1
Training nGPT
把 Transformer 参数与激活约束到超球面，nGPT 带来表示学习新范式，一文看懂归一化训练的关键设计。
arXiv:2608.01284v1 Announce Type: cross Abstract: The normalized Transformer (nGPT) realizes hyperspherical representation learning by constraining mo…
2
Divisive Normalization Shapes Low-Rank Slow Manifolds for Continuous Working Memory
揭秘大脑连续工作记忆的数学机制，看归一化如何塑造低秩慢流形。
arXiv:2608.01947v1 Announce Type: cross Abstract: The ability to robustly maintain and update continuous variables is a hallmark of working memory. Wh…
3
Normalized Rewards for Preference Optimization
提出归一化奖励方法，提升偏好优化训练稳定性与效果
arXiv:2607.16240v1 Announce Type: cross Abstract: Direct Alignment Algorithms (DAAs) such as DPO have become a common way to post-train and align LLMs…
4
Controllable Accent Normalization via Discrete Diffusion
口音归一化新研究，离散扩散模型实现可控语音转换，论文思路前沿，值得语音方向开发者深挖。
arXiv:2603.14275v2 Announce Type: replace-cross Abstract: Existing accent normalization methods do not typically offer control over accent strength, y…
5
Stabilizing On-Policy Distillation for MLLM Reasoning with Global Normalization
用全局归一化稳定多模态大模型基于策略的蒸馏，提升推理性能与训练效率的创新方法。
arXiv:2606.09091v1 Announce Type: new Abstract: On-policy distillation (OPD) has recently emerged as an important post-training paradigm. By using a s…
6
SpanNorm: Reconciling Training Stability and Performance in Deep Transformers
突破PreNorm与PostNorm的困境：SpanNorm在提升深度Transformer训练稳定性同时保持高性能
arXiv:2601.22580v2 Announce Type: replace-cross Abstract: The success of Large Language Models (LLMs) hinges on the stable training of deep Transforme…
7
Restoring the Sweet Spot: Pass-Rate Weighted Self-Distillation for LLM Reasoning
通过pass-rate加权自蒸馏，恢复LLM推理的“甜蜜点”，破解GRPO归一化带来的学习偏差。
arXiv:2605.27765v1 Announce Type: cross Abstract: Self-Distillation Policy Optimization (SDPO) provides dense token-level credit assignment for reinfo…
8
Normalization Equivariance for Arbitrary Backbones, with Application to Image Denoising
提出归一化等变性的结构先验，可应用于任意骨干网络的图像去噪，有效提升分布偏移健壮性。
arXiv:2605.08193v2 Announce Type: replace-cross Abstract: Normalization Equivariance (NE) is a structural prior that improves robustness to distributi…