Year: 2025
Type: preprint
Abstract: Reasoning models often outperform smaller models but at 3–5°ø higher cost and added latency. We present entropy-guided refinement: a lightweight, test-time loop that uses token-level uncertainty to trigger a single, targeted refinement pass. We extract logprobs, compute Shannon entropy on top-k alternatives, and apply a simple OR-logic trigger over perplexity, maximum token entropy, and low-confidencetoken count. Unlike approaches that use entropy only for measurement or decoding, we pass a compact uncertainty report (tokens, confidences, alternatives, context) back to the model to guide corrective edits. On representative technical queries across reasoning, mathematics, and code generation tasks, a small model with our loop approaches 95% of a reference reasoning model’s quality at approximately one-third of the cost. The method achieves selective refinement on 31% of responses while improving accuracy by 16 percentage points over single-pass inference. We demonstrate that this uncertainty-aware loop provides an effective middle ground between single-pass inference and expensive reasoning chains, making it practical for production deployments where both quality and cost matter.
Source: arXiv (Cornell University)
Authors: Araujo Correa, Andrew Gabriel, Hermogenes de Matos, Ana Carolina
Institution: Pontifícia Universidade Católica do Rio de Janeiro
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4416681964
Cites:
Cited by:
Related to: