AutoNorm: Understanding Adaptive Normalization in Transformers through Differentiable Gating

• URL: http://arxiv.org/abs/2607.10593v1
• Date: Sun, 12 Jul 2026 06:12:37 GMT
• Title: AutoNorm: Understanding Adaptive Normalization in Transformers through Differentiable Gating
• Abstract summary: Normalization is a critical component for stabilizing Transformer training.<n>In this paper, we investigate a key optimization challenge in differentiable normalization gating.<n>We propose AutoNorm-S (Stabilized), a training strategy that mitigates optimization instability through a gate-freezing schedule.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Normalization is a critical component for stabilizing Transformer training, yet the choice between static strategies such as Layer Normalization (LN) and adaptive alternatives remains largely task-dependent. In this paper, we investigate a key optimization challenge in differentiable normalization gating. Our experiments show that, on relatively stationary vision tasks, the high gradient variance introduced by Gumbel-Softmax gating can hinder convergence of the routing mechanism, causing learned gates to underperform simple random selection. In contrast, on non-stationary language modeling and classification tasks, sustained gating diversity enables the model to learn more effective layer-wise normalization policies. Motivated by these observations, we propose AutoNorm-S (Stabilized), a training strategy that mitigates optimization instability through a gate-freezing schedule. AutoNorm-S achieves competitive or improved performance across multiple benchmarks, outperforming adaptive normalization baselines on NLP datasets, including PTB and SST-2, while remaining competitive on standard vision benchmarks. These results suggest that decoupling normalization selection from optimization noise provides a practical and principled approach for adaptive normalization in Transformer architectures.

Related papers

• LiFT: Local Search via Linear Programming for Overfitting-Controlled Transformers [0.16866396683295748]
  This paper proposes a Linear Programming (LP)-based local search framework for fine-tuning transformer models with explicit control against overfitting.<n>Experiments on GPT-2 Small on WikiText-2 demonstrate that LiFT enables effective adaptation through selective tuning of transformer blocks and regularization parameters.
  arXiv  Detail & Related papers  (2026-06-15T05:42:59Z)
• DVAO: Dynamic Variance-adaptive Advantage Optimization for Multi-reward Reinforcement Learning [4.844785181175473]
  Group Relative Policy Optimization offers an efficient, value-model-free alternative to Proximal Policy Optimization.<n>Standard scalarization practices, such as Reward Combination and Advantage Combination, suffer from significant drawbacks.<n>We propose Dynamic Variance-adaptive Advantage Optimization (DVAO), which dynamically adjusts combination weights based on the empirical reward variance of each objective.
  arXiv  Detail & Related papers  (2026-05-25T08:55:16Z)
• Too Correct to Learn: Reinforcement Learning on Saturated Reasoning Data [55.84428098924793]
  We propose Constrained Uniform Top-K Sampling (CUTS), a parameter-free decoding strategy enforcing structure-preserving exploration.<n>We integrate this into Mixed-CUTS, a training framework synergizing exploitative and exploratory rollouts to amplify intra-group advantage variance.<n> Notably, Mixed-CUTS improves Pass@1 accuracy on the challenging AIME25 benchmark by up to 15.1% over standard GRPO.
  arXiv  Detail & Related papers  (2026-04-20T16:43:28Z)
• Not All Preferences Are Created Equal: Stability-Aware and Gradient-Efficient Alignment for Reasoning Models [52.48582333951919]
  We propose a dynamic framework designed to enhance alignment reliability by maximizing the Signal-to-Noise Ratio of policy updates.<n>SAGE (Stability-Aware Gradient Efficiency) integrates a coarse-grained curriculum mechanism that refreshes candidate pools based on model competence.<n> Experiments on multiple mathematical reasoning benchmarks demonstrate that SAGE significantly accelerates convergence and outperforms static baselines.
  arXiv  Detail & Related papers  (2026-02-01T12:56:10Z)
• A Constrained Optimization Perspective of Unrolled Transformers [77.12297732942095]
  We introduce a constrained optimization framework for training transformers that behave like optimization descent algorithms.<n>We observe constrained transformers achieve stronger to perturbations robustness and maintain higher out-of-distribution generalization.
  arXiv  Detail & Related papers  (2026-01-24T02:12:39Z)
• MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization [56.074760766965085]
  Group-Relative Policy Optimization has emerged as an efficient paradigm for aligning Large Language Models (LLMs)<n>We propose MAESTRO, which treats reward scalarization as a dynamic latent policy, leveraging the model's terminal hidden states as a semantic bottleneck.<n>We formulate this as a contextual bandit problem within a bi-level optimization framework, where a lightweight Conductor network co-evolves with the policy by utilizing group-relative advantages as a meta-reward signal.
  arXiv  Detail & Related papers  (2026-01-12T05:02:48Z)
• Reliable Optimization Under Noise in Quantum Variational Algorithms [0.05219568203653522]
  We show that Variational Quantum Eigensolver is severely challenged by finite-shot sampling noise.<n>We identify adaptive metaheuristics as the most effective and resilient strategies.
  arXiv  Detail & Related papers  (2025-11-11T14:21:43Z)
• Gradient Correction in Federated Learning with Adaptive Optimization [19.93709245766609]
  We propose tt FAdamGC, the first algorithm to integrate client-drift compensation into adaptive optimization.<n>We show that tt FAdamGC consistently outperforms existing methods in total communication and cost across varying levels of data.
  arXiv  Detail & Related papers  (2025-02-04T21:21:30Z)
• Adaptive Experimentation at Scale: A Computational Framework for Flexible Batches [7.390918770007728]
  Motivated by practical instances involving a handful of reallocations in which outcomes are measured in batches, we develop an adaptive-driven experimentation framework. Our main observation is that normal approximations, which are universal in statistical inference, can also guide the design of adaptive algorithms.
  arXiv  Detail & Related papers  (2023-03-21T04:17:03Z)
• Efficient Semantic Image Synthesis via Class-Adaptive Normalization [116.63715955932174]
  Class-adaptive normalization (CLADE) is a lightweight but equally-effective variant that is only adaptive to semantic class. We introduce intra-class positional map encoding calculated from semantic layouts to modulate the normalization parameters of CLADE. The proposed CLADE can be generalized to different SPADE-based methods while achieving comparable generation quality compared to SPADE.
  arXiv  Detail & Related papers  (2020-12-08T18:59:32Z)
• Stochastic batch size for adaptive regularization in deep network optimization [63.68104397173262]
  We propose a first-order optimization algorithm incorporating adaptive regularization applicable to machine learning problems in deep learning framework. We empirically demonstrate the effectiveness of our algorithm using an image classification task based on conventional network models applied to commonly used benchmark datasets.
  arXiv  Detail & Related papers  (2020-04-14T07:54:53Z)

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.