Zengfeng Huang 외 arxiv

Rotary Position Embeddings (RoPE) have become a standard for encoding sequence order in Large Language Models (LLMs) by applying rotations to query and key vectors in the complex plane. Standard implementations, however,…

The Rotary Position Embedding May Cause Dimension Inefficiency in Attention Heads for Long-Distance Retrieval

2025-02-16 · Ting-Rui Chiang, Dani Yogatama

The Rotary Position Embedding (RoPE) is widely used in the attention heads of many large language models (LLM). It rotates dimensions in the query and the key vectors by different angles according to their positions in t…

PositionQuestion Answering

Remember to Forget: Gated Adaptive Positional Encoding

2026-05-11 · Riccardo Ali, Alessio Borgi, Christopher Irwin, Mario Severino 외 arxiv

Rotary Positional Encoding (RoPE) is widely used in modern large language models. However, when sequences are extended beyond the range seen during training, rotary phases can enter out-of-distribution regimes, leading t…