首页 > AI前沿 > Beyond the Previous Layer: Residual Predictive Structure in Sparse MoE Routing

Beyond the Previous Layer: Residual Predictive Structure in Sparse MoE Routing

arXiv机器学习 2026-09-16 07:57 2 阅读 查看原文

Sparse mixture-of-experts models route each token through a sequence of expert selections.

We ask whether the immediately preceding selection adequately summarizes this trajectory for predicting the next router.

Using frozen OLMoE and JetMoE models, we measure the held-out predictive gain from earlier expert selections while retaining the most recent selection as a common baseline.

OLMoE Results

In OLMoE, extending the history from one to eleven layers raises router-logit $R^2$ from 0.59879 to 0.66544.

JetMoE Replication

A preregistered JetMoE replication yields four-layer gains of 0.14275 and 0.20528 at two target depths, with paired bootstrap intervals above zero.

These gains survive nonlinear decoding: adding history to a small multilayer perceptron improves $R^2$ by 0.17137 and 0.21861, whereas nonlinear decoding of the recent state alone adds 0.00139 and 0.00936 over a linear probe.

Parameter-matched Controls

Parameter-matched controls preserve the advantage.

Cross-fitted History Residuals

Cross-fitted history residuals predict target residuals with $R^2$ of 0.20549 and 0.23556.

These findings identify residual predictive structure in expert-selection trajectories beyond adjacent-layer persistence.