1.8 KiB
1.8 KiB
title, created, type, arxiv, authors, venue, date
| title | created | type | arxiv | authors | venue | date |
|---|---|---|---|---|---|---|
| The Verification Horizon: No Silver Bullet for Coding Agent Rewards | 2026-07-02 | raw | 2606.26300 | Binghai Wang, Chenlong Zhang, Dayiheng Liu, Jiajun Zhang, Jiawei Chen, Mingze Li, Mouxiang Chen, Rongyao Fang, Siyuan Zhang, Xuwu Wang, Yuheng Jing, Zeyao Ma, Zeyu Cui (Qwen Team, Alibaba) | arXiv | 2026-06-24 |
Raw: The Verification Horizon
A classical intuition holds that verifying a solution is easier than producing one. For today's coding agents, this intuition is being inverted: generating complex candidate solutions is no longer difficult — reliably verifying them has become the harder problem.
Core Framework
Every verifier is a proxy for human intent, never the intent itself. Verification quality characterized along three dimensions:
- Scalability: can the signal be produced cheaply at training scale?
- Faithfulness: how much of true user intent does the signal reflect?
- Robustness: can the verifier hold across diverse inputs and optimization pressure?
Achieving all three simultaneously is the central challenge.
Four Reward Constructions
- Test Verifier (§2): execution-based test suites + agentic quality judge + behavior monitoring. Hacked resolved rate: 28.57% → 0.56%; Clean resolved: 40.22% → 60.53%.
- Interactive Judge (§3): rubric-based static judge → agentic interactive judge with Playwright. Resists length exploitation.
- User Feedback Verifier (§4): extract HIRS from interaction data → Span-KTO training. +13.3pp on Aone-bench.
- Automated Agent Verifier (§5): autonomous evaluator for long-horizon tasks. Evaluator quality is metric-dependent.
Core Claim
No fixed reward function can remain effective as policy capability continues to grow; verification must co-evolve with the generator.
Source: https://arxiv.org/abs/2606.26300