analiza aiqbench
TIDE addresses a critical bottleneck in training multi-turn agentic models by dynamically balancing teacher distillation and reinforcement learning signals to prevent premature convergence or policy drift. By adapting the weight of guidance based on student-teacher disagreement trends, this method enables smaller models to explore more effectively while leveraging larger teachers without being constrained by them. This advancement represents a significant step forward in optimizing the training efficiency and capability scaling of autonomous AI agents.
TIDE: Teacher-Student Transition via Informative Distillation and Exploration for Agentic RL
arXiv:2609.35058v1 Announce Type: new Abstract: Effective multi-turn agents require interaction strategies that coordinate information gathering, actions, and feedback over long horizons. GRPO is a reinforcement learning algorithm used to train these agents, but sparse trajectory-level rewards limit early exploration in small models. Recent methods augment RL with on-policy distillation (OPD) from a stronger teacher. However, a fixed mixture assumes that teacher guidance and reward optimization should retain a constant relative role throughout training and across interaction turns. This assumption can fail at two scales. Globally, as training progresses, maintaining strong distillation pressure can constrain the model from moving beyond the teacher's capabilities. Locally, teacher--student disagreement identifies where the student departs from the teacher, but cannot tell whether that departure is exploration supported by better outcomes or low-quality policy drift. Our methodological insight is that teacher guidance and reward optimization should be dynamically rebalanced over training and jointly allocated across turns. We instantiate this insight in \tide. Globally, \tide uses the measured disagreement trend as a practical schedule signal, advancing an OPD-to-RL handoff when discrepancy reduction becomes slow but remains positive and progressively increasing the relative weight of RL. Locally, \tide jointly modulates teacher-guided and reward-driven updates: relative action value and disagreement prioritize the OPD signal, whereas relative action value supplies the RL advantage and normalized disagreement reweights it across turns. Coupled with the global handoff, \tide allocates stronger teacher guidance early and gives reward-driven updates greater relative weight later in training. Experiments across multiple benchmarks, student scales, and controlled ablations support the effectiveness of TIDE's adaptive OPD--RL coordination.
Abstract:Effective multi-turn agents require interaction strategies that coordinate information gathering, actions, and feedback over long horizons. GRPO is a reinforcement learning algorithm used to train these agents, but sparse trajectory-level rewards limit early exploration in small models. Recent methods augment RL with on-policy distillation (OPD) from a stronger teacher. However, a fixed mixture assumes that teacher guidance and reward optimization should retain a constant relative role throughout training and across interaction turns. This assumption can fail at two scales. Globally, as training progresses, maintaining strong distillation pressure can constrain the model from moving beyond the teacher's capabilities. Locally, teacher--student disagreement identifies where the student departs from the teacher, but cannot tell whether that departure is exploration supported by better outcomes or low-quality policy drift. Our methodological insight is that teacher guidance and reward optimization should be dynamically rebalanced over training and jointly allocated across turns. We instantiate this insight in \tide. Globally, \tide uses the measured disagreement trend as a practical schedule signal, advancing an OPD-to-RL handoff when discrepancy reduction becomes slow but remains positive and progressively increasing the relative weight of RL. Locally, \tide jointly modulates teacher-guided and reward-driven updates: relative action value and disagreement prioritize the OPD signal, whereas relative action value supplies the RL advantage and normalized disagreement reweights it across turns. Coupled with the global handoff, \tide allocates stronger teacher guidance early and gives reward-driven updates greater relative weight later in training. Experiments across multiple benchmarks, student scales, and controlled ablations support the effectiveness of TIDE's adaptive OPD--RL coordination.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2609.35058 [cs.LG]
(or arXiv:2609.35058v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.35058 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 350 cuvinte
Subiecte conexe
Frontier post-training recipe review with Finbarr Timbers
Interconnects · 16 iun. 2026
Triton-Based Optimization of Video Sparse Attention on ROCm
AMD · 13 iul. 2026
Projection sampling preserves capabilities but diminishes gains from alignment fine-tuning
LessWrong · 16 oct. 2026
Debating RSI, the US-China Gap, and Jaggedness with JS Denain of Epoch AI
Interconnects · 22 sept. 2026