HuggingFace Daily Papers·· 1 天前
Tev：用稠密进度奖励扩展长时长智能体训练环境
Tev: Scaling Long-Horizon Agentic Environments with Dense Progress Rewards
AI 导读
论文提出 Tev-Gym，首个带可验证稠密渐进奖励的长时长智能体训练数据集，覆盖 7 个领域共 1,170 个任务环境，通过 latent-world-first 方法逐级渲染出复杂度可控的环境。
正文
Abstract:Long-horizon agents have drawn growing attention for their ability to work autonomously -- from hours to days -- on economically valuable tasks, a capability essential to automated discovery and recursive self-improvement (RSI). Yet open-source models still struggle on long-horizon tasks, largely for lack of both training data and a working recipe, and scaling both is hard: long, complex tasks are difficult to curate and verify, and their rewards are sparse. We address both. First, we introduce Tev-Gym, the first long-horizon agent training dataset with verifiable, dense, progressive rewards -- 1,170 verified task environments across 7 domains. It is built on a latent-world-first approach that first constructs a latent world: a complete and correct hidden specification of the task domain, workflow, and expected state after each step. Each task is then progressively rendered from this specification into an agent-facing environment, enabling controllable complexity and a natural progress metric. Second, we turn this metric into a training recipe: via progress-potential reward shaping (PPRS), we convert progress into a dense per-step reward that improves training efficiency and makes otherwise-intractable tasks trainable. Training with this recipe yields Tev-27B, which achieves substantial gains on the Agents' Last Exam computing and math sciences domains (ALE-CM), matching GPT-6 Astra and Claude Opus 5 at Pass@5. Despite training only on ALE-related data, it generalizes to the Terminal-Bench (TB) 3.0 and 4.0 subsets, surpassing GLM 5.2 and Gemini 3.7 Flash, and reaches Claude Opus 4.8-level performance on middle-horizon tasks such as TB 2.1.
| Comments: | this https URL |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.18868 [cs.CL] |
| (or arXiv:2610.18868v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.18868 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Wenbo Hu [view email]
[v1]
Tue, 13 Oct 2026 17:59:51 UTC (1,018 KB)
来源：HuggingFace Daily Papers · arxiv.org