EGSS: Entropy-guided Stepwise Scaling for Reliable Software Engineering

2026-02-05

AI Summary

This work addresses the limitations of Agentic Test-Time Scaling in software engineering tasks, which suffers from high computational overhead and unreliable candidate selection. To overcome these challenges, the authors propose Entropy-Guided Stepwise Scaling (EGSS), a novel framework that leverages entropy signals to dynamically guide the search process. By integrating test suite augmentation with dynamic candidate pruning, EGSS achieves a synergistic optimization of both efficiency and effectiveness. Evaluated on SWE-Bench-Verified, the method consistently improves pass rates by 5–10% across all models—reaching 74.6% for GLM-4.6—while reducing inference token consumption by over 28%.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Planning with Language Models
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Agentic Test-Time Scaling (TTS) has delivered state-of-the-art (SOTA) performance on complex software engineering tasks such as code generation and bug fixing. However, its practical adoption remains limited due to significant computational overhead, primarily driven by two key challenges: (1) the high cost associated with deploying excessively large ensembles, and (2) the lack of a reliable mechanism for selecting the optimal candidate solution, ultimately constraining the performance gains that can be realized. To address these challenges, we propose Entropy-Guided Stepwise Scaling (EGSS), a novel TTS framework that dynamically balances efficiency and effectiveness through entropy-guided adaptive search and robust test-suite augmentation. Extensive experiments on SWE-Bench-Verified demonstrate that EGSS consistently boosts performance by 5-10% across all evaluated models. Specifically, it increases the resolved ratio of Kimi-K2-Intruct from 63.2% to 72.2%, and GLM-4.6 from 65.8% to 74.6%. Furthermore, when paired with GLM-4.6, EGSS achieves a new state-of-the-art among open-source large language models. In addition to these accuracy improvements, EGSS reduces inference-time token usage by over 28% compared to existing TTS methods, achieving simultaneous gains in both effectiveness and computational efficiency.

Problem

• Test-Time Scaling
• Computational Overhead
• Candidate Selection
• Software Engineering
• Large Language Models

Innovation

• Entropy-guided scaling
• Test-Time Scaling
• Software engineering
• Efficiency-effectiveness tradeoff
• LLM code generation