Certainty-Guided Reasoning in Large Language Models: A Dynamic Thinking Budget Approach

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nogueira, João Paulo, Sun, Wentao, Silva, Alonso, Zumot, Laith
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912889342787584
author Nogueira, João Paulo
Sun, Wentao
Silva, Alonso
Zumot, Laith
author_facet Nogueira, João Paulo
Sun, Wentao
Silva, Alonso
Zumot, Laith
contents Large reasoning language models are typically run with fixed inference budgets, which can waste computation or terminate reasoning prematurely. We introduce Certainty-Guided Reasoning (CGR), a model-agnostic adaptive inference procedure that periodically probes whether the current reasoning supports a confident final answer and terminates early once a target certainty threshold is reached, otherwise continuing until the end-of-thinking token or the budget limit. Certainty is estimated from the model's predicted probabilities over the answer tokens, yielding a lightweight stopping criterion. On AIME2025, CGR preserves baseline accuracy while reducing token usage, providing a tunable certainty-efficiency trade-off that can eliminate millions of tokens in aggregate. Across 64 random seeds, CGR exhibits consistent behavior. We also introduce a Grade metric that penalizes incorrect answers and permits abstention, capturing risk-sensitive performance. Results show that CGR improves Grade by abstaining when certainty remains low.
format Preprint
id arxiv_https___arxiv_org_abs_2509_07820
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Certainty-Guided Reasoning in Large Language Models: A Dynamic Thinking Budget Approach
Nogueira, João Paulo
Sun, Wentao
Silva, Alonso
Zumot, Laith
Artificial Intelligence
Computation and Language
Large reasoning language models are typically run with fixed inference budgets, which can waste computation or terminate reasoning prematurely. We introduce Certainty-Guided Reasoning (CGR), a model-agnostic adaptive inference procedure that periodically probes whether the current reasoning supports a confident final answer and terminates early once a target certainty threshold is reached, otherwise continuing until the end-of-thinking token or the budget limit. Certainty is estimated from the model's predicted probabilities over the answer tokens, yielding a lightweight stopping criterion. On AIME2025, CGR preserves baseline accuracy while reducing token usage, providing a tunable certainty-efficiency trade-off that can eliminate millions of tokens in aggregate. Across 64 random seeds, CGR exhibits consistent behavior. We also introduce a Grade metric that penalizes incorrect answers and permits abstention, capturing risk-sensitive performance. Results show that CGR improves Grade by abstaining when certainty remains low.
title Certainty-Guided Reasoning in Large Language Models: A Dynamic Thinking Budget Approach
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2509.07820