arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2610.21055cs.AIcs.CLcs.LGcs.MA
收藏

Athanor:面向自主后训练研究的智能体原生系统

Athanor: An Agent-Native System for Autonomous Post-Training Research

Duke University(杜克大学);The University of Chicago(芝加哥大学)

Yufa Zhou, Yichen Wang, Jiakun Fan, Ziqi Shu, Zhichao Liu, Anru R. Zhang, Neiwen Ling, Zijun Wei

AI总结:

本文提出Athanor系统,首次证明Claude Opus 5.5智能体团队可自主后训练Qwen3-4B-Base,使其性能接近或超过官方指令微调版本,为自主AI研究提供基础。

摘要:

递归自我改进(即AI系统构建更强的AI系统)是实现超级智能的一条有前景的路径。本文研究了一种当前可衡量的较弱形式:前沿智能体利用公开数据对小型原始基础模型进行后训练,以此作为其研究潜力的测试。我们构建了Athanor,这是一个面向真实GPU的自主后训练研究(AutoPostTrain)的开源智能体原生系统。我们为智能体提供完全的研究自主权、低摩擦的工程工具和高效的入门代码。智能体自行开展研究:它们读取数据集和评估结果、构建训练数据、编写和修改训练算法、调整超参数,并分析结果以决定下一步尝试的方向。它们在沙箱中工作,无法使用外部教师模型。Athanor的服务提供冻结评估、托管作业、去污染数据、持久研究记录以及智能体未知的保留测试集。在一场为期104小时的实验中,由四个Claude Opus 5.5智能体组成的团队(每个智能体配备8个GPU节点且未获得任何研究建议),在包含6个任务的开发套件上将Qwen3-4B-Base的性能从15.9提升至67.3,在保留测试集上从9.2提升至48.2,而官方Qwen3-4B-Instruct-2507的对应性能分别为66.5和47.9。据我们所知,这是首次证明前沿智能体团队可自主将原始基础模型后训练为跨领域通用模型,其平均性能与官方指令微调版本相当或更优。该实验的完整记录显示,智能体通过组合已知方法和彼此的模型达到这一水平:在公开推理数据上的监督微调(SFT)、采用某一智能体引入的rollout变化的强化学习(RL)、跨智能体的权重合并以及多轮自蒸馏。随着执行环节的解决,当前的智能体已具备完成该任务的工程和研究能力。我们希望Athanor能为扩大自主AI研究提供基础。

英文摘要:

Recursive self-improvement, in which AI systems build stronger AI systems, is one promising route to superintelligence. We study a weaker form that can be measured today: frontier agents post-training a small raw base model with public data, as a test of their research potential. We build Athanor, an open agent-native system for autonomous post-training research (AutoPostTrain) on real GPUs. We give the agents full research autonomy, low-friction engineering tools, and efficient starter code. The agents carry out the research themselves: they read datasets and evaluation results, build their training data, write and modify training algorithms, tune hyperparameters, and analyze the results to decide what to try next. They work in a sandbox and cannot use external teacher models. Athanor's services supply frozen evaluation, managed jobs, decontaminated data, a persistent research record, and a held-out suite the agents are unaware of. In a 104-hour campaign, a team of four Claude Opus 5.5 agents, each on an eight-GPU node and given no research advice, improves Qwen3-4B-Base from 15.9 to 67.3 on a six-task development suite and from 9.2 to 48.2 on a held-out suite, against 66.5 and 47.9 for the official Qwen3-4B-Instruct-2507. To our knowledge, this is the first demonstration that a team of frontier agents can autonomously post-train a raw base model into a cross-domain generalist model that matches or surpasses the official instruct version on average. The campaign's full record shows that the agents reach this level by recombining known methods and each other's models: SFT on public reasoning data, RL with a rollout change one agent introduced, weight merges across agents, and rounds of self-distillation. With execution handled, today's agents already have the engineering and research ability to conquer this task. We hope Athanor provides a foundation for scaling autonomous AI research.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑