readout-feedback

Tag

Cards List
#readout-feedback

Steering Recurrent Reasoners at Inference Time with Readout Feedback

arXiv cs.LG · 2026-08-26 Cached

The paper introduces Readout Feedback (RoFB), a test-time intervention that steers latent dynamics in recurrent reasoning models using their own readout probabilities, achieving performance gains on Sudoku and Maze tasks without retraining.

0 favorites 0 likes
← Back to home

Submit Feedback