2026-09-29 · America/Los_Angeles · 论文 · #17

DepthBench: Measuring How Residual Connections Enable More Computational Depth

Depth is a natural way to increase the computational capacity in Transformers, yet the contribution of deeper layers can diminish as depth grows larger. Recent approaches enhance normalization (e.g., LayerNorm Scaling) or residual connections (e.g., mHC, AttnRes) to enable better information flow and depth utilization. However, it remains unclear whether they truly translate increased architectural depth into effective computational depth, and whether their reported gains stem from better access to information across depth, or unaccounted-for confounding factors. In this paper, we introduce…

热度 47.2 / 100；排名与评分保留该期记录。