Saved in:
Bibliographic Details
Main Authors: Quamar, Mohammad Atif, Areeb, Mohammad
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.04654
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909890737340416
author Quamar, Mohammad Atif
Areeb, Mohammad
author_facet Quamar, Mohammad Atif
Areeb, Mohammad
contents Chain-of-Thought (CoT) prompting is a key technique for enabling complex reasoning in large language models. However, generating full, fixed-length rationales is computationally wasteful, inflating both token usage and latency. We introduce LEASH: Logit-Entropy Adaptive Stopping Heuristic, a training-free decoding algorithm that adaptively halts rationale generation. LEASH monitors two intrinsic signals: the slope of token-level entropy and the improvement in the top-logit margin. It terminates the generation once both signals plateau, indicating the model has reached a stable reasoning state. Across four instruction-tuned models on the GSM8K and AQuA-RAT benchmarks, LEASH reduces average token generation by 30--35% and latency by 27%, while incurring a 10 p.p. accuracy drop relative to CoT. LEASH is model-agnostic and requires no additional training or supervision, offering a simple and efficient alternative to CoT decoding.
format Preprint
id arxiv_https___arxiv_org_abs_2511_04654
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Logit-Entropy Adaptive Stopping Heuristic for Efficient Chain-of-Thought Reasoning
Quamar, Mohammad Atif
Areeb, Mohammad
Computation and Language
Chain-of-Thought (CoT) prompting is a key technique for enabling complex reasoning in large language models. However, generating full, fixed-length rationales is computationally wasteful, inflating both token usage and latency. We introduce LEASH: Logit-Entropy Adaptive Stopping Heuristic, a training-free decoding algorithm that adaptively halts rationale generation. LEASH monitors two intrinsic signals: the slope of token-level entropy and the improvement in the top-logit margin. It terminates the generation once both signals plateau, indicating the model has reached a stable reasoning state. Across four instruction-tuned models on the GSM8K and AQuA-RAT benchmarks, LEASH reduces average token generation by 30--35% and latency by 27%, while incurring a 10 p.p. accuracy drop relative to CoT. LEASH is model-agnostic and requires no additional training or supervision, offering a simple and efficient alternative to CoT decoding.
title Logit-Entropy Adaptive Stopping Heuristic for Efficient Chain-of-Thought Reasoning
topic Computation and Language
url https://arxiv.org/abs/2511.04654