SPARK: Adaptive Parametric Knowledge Participation for Faithful Retrieval-Augmented Generation

arXiv:2606.27786 · cs.CL, cs.AI · Submitted 2026-06-26 · Read on arXiv

cs.CL, cs.AI

Submitted: 2026-06-26

Updated: 2026-10-13

Code: https://github.com/OpenBMB/SHIFT

Terminology

• RAG — 2× in this paper · explained in 3 episodes
• Retrieval-Augmented Generation (RAG) — 1× in this paper · explained in 16 episodes

Sources

• Context-DPO: Aligning Language Models for Context-Faithfulness
• Parameters vs. Context: Fine-Grained Control of Knowledge Reliance in Language Models
• SearchQA: A New Q&A Dataset Augmented with Context from a Search Engine
• Transformer Feed-Forward Layers Are Key-Value Memories
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• Measuring Massive Multitask Language Understanding
• ParamMute: Suppressing Knowledge-Critical FFNs for Faithful Retrieval-Augmented Generation
• Taming Knowledge Conflicts in Language Models
• Resisting Contextual Interference in RAG via Parametric-Knowledge Reinforcement
• CtrlA: Adaptive Retrieval-Augmented Generation via Inherent Control
• Entity-Based Knowledge Conflicts in Question Answering
• Locating and Editing Factual Associations in GPT
• Beyond Accuracy: Evaluating Self-Consistency of Code Large Language Models with IdentityChain
• What does the Knowledge Neuron Thesis Have to do with Knowledge?
• Training language models to follow instructions with human feedback
• ToolRL: Reward is All Tool Learning Needs
• In-Context Retrieval-Augmented Language Models
• IRCAN: Mitigating Knowledge Conflicts in LLM Generation via Identifying and Reweighting Context-Aware Neurons
• AdaCAD: Adaptively Decoding to Balance Conflicts between Contextual and Parametric Knowledge
• Self-Consistency Improves Chain of Thought Reasoning in Language Models

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