SWE-Prot\'eg\'e: Learning to Selectively Collaborate With an Expert Unlocks Small Language Models as Software Engineering Agents

TL;DR

SWE-Protégé enhances small language models' software engineering capabilities by enabling selective expert collaboration, significantly improving performance on long-horizon tasks with minimal expert guidance.

Contribution

Introduces a post-training framework that trains small language models to selectively seek expert guidance, reducing looping and improving task success in software engineering.

Findings

Achieves 42.4% Pass@1 on SWE-bench Verified, a +25.4% improvement.

Uses expert assistance sparingly (~4 calls per task).

Combines supervised fine-tuning with reinforcement learning to improve decision-making.

Abstract

Small language models (SLMs) offer compelling advantages in cost, latency, and adaptability, but have so far lagged behind larger models on long-horizon software engineering tasks such as SWE-bench, where they suffer from pervasive action looping and low resolution rates. We introduce SWE-Prot\'eg\'e, a post-training framework that reframes software repair as an expert-prot\'eg\'e collaboration problem. In SWE-Prot\'eg\'e, an SLM remains the sole decision-maker while learning to selectively seek guidance from a strong expert model, recognize stalled states, and follow through on expert feedback. Our approach combines supervised fine-tuning on expert-augmented trajectories with agentic reinforcement learning that explicitly discourages degenerative looping and unproductive expert collaboration. We lightly post-train Qwen2.5-Coder-7B-Instruct to achieve 42.4% Pass@1 on SWE-bench Verified, a +25.4% improvement over the prior SLM state of the art, while using expert assistance sparsely (~4 calls per task and 11% of total tokens).