Route-Induced Density and Stability (RIDE): Controlled Intervention and Mechanism Analysis of Routing-Style Meta Prompts on LLM Internal States

Dianxing Zhang, Gang Li, Sheng Li · 2026

Read Route-Induced Density and Stability (RIDE): Controlled Intervention and Mechanism Analysis of Routing-Style Meta Prompts on LLM Internal States by Dianxing Zhang, Gang Li, Sheng Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-29206.txt.

Description

Zhang and co-authors test whether routing prompts make language models sparser inside. A common belief links routing to expert-like, stable outputs. They inject routing-style meta prompts before frozen instruction-tuned models. Then they measure activation sparsity, keyword attention, and output stability. Across three models, the prompts made early layers denser, not sparser. Plain expert instructions often beat structured tags. The link between density and stability was weak overall. The probe helps calibrate routing design choices.

Catalog record

CREATORDianxing Zhang, Gang Li, Sheng Li
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE68392
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.29206

Browse the full library · About the Tower of Babel