SOURCE-LINKED INTELLIGENCE

Steering Recurrent Reasoners at Inference Time with Readout Feedback

arXiv · AI, language, vision and robotics · article · Aug 25, 2026 · UTC

Recurrent models, which repeatedly update latent states with shared computation blocks, have emerged as powerful architectures for solving complex reasoning tasks. Existing inference-time methods scale computation by running more steps or sampling more trajectories, but ignore information revealed within each trajectory. Here we show that recurrent models can be improved at inference time by using their own readout probabilities to steer latent dynamics without retraining. We introduce Readout Feedback (RoFB), a test-time intervention that converts intermediate predictions into token-wise pair

Read original source ↗ Open in workspace

recordType

paper

region

Global