EPISODE · Nov 18, 2025 · 19 MIN
DoPE: Denoising Rotary Position Embedding
from Daily Paper Cast · host Jingwen Liang, Gengyu Wang
🤗 Upvotes: 64 | cs.CL Authors: Jing Xiong, Liyang Fan, Hui Shen, Zunhai Su, Min Yang, Lingpeng Kong, Ngai Wong Title: DoPE: Denoising Rotary Position Embedding Arxiv: http://arxiv.org/abs/2511.09146v1 Abstract: Rotary Position Embedding (RoPE) in Transformer models has inherent limits that weaken length extrapolation. We reinterpret the attention map with positional encoding as a noisy feature map, and propose Denoising Positional Encoding (DoPE), a training-free method based on truncated matrix entropy to detect outlier frequency bands in the feature map. Leveraging the noise characteristics of the feature map, we further reparameterize it with a parameter-free Gaussian distribution to achieve robust extrapolation. Our method theoretically reveals the underlying cause of the attention sink phenomenon and its connection to truncated matrix entropy. Experiments on needle-in-a-haystack and many-shot in-context learning tasks demonstrate that DoPE significantly improves retrieval accuracy and reasoning stability across extended contexts (up to 64K tokens). The results show that the denoising strategy for positional embeddings effectively mitigates attention sinks and restores balanced attention patterns, providing a simple yet powerful solution for improving length generalization. Our project page is Project: https://The-physical-picture-of-LLMs.github.io
What this episode covers
🤗 Upvotes: 64 | cs.CL Authors: Jing Xiong, Liyang Fan, Hui Shen, Zunhai Su, Min Yang, Lingpeng Kong, Ngai Wong Title: DoPE: Denoising Rotary Position Embedding Arxiv: http://arxiv.org/abs/2511.09146v1 Abstract: Rotary Position Embedding (RoPE) in Transformer models has inherent limits that weaken length extrapolation. We reinterpret the attention map with positional encoding as a noisy feature map, and propose Denoising Positional Encoding (DoPE), a training-free method based on truncated matrix entropy to detect outlier frequency bands in the feature map. Leveraging the noise characteristics of the feature map, we further reparameterize it with a parameter-free Gaussian distribution to achieve robust extrapolation. Our method theoretically reveals the underlying cause of the attention sink phenomenon and its connection to truncated matrix entropy. Experiments on needle-in-a-haystack and many-shot in-context learning tasks demonstrate that DoPE significantly improves retrieval accuracy and reasoning stability across extended contexts (up to 64K tokens). The results show that the denoising strategy for positional embeddings effectively mitigates attention sinks and restores balanced attention patterns, providing a simple yet powerful solution for improving length generalization. Our project page is Project: https://The-physical-picture-of-LLMs.github.io
NOW PLAYING
DoPE: Denoising Rotary Position Embedding
No transcript for this episode yet
Similar Episodes
No similar episodes found.