Metraipolis › archive › dc_efe9c9dfc02b
Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Models — arXiv:2609.39445
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.39445
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T13:46:12.594Z
- SHA-256 of the archived text
1b507a959030f905b91f39c908e61d7b78a5f87a7b0e06542d5c499351e28ba3- Stable URL of this page
https://metraipolis.com/documento/dc_efe9c9dfc02b
How to cite
Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Models — arXiv:2609.39445 — arXiv (en), https://arxiv.org/abs/2609.39445 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_efe9c9dfc02b (Metraipolis archive dc_efe9c9dfc02b, sha256 1b507a959030f905b91f39c908e61d7b78a5f87a7b0e06542d5c499351e28ba3)
BibTeX
@misc{dcefe9c9dfc02b,
title = {Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Models — arXiv:2609.39445},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.39445},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_efe9c9dfc02b; sha256 1b507a959030f905b91f39c908e61d7b78a5f87a7b0e06542d5c499351e28ba3},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39445. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Time series foundation models (TSFMs) commonly adapt to new data by attaching a single trainable head to a frozen backbone, a one-size-fits-all setup that underfits heterogeneous regimes. Replacing the head with a mixture of experts is the standard upgrade, but on instance-normalized backbones (the dominant TSFM design class) it fails: routing entropy collapses to zero and one expert absorbs every input, a failure we call normalization-induced routing collapse. Standard MoE rescue mechanisms do not repair it, because the cause is in the router's input, not its optimization. Pre-encoder normalization strips the statistics a router would need to tell regimes apart. A mutual-information decomposition makes this precise and yields a signal-ratio that, computed before training, predicts dataset vulnerability (Spearman $\rho = -0.88$). Eight causal controls, including a vision-modality replication, isolate instance normalization as the cause. The prescription is a minimal causal intervention: Raw-Routed Mixture of Adapters (RR-MoA), which routes on the raw, pre-normalization input. Under a strictly frozen backbone, RR-MoA wins 54/54 comparisons against the strongest fixed adapter and significantly outperforms LoRA, TRACE, AdaMix, and full fine-tuning. The effect generalizes across six backbones and an imputation task. Frozen RR-MoA also beats full fine-tuning by 12-79% (the Frozen Paradox); two architecturally distinct variants confirm the principle generalizes beyond this specific router. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.LG). Categorie arXiv: Machine Learning (cs.LG) ; Machine Learning (stat.ML). Pagina del lavoro: https://arxiv.org/abs/2609.39445 — 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 2026-10-07T13:18:10.386Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).