MAIGO: Mitigating Lost-in-Conversation with History-Cleaned On-Policy Self-Distillation

arXiv:2605.27186 · cs.CL · Submitted 2026-05-26 · Read on arXiv

cs.CL

Submitted: 2026-05-26

Updated: 2026-08-30

Terminology

• Accuracy — 30× in this paper · explained in 1 episode
• Distillation — 22× in this paper · explained in 2 episodes
• Verifier — 20× in this paper · explained in 2 episodes
• GRPO — 9× in this paper · explained in 6 episodes
• Self-Distillation — 9× in this paper · explained in 7 episodes
• Ablation — 7× in this paper · explained in 1 episode
• On-Policy Distillation — 7× in this paper · explained in 3 episodes
• Fine-Tuning — 5× in this paper · explained in 5 episodes
• Scaffolding — 5× in this paper · explained in 3 episodes
• Text-to-SQL — 5× in this paper · explained in 3 episodes
• Trace — 4× in this paper · explained in 2 episodes
• Adaptation — 3× in this paper · explained in 1 episode
• Abstention — 2× in this paper · explained in 1 episode
• Adaptive Weighting — 2× in this paper · explained in 1 episode
• Agentic Evaluation — 1× in this paper · explained in 1 episode
• Benchmark — 1× in this paper · explained in 2 episodes
• Benchmarking — 1× in this paper · explained in 3 episodes
• Consistency — 1× in this paper · explained in 3 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Low-Rank Adaptation (LoRA) — 1× in this paper · explained in 9 episodes
• Regret — 1× in this paper · explained in 3 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Robustness — 1× in this paper · explained in 5 episodes
• Temporal consistency — 1× in this paper · explained in 2 episodes
• Verifiable Rewards — 1× in this paper · explained in 2 episodes
• Verification — 1× in this paper · explained in 2 episodes
• alpha — 1× in this paper · explained in 1 episode

Sources

• Training Verifiers to Solve Math Word Problems
• When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction
• The Llama 3 Herd of Models
• Grounded Continuation: A Linear-Time Runtime Verifier for LLM Conversations
• Models Recall What They Violate: Constraint Adherence in Multi-Turn LLM Ideation
• LLMs Get Lost In Multi-Turn Conversation
• Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards
• Evaluating Temporal Consistency in Multi-Turn Language Models
• DeepSeek-V3 Technical Report
• Intent Mismatch Causes LLMs to Get Lost in Multi-Turn Conversation
• SEQUOR: A Multi-Turn Benchmark for Realistic Constraint Following
• Evaluating Large Language Models Trained on Code
• Breaking Contextual Inertia: Reinforcement Learning with Single-Turn Anchors for Stable Multi-Turn Interaction
• Multi-Rollout On-Policy Distillation via Peer Successes and Failures
• Pause or Fabricate? Training Language Models for Grounded Reasoning
• OPSDL: On-Policy Self-Distillation for Long-Context Language Models
• Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• SCOPE: Signal-Calibrated On-Policy Distillation Enhancement with Dual-Path Adaptive Weighting
• Mitigating Conversational Inertia in Multi-Turn Agents

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering