Metraipolis › archive › dc_a6ef4fc0f6d7
The Residual Stream's Effective Depth — arXiv:2609.31098
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.31098
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T02:13:07.205Z
- SHA-256 of the archived text
6c183f6d0756a26d11699ef4f6226f1b9b3d0c2f3ff2087bc883eb602e440d17- Stable URL of this page
https://metraipolis.com/documento/dc_a6ef4fc0f6d7
How to cite
The Residual Stream's Effective Depth — arXiv:2609.31098 — arXiv (en), https://arxiv.org/abs/2609.31098 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a6ef4fc0f6d7 (Metraipolis archive dc_a6ef4fc0f6d7, sha256 6c183f6d0756a26d11699ef4f6226f1b9b3d0c2f3ff2087bc883eb602e440d17)
BibTeX
@misc{dca6ef4fc0f6d7,
title = {The Residual Stream's Effective Depth — arXiv:2609.31098},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.31098},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a6ef4fc0f6d7; sha256 6c183f6d0756a26d11699ef4f6226f1b9b3d0c2f3ff2087bc883eb602e440d17},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.31098. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We introduce \emph{effective depth} ($\Deff$), a scalar diagnostic that treats the layer-wise residual stream of a transformer as a discrete-time process, measures how representation similarity decays with layer distance, and aggregates that profile into one number. Across sixteen decoder-only language models, $\Deff$ separates a structural consequence of residual accumulation from an empirical one: even maximally diverse orthogonal updates have the closed-form reference $F_L = 2L/(L+1)<2$, yet fifteen of sixteen default measurements lie below $F_L$ (Qwen3.5: 32--44\%, OLMo-2: 40--41\%, Pythia: 23--28\%). Matched references show that the gap is not caused by the persistent initial state or update-size imbalance, but is largely a calibrated signature of correlated residual updates rather than evidence that depth is unused. Symmetric position-0, token-normalisation, and top-PC controls show the regime is not reducible to BOS or top-PC artefacts: the lone above-reference default outlier joins the same regime, and all sixteen models are sub-reference after token-normalisation or top-1-PC removal. Intermediate checkpoints show that the regime is established early in OLMo-2 and stable through 5T tokens, while Pythia-1.4B follows a distinct decreasing trajectory. A controlled residual-carry intervention supports the mechanism, and $\Deff$ is best read as a \emph{global} accumulated-state diagnostic, not as a capability score or pruning method.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.31098 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T01:43:06.728Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).