AIIC AI Intelligence Centre

SOURCE-LINKED INTELLIGENCE

One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents

arXiv · AI, language, vision and robotics · article · Sep 20, 2026 · UTC

Repository-level software engineering (SWE) comprises heterogeneous task categories, whose progress under pooled agentic reinforcement learning can be uneven: gains in some categories coincide with regressions in others, while aggregate resolution obscures these changes. Motivated by this category see-saw, we develop a category-aware expert-training and policy-integration framework. Executable task construction and SWE Labeler, an evidence-grounded multi-axis labeling system, organize the training pools. Initial category-specific RL improves average training success while leaving uneven instan

Read original source ↗ Open in workspace

recordType
paper
region
Global

Evidence & attribution

First collected: 2026-09-23T10:01:48.231Z. This is not the publication date.