Act First, Reason Later: Accelerating On-Policy Distillation for Multi-Turn Agents via Reference-Conditioned Inverse Dynamics

arXiv:2609.36608 · cs.LG, cs.CL · Submitted 2026-09-29 · Read on arXiv

cs.LG, cs.CL

Submitted: 2026-09-29

Updated: 2026-09-29

Code: https://github.com/alfworld/alfworld

Terminology

• Distillation — 47× in this paper · explained in 2 episodes
• Benchmark — 41× in this paper · explained in 2 episodes
• On-Policy Distillation — 19× in this paper · explained in 3 episodes
• Alignment — 14× in this paper · explained in 1 episode
• Ablation — 6× in this paper · explained in 1 episode
• Consistency — 5× in this paper · explained in 3 episodes
• Adam — 3× in this paper · explained in 2 episodes
• Exposure Bias — 3× in this paper · explained in 2 episodes
• In International Conference on Learning Representations — 3× in this paper
• Reinforcement Learning — 3× in this paper · explained in 4 episodes
• Verification — 3× in this paper · explained in 2 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Thinking mode — 2× in this paper · explained in 2 episodes
• Tokenization — 2× in this paper · explained in 2 episodes
• Ablation Studies — 1× in this paper · explained in 1 episode
• Agent Harness — 1× in this paper · explained in 2 episodes
• Aggregation — 1× in this paper · explained in 1 episode
• Credit assignment — 1× in this paper · explained in 3 episodes
• Distribution Shift — 1× in this paper · explained in 5 episodes
• Knowledge Distillation — 1× in this paper · explained in 11 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Proximal Policy Optimization (PPO) — 1× in this paper · explained in 5 episodes
• Regularization — 1× in this paper · explained in 3 episodes
• Supervised Fine-Tuning (SFT) — 1× in this paper · explained in 8 episodes
• Transformer — 1× in this paper · explained in 2 episodes
• Transformer Models — 1× in this paper · explained in 2 episodes
• Unified Framework — 1× in this paper · explained in 5 episodes

Sources

• DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
• DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models
• On-Policy Distillation with Curriculum Turn-level Guidance for Multi-turn Agents
• Turning Video Models into Generalist Robot Policies
• Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe
• Multi-Turn On-Policy Distillation with Prefix Replay
• Code as Agent Harness
• Trinity-RFT: A General-Purpose and Unified Framework for Reinforcement Fine-Tuning of Large Language Models
• Qwen3 Technical Report
• Near-Policy: Accelerating On-Policy Distillation via Asynchronous Generation and Selective Packing
• Proximal Policy Optimization Algorithms
• A Survey of On-Policy Distillation for Large Language Models
• Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation
• Multi-Rollout On-Policy Distillation via Peer Successes and Failures
• SAGE-OPD: Selective Agent-Guided Intervention for Multi-Turn On-Policy Distillation
• TurnOPD: Making On-Policy Distillation Turn-Aware for Efficient Long-Horizon Agent Training

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks