From Memorization to Absorption: Mixed-Policy RL for Continual Knowledge Injection

arXiv:2608.25243 · cs.CL, cs.LG · Submitted 2026-08-26 · Read on arXiv

cs.CL, cs.LG

Submitted: 2026-08-26

Updated: 2026-08-26

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• GRPO — 40× in this paper · explained in 6 episodes
• Benchmark — 20× in this paper · explained in 2 episodes
• Accuracy — 16× in this paper · explained in 1 episode
• Reinforcement Learning — 16× in this paper · explained in 4 episodes
• RAG — 8× in this paper · explained in 3 episodes
• Fine-Tuning — 7× in this paper · explained in 5 episodes
• Generalization — 5× in this paper · explained in 13 episodes
• Commentary — 4× in this paper · explained in 2 episodes
• Supervised Fine-Tuning (SFT) — 4× in this paper · explained in 8 episodes
• Reward Function — 3× in this paper · explained in 2 episodes
• Temporal Reasoning — 3× in this paper · explained in 2 episodes
• Ablation — 2× in this paper · explained in 1 episode
• Adaptation — 2× in this paper · explained in 1 episode
• Alignment — 2× in this paper · explained in 1 episode
• Autonomous Learning — 2× in this paper
• Foundation Model — 2× in this paper · explained in 3 episodes
• Hallucination — 2× in this paper · explained in 4 episodes
• Retrieval-Augmented Generation — 2× in this paper
• Trace — 2× in this paper · explained in 2 episodes
• Catastrophic Forgetting — 1× in this paper · explained in 5 episodes
• Consistency — 1× in this paper · explained in 3 episodes
• Continual Learning — 1× in this paper · explained in 2 episodes
• Core — 1× in this paper · explained in 2 episodes
• Direct Preference Optimization (DPO) — 1× in this paper · explained in 9 episodes
• Domain Adaptation — 1× in this paper · explained in 5 episodes
• Failure Modes — 1× in this paper · explained in 3 episodes
• Few-Shot Learning — 1× in this paper · explained in 4 episodes
• Group Relative Policy Optimization — 1× in this paper
• Group Relative Policy Optimization (GRPO) — 1× in this paper · explained in 5 episodes
• Knowledge Graph — 1× in this paper · explained in 7 episodes
• New York Stock Exchange — 1× in this paper
• Procedural Knowledge — 1× in this paper · explained in 2 episodes
• Reasoning Traces — 1× in this paper · explained in 3 episodes
• Regularization — 1× in this paper · explained in 3 episodes
• Retrieval-Augmented Generation (RAG) — 1× in this paper · explained in 16 episodes

Sources

• A Dataset for Answering Time-Sensitive Questions
• SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training
• The Llama 3 Herd of Models
• Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning
• DeepSeek-V3 Technical Report
• Understanding R1-Zero-Like Training: A Critical Perspective
• Towards Continual Knowledge Learning of Language Models
• Decoupled Weight Decay Regularization
• Learning What Reinforcement Learning Can't: Interleaved Online Fine-Tuning for Hardest Questions
• KORE: Enhancing Knowledge Injection for Large Multimodal Models via Knowledge-Oriented Controls
• Injecting New Knowledge into Large Language Models via Supervised Fine-Tuning
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Temporal Reasoning with Large Language Models Augmented by Evolving Knowledge Graphs
• Attention Residuals
• LLaMA: Open and Efficient Foundation Language Models
• A Survey of Large Language Model Agents for Question Answering
• Critique-GRPO: Advancing LLM Reasoning with Natural Language and Numerical Feedback
• When Silence Is Golden: Can LLMs Learn to Abstain in Temporal QA and Beyond?
• Continual Learning for Large Language Models: A Survey
• Qwen3 Technical Report

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