A Unified View of Attention and Residual Sinks: Outlier-Driven Rescaling is Essential for Transformer Training

• URL: http://arxiv.org/abs/2601.22966v1
• Date: Fri, 30 Jan 2026 13:29:45 GMT
• Title: A Unified View of Attention and Residual Sinks: Outlier-Driven Rescaling is Essential for Transformer Training
• Authors: Zihan Qiu, Zeyu Huang, Kaiyue Wen, Peng Jin, Bo Zheng, Yuxin Zhou, Haofeng Huang, Zekun Wang, Xiao Li, Huaqing Zhang, Yang Xu, Haoran Lian, Siqi Zhang, Rui Men, Jianwei Zhang, Ivan Titov, Dayiheng Liu, Jingren Zhou, Junyang Lin,
• Abstract summary: Outliers function jointly with normalization.<n>Outliers serve more as rescale factors rather than contributors.<n>Outliers can be absorbed into learnable parameters or mitigated via explicit gated rescaling.
• Score: 86.64715217940274
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We investigate the functional role of emergent outliers in large language models, specifically attention sinks (a few tokens that consistently receive large attention logits) and residual sinks (a few fixed dimensions with persistently large activations across most tokens). We hypothesize that these outliers, in conjunction with the corresponding normalizations (\textit{e.g.}, softmax attention and RMSNorm), effectively rescale other non-outlier components. We term this phenomenon \textit{outlier-driven rescaling} and validate this hypothesis across different model architectures and training token counts. This view unifies the origin and mitigation of both sink types. Our main conclusions and observations include: (1) Outliers function jointly with normalization: removing normalization eliminates the corresponding outliers but degrades training stability and performance; directly clipping outliers while retaining normalization leads to degradation, indicating that outlier-driven rescaling contributes to training stability. (2) Outliers serve more as rescale factors rather than contributors, as the final contributions of attention and residual sinks are significantly smaller than those of non-outliers. (3) Outliers can be absorbed into learnable parameters or mitigated via explicit gated rescaling, leading to improved training performance (average gain of 2 points) and enhanced quantization robustness (1.2 points degradation under W4A4 quantization).

Related papers

• Attention Projection Mixing with Exogenous Anchors [0.0]
  Cross-layer reuse of early attention projections can improve data efficiency, but it creates a structural conflict.<n>We show this conflict is a hidden limiter of internal-anchor designs.<n>We propose ExoFormer, which resolves the conflict by learning anchor projections outside the sequential layer stack.
  arXiv  Detail & Related papers  (2026-01-13T01:52:19Z)
• STORE: Semantic Tokenization, Orthogonal Rotation and Efficient Attention for Scaling Up Ranking Models [11.965535230928372]
  Store is a unified and scalable token-based ranking framework built upon three core innovations.<n>Our framework consistently improves prediction accuracy(online CTR by 2.71%, AUC by 1.195%) and training effeciency (1.84 throughput)
  arXiv  Detail & Related papers  (2025-11-24T06:20:02Z)
• UPCORE: Utility-Preserving Coreset Selection for Balanced Unlearning [57.081646768835704]
  User specifications or legal frameworks often require information to be removed from pretrained models, including large language models (LLMs)<n>This requires deleting or "forgetting" a set of data points from an already-trained model, which typically degrades its performance on other data points.<n>We propose UPCORE, a method-agnostic data selection framework for mitigating collateral damage during unlearning.
  arXiv  Detail & Related papers  (2025-02-20T22:51:10Z)
• ALTBI: Constructing Improved Outlier Detection Models via Optimization of Inlier-Memorization Effect [2.3961612657966946]
  Outlier detection (OD) is the task of identifying unusual observations (or outliers) from a given or upcoming data.<n>Inlier-memorization (IM) effect suggests that generative models memorize inliers before outliers in early learning stages.<n>We propose a theoretically principled method to address UOD tasks by maximally utilizing the IM effect.
  arXiv  Detail & Related papers  (2024-08-19T08:40:53Z)
• SINDER: Repairing the Singular Defects of DINOv2 [61.98878352956125]
  Vision Transformer models trained on large-scale datasets often exhibit artifacts in the patch token they extract. We propose a novel fine-tuning smooth regularization that rectifies structural deficiencies using only a small dataset.
  arXiv  Detail & Related papers  (2024-07-23T20:34:23Z)
• Regularized Linear Regression for Binary Classification [20.710343135282116]
  Regularized linear regression is a promising approach for binary classification problems in which the training set has noisy labels. We show that for large enough regularization strength, the optimal weights concentrate around two values of opposite sign. We observe that in many cases the corresponding "compression" of each weight to a single bit leads to very little loss in performance.
  arXiv  Detail & Related papers  (2023-11-03T23:18:21Z)
• Robust Outlier Rejection for 3D Registration with Variational Bayes [70.98659381852787]
  We develop a novel variational non-local network-based outlier rejection framework for robust alignment. We propose a voting-based inlier searching strategy to cluster the high-quality hypothetical inliers for transformation estimation.
  arXiv  Detail & Related papers  (2023-04-04T03:48:56Z)
• Monotonicity and Double Descent in Uncertainty Estimation with Gaussian Processes [52.92110730286403]
  It is commonly believed that the marginal likelihood should be reminiscent of cross-validation metrics and that both should deteriorate with larger input dimensions. We prove that by tuning hyper parameters, the performance, as measured by the marginal likelihood, improves monotonically with the input dimension. We also prove that cross-validation metrics exhibit qualitatively different behavior that is characteristic of double descent.
  arXiv  Detail & Related papers  (2022-10-14T08:09:33Z)
• Outlier Suppression: Pushing the Limit of Low-bit Transformer Language Models [57.933500846742234]
  Recent work recognizes that structured outliers are the critical bottleneck for quantization performance. We propose an outlier suppression framework including two components: Gamma Migration and Token-Wise Clipping. This framework effectively suppresses the outliers and can be used in a plug-and-play mode.
  arXiv  Detail & Related papers  (2022-09-27T12:05:59Z)
• Deconfounding Scores: Feature Representations for Causal Effect Estimation with Weak Overlap [140.98628848491146]
  We introduce deconfounding scores, which induce better overlap without biasing the target of estimation. We show that deconfounding scores satisfy a zero-covariance condition that is identifiable in observed data. In particular, we show that this technique could be an attractive alternative to standard regularizations.
  arXiv  Detail & Related papers  (2021-04-12T18:50:11Z)

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.