arXiv AI

VACE: Validation-Gated Alternating Co-Evolution of Agent Models and Harnesses

arXiv Computation and Language
Oct 8

CoTrace: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution

arXiv:2610.10426v1 Announce Type: new Abstract: Terminal-agent capability depends jointly on model weights and the runtime harness that formats prompts, binds tools, and handles error recovery. Exist...

By Jixuan Chen, Jiaxin Zhang, Qinyuan Ye, Yada Pruksachatkun, Haoxiang Zhang, Jingming Zhuo, Yifan Zhang, Yutong Dai, Juntao Tan, Xiangyu Peng, Silvio Savarese, Zeyuan Chen, Lianhui Qin, Chien-Sheng Wu
arXiv AI
Sep 2

WHALE: A Simple Recipe for Joint Harness-Weight Optimization

The paper introduces WHALE, a method that alternates between updating a language model’s weights and searching for a better harness (the code that manages context and control flow). By iteratively fine‑tuning the model under the current harness and then optimizing the harness under the updated model, WHALE improves performance across search QA, math reasoning, and chess puzzles, outperforming weight‑only, harness‑only, and Fast‑Slow Training by 4.15–24.38 percentage points in mean@8 accuracy. The approach uses either fixed phase lengths or an adaptive patience rule to decide when to switch phases, and the authors provide code on GitHub.

By Haechan Kim, Yoonho Lee, Gisang Lee, Chelsea Finn, Kangwook Lee
arXiv AI
Oct 12

Multi-Harness Reinforcement Learning for Coding Agents: Training Gains and Harness Dependence

The study evaluates multi-harness reinforcement learning (RL) for coding agents by training two Qwen3-8B RL variants—one with rewards normalized within each harness and one across harnesses—using identical replay data. Results show modest gains over the supervised fine-tuning baseline (+0.48 and +0.77 percentage points for within and cross normalization, respectively), but no consistent advantage of cross normalization on unseen harnesses. Performance differences across harnesses are much larger than the training gains, indicating that harness dependence remains a significant challenge.

By Chenqian Le, Jiayi Cheng, Qijia He, Runhao Li, Yinghao Li, Xupeng Chen
arXiv Computation and Language
Oct 12

Evo$^2$: Evolving the Evolver through Weight-Harness Co-Evolution

Evo$^2$ is a framework that jointly evolves a language model’s weights and its harness—the set of prompts, tools, and feedback mechanisms—using a dual-role pipeline where a single policy acts as both a Proposer and a Reflector. The training pipeline EvoTIDE combines supervised fine-tuning with interleaved GRPO to refine both roles based on harness edits and execution outcomes. Experiments on coding benchmarks show Evo$^2$ outperforms existing self‑evolution methods, achieving higher scores on Polyglot and SWE‑bench Verified, and its evolved harnesses transfer across tasks and models.

By Chejian Xu, Shujian Zhang, Wenxuan Zhou, John Lambert, Beck Chen, Bo Li, Andrew Hard, Rajiv Mathews
arXiv AI
Oct 23

Forging More Tasks into Better Harnesses: Agentic Harness Evolution via Successive Scale Racing

The paper introduces Successive Scale Racing (SSR), a framework that improves agentic harness evolution by adaptively selecting optimal evidence combinations during each evolution round. SSR constructs diverse evidence sets based on task quality and coverage, then efficiently races candidate harnesses under a limited compute budget. Across three benchmarks, SSR outperforms baselines, steadily raising pass rates as the task pool expands from 40 to 200 while cutting validation executions by 60%.

By Jingzhe Liu, Jiliang Tang, Tong Zhao, Neil Shah, Clark Mingxuan Ju
arXiv AI
Oct 5

VERSE: Verified Self-Evolving Optimizer for Agent Harnesses

The paper introduces VERSE, a Verified Self‑Evolving optimizer that enhances LLM agent harnesses by allowing the optimizer to test edits, replay failures, and perturb steps while tracking fixes and regressions. VERSE builds its own tools for failure analysis, verification, training audits, and workflow control, and uses this feedback to revise the harness’s prompts, skills, tools, hooks, and notes without changing model weights. In experiments across five executors and multiple languages, VERSE improves all evaluated harness optimizers, achieving higher accuracy on held‑out and out‑of‑distribution tasks compared to the strongest baselines.

By Zekai Wang, Yingqiang Ge, Zekun Wang, Hai Wang, Yuhui Xu, Joshua Frandsen, Shancong Fu, Ashia C. Wilson, Chandan K. Reddy
arXiv AI
3d ago

Verify by Doing: Co-Evolving Agent Execution and Evaluation without External Feedback

The paper introduces Verify by Doing (VbD), a method that co‑evolves an Actor and an Evaluator within a single task while keeping model weights fixed. The Evaluator actively inspects artifacts and environment state using tools, generating an internal signal that guides both output selection and updates to execution and verification procedures. Experiments on AutomationBench, τ³ Banking, and GDPval show VbD outperforms other methods that lack external feedback, with gains of 3.3, 4.1, and 2.3 percentage points on the main backbone, and ablations confirm the importance of active verification and Evaluator evolution.

By Yitong Lin, Jianda Chen, Yihao Hu, Pan Wang, Hang Wang, Zhihao Wen, Xiaoxue Ren, Wei Wu