One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents
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 instance-level progress, motivating explicit consolidation of successful behavior and policy-adaptive t...
| Authors | Jie Zhao, Ziyu Jiang, Suhang Zheng et al. |
|---|---|
| Primary category | cs.SE |
| Published | 20 September 2026 |
| Last updated at source | 20 September 2026 |
Connected in the knowledge graph
- Logics-MLLM/Logics-SWE-Env-2.5KDataset · cited by
- Logics-MLLM/Logics-SWE-Qwen3.6-27BModel · cited by
Models and datasets referencing this paper
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.