Tag
该论文提出 Stability-Aware Generalization Objective（SAGO）框架，主张 LLM 的泛化应评估为稳定性而非准确率，通过生成一致性、内部激活、置信度和响应镜像等多维度变异测量揭示了现有模型普遍存在且与排名相关的泛化不稳定问题。
DeepSeek has published a paper introducing mHC (Manifold-Constrained Hyper-Connections), a fundamental rewrite of the Transformer architecture that stabilizes large models by replacing standard residual connections with mathematically constrained multi-stream pathways.