aiqbench analysis
This research introduces UPTURN, a novel framework that improves LLM reasoning accuracy by learning optimal intervention points to redirect model outputs via new user turns rather than fine-tuning or inline injections. The approach demonstrates superior performance across multiple math benchmarks and reasoners while maintaining efficiency without requiring model retraining.
Inject a Turn, Not a Token: Learning Where to Redirect LLM Reasoning for Self-Correction
arXiv:2610.13672v1 Announce Type: new Abstract: Large language models struggle to reliably identify and correct errors in their own reasoning. Unlike prior training-based approaches that fine-tune an LLM to generate revisions or feedback, we learn where to intervene in failed rollouts and determine how to redirect them. To choose where to intervene, we train a 130K-parameter MLP head on a frozen encoder using only final-answer correctness. For redirection, we find that under a fixed generation budget, the tested inline injections fail to reliably redirect reasoning, whereas presenting the same instruction in a new user turn changes the model's course. The resulting correction framework, UPTURN (User-Prompt TURN injection at a learned position), requires no LLM fine-tuning: it retains a selected prefix, injects an instruction as a new user turn, and regenerates the suffix. Across four reasoners and six math benchmarks, UPTURN improves average accuracy in every setting over a 32-rollout baseline. It outperforms self-correction and reflection-token injection baselines under matched correction budgets, transfers across reasoners and to GPQA-Diamond without retraining, and complements GRPO and self-correction methods.
Abstract:Large language models struggle to reliably identify and correct errors in their own reasoning. Unlike prior training-based approaches that fine-tune an LLM to generate revisions or feedback, we learn where to intervene in failed rollouts and determine how to redirect them. To choose where to intervene, we train a 130K-parameter MLP head on a frozen encoder using only final-answer correctness. For redirection, we find that under a fixed generation budget, the tested inline injections fail to reliably redirect reasoning, whereas presenting the same instruction in a new user turn changes the model's course. The resulting correction framework, UPTURN (User-Prompt TURN injection at a learned position), requires no LLM fine-tuning: it retains a selected prefix, injects an instruction as a new user turn, and regenerates the suffix. Across four reasoners and six math benchmarks, UPTURN improves average accuracy in every setting over a 32-rollout baseline. It outperforms self-correction and reflection-token injection baselines under matched correction budgets, transfers across reasoners and to GPQA-Diamond without retraining, and complements GRPO and self-correction methods.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2610.13672 [cs.CL]
(or arXiv:2610.13672v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.13672 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 267 words
Related coverage
Implementing and Evaluating a Basic Per-Action Monitor for Safer Evals
METR · Sep 27, 2026
Low-Background Minds
LessWrong · Oct 3, 2026
GitLab 19.4 release notes
GitLab · Sep 17, 2026
It’s safe to close your laptop now: Hosting coding agents on Amazon Bedrock AgentCore
AWS · Jun 8, 2026