A Few Teacher Steps Go a Long Way: Cost-Efficient On-Policy Data Augmentation for Agent Post-Training

July 06, 2026 ยท Grace Period ยท ๐Ÿ› ICML 2026 Workshop

โณ Grace Period
This paper is less than 90 days old. We give authors time to release their code before passing judgment.
Authors Junze Ye, Jiayi Cheng, Miao Lu, Michal Mankowski, Jose Blanchet, Mohsen Bayati arXiv ID 2607.04574 Category cs.LG: Machine Learning Cross-listed cs.AI Citations 0 Venue ICML 2026 Workshop
Abstract
For LLM agents, supervised fine-tuning is not only about teacher labels' quality, but also about which interaction contexts those labels condition on. Pure behavioral cloning uses full teacher demonstrations, creating a mismatch between teacher-induced contexts seen in training and student-induced contexts encountered at test time. Recent work addresses this mismatch by querying a teacher at contexts reached by the student, often with increasingly elaborate filtering of the teacher's continuations. We instead frame on-policy data construction as a budget-allocation problem: under matched supervision resources, should teacher output be spent on more start-to-finish demos, longer continuations, outcome filtering, or broader coverage of learner-induced contexts? We formalize this design space through the rollout policy, switch-time distribution, continuation horizon, filtering rules, and two complementary costs: teacher inference generated before filtering and teacher supervision retained for SFT. Across HotpotQA, ALFWorld, and Terminal-Bench-Dev, bounded unfiltered teacher continuations at learner-induced contexts improve over pure behavioral cloning at matched budgets. On HotpotQA and ALFWorld, where we run the full comparison, few-step continuations match or exceed success-filtered and critical-context-filtered alternatives. Our findings suggest that a few teacher steps, placed at learner-induced contexts, can be a more cost-efficient supervision allocation than longer or more heavily curated teacher completions.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Machine Learning