用统一框架解读模型训练状态，揭示内部信号的深层关联。
Model-Centric Diagnostics: A Framework for Internal State Readouts
- 将训练状态视为潜在变量，整合梯度、置信度等信号为投影
- 不同读出方式反映局部损失曲面几何特性，互补性强
- 适合关注模型诊断与早期停止的从业者
我们提出一种以模型为中心的诊断框架，将训练状态视为潜变量，并将一系列内部读出信号——头梯度范数、置信度、熵、预测间隔及相关信号——统一为该状态的锚点相对投影。初步版本曾引入头梯度探针用于检查点选择，本版聚焦统一视角与结构诊断；完整算法细节、理论分析及实验验证将见于后续论文。核心思想是：任意预测头均诱导出局部损失曲面，其几何特征（梯度大小、曲率、尖锐度）反映上游特征与任务的对齐程度。不同读出方式（梯度范数、Softmax熵、预测间隔）对应该几何的不同投影，各有优势。框架表明，检查点选择、早停和轻量架构预筛选均可视作通过不同视角查询同一底层状态。在ImageNet分类与COCO检测/分割上的初步实验展示了实际潜力；严格基准测试与消融实验将延后至完整论文。
原文摘要 · Abstract (English)
We present a model-centric diagnostic framework that treats training state as a latent variable and unifies a family of internal readouts -- head-gradient norms, confidence, entropy, margin, and related signals -- as anchor-relative projections of that state. A preliminary version of this work introduced a head-gradient probe for checkpoint selection. In this version, we focus on the unifying perspective and structural diagnostics; full algorithmic details, theoretical analysis, and experimental validation will appear in a forthcoming paper. We outline the conceptual scaffold: any prediction head induces a local loss landscape whose geometry (gradient magnitude, curvature, sharpness) reflects how well the upstream features are aligned with the task. Different readout choices -- gradient norms, softmax entropy, predictive margin -- correspond to different projections of this geometry, each with complementary strengths. The framework suggests that checkpoint selection, early stopping, and lightweight architecture pre-screening can all be viewed as querying the same underlying state through different lenses. Illustrative experiments on ImageNet classification and COCO detection/segmentation hint at the practical potential; rigorous benchmarks and ablations are deferred to the full paper.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。