Enregistré dans:
| Auteurs principaux: | , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2509.00079 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908510395039744 |
|---|---|
| author | Correa, Andrew G. A. de Matos, Ana C. H |
| author_facet | Correa, Andrew G. A. de Matos, Ana C. H |
| contents | Reasoning models often outperform smaller models but at 3--5$\times$ 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-confidence-token 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_00079 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Entropy-Guided Loop: Achieving Reasoning through Uncertainty-Aware Generation Correa, Andrew G. A. de Matos, Ana C. H Artificial Intelligence Machine Learning Reasoning models often outperform smaller models but at 3--5$\times$ 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-confidence-token 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. |
| title | Entropy-Guided Loop: Achieving Reasoning through Uncertainty-Aware Generation |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2509.00079 |