arXiv：cs.LG· Alexander Gill, Guram Gogia, Francisco Guzm\'an·· 4 小时前
Precog-RM：预测 LLM 智能体轨迹结果的学习表示
Divergent Paths: Learning an Outcome-Predictive Representation of LLM Agent Trajectories
AI 导读
研究者提出 Precog-RM，一种基于 transformer 的模型，通过元学习在少样本设定下预测 LLM 智能体轨迹的成功或失败，可泛化到未见任务和模型。在三个智能体基准和三个策略模型上，简单判别器预测结果的 AUROC 最高达 0.95。该模型作为 GRPO 在 Tau2-Bench 上的早停控制器，训练 FLOPs 减少最多 47.3%，性能与基线相当。
正文
Abstract:In this work, we reframe agentic trajectories as paths through a shared embedding space of actions and environmental observations. We show that successful and failed rollouts of the same task separate in this space, with simple per-task discriminators predicting held-out rollout outcomes at up to 0.95 AUROC across three agent benchmarks and three policy models. To exploit this signal across tasks, we introduce Precog-RM, a transformer-based model that meta-learns a representation in which outcome prediction can be performed in a few-shot setting. We show that Precog-RM generalizes to unseen tasks and models, and gives confident predictions of trajectory success long before termination. Deployed as an early stopping controller for Group Relative Policy Optimization (GRPO) on Tau2-Bench, Precog-RM cuts training FLOPs by up to 47.3\% while maintaining performance comparable to baseline GRPO.
| Comments: | 9 pages, 8 pages appendix |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.19614 [cs.LG] |
| (or arXiv:2610.19614v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.19614 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Alexander Gill [view email]
[v1]
Wed, 14 Oct 2026 05:36:22 UTC (377 KB)
来源：arXiv：cs.LG · arxiv.org