Re-FORC: Adaptive Reward Prediction for Efficient Chain-of-Thought Reasoning
Renos Zabounidis, Aditya Golatkar, Michael Kleinman, Alessandro Achille, Wei Xia, Stefano Soatto
Abstract
We propose Re-FORC, an adaptive reward prediction method that, given a context, enables prediction of the expected future rewards as a function of the number of future thinking tokens. Re-FORC trains a lightweight adapter on reasoning models, demonstrating improved prediction with longer reasoning and larger models. Re-FORC enables: 1) early stopping of unpromising reasoning chains, reducing compute by 26\% while maintaining accuracy, 2) optimized model and thinking length selection that achieves 4\% higher accuracy at equal compute and 55\% less compute at equal accuracy compared to the largest model, 3) adaptive test-time scaling, which increases accuracy by 11\% in high compute regime, and 7\% in low compute regime. Re-FORC allows dynamic reasoning with length control via cost-per-token thresholds while estimating computation time upfront.
BibTeX
@inproceedings{
zabounidis2026reforc,
title={Re-{FORC}: Adaptive Reward Prediction for Efficient Chain-of-Thought Reasoning},
author={Renos Zabounidis and Aditya Golatkar and Michael Kleinman and Alessandro Achille and Wei Xia and Stefano Soatto},
booktitle={Forty-third International Conference on Machine Learning},
year={2026},
url={https://openreview.net/forum?id=8NNPSNAvTX}
}