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

This repository contains the model checkpoints for SPARK, a lightweight framework designed to resolve knowledge conflicts in retrieval-augmented generation (RAG).

• Paper: SPARK: Adaptive Parametric Knowledge Participation for Faithful Retrieval-Augmented Generation
• Repository: GitHub - OpenBMB/SPARK

Method Overview

SPARK is a lightweight framework that learns adaptive knowledge participation policies for faithful retrieval-augmented generation. With the backbone LLM frozen, its participation policy dynamically regulates layer-wise FFN contributions, determining when and to what extent parametric knowledge participates during generation. Optimized through faithfulness-guided reinforcement learning using response-level rewards, SPARK mitigates catastrophic forgetting and better preserves general generation capabilities compared to full-parameter supervised fine-tuning.

Setup and Usage

Please refer to the official GitHub Repository for detailed environment setup, training, and evaluation scripts.

Citation

If you find this work useful, please cite the paper:

@misc{li2026sparkadaptiveparametricknowledge,
      title={SPARK: Adaptive Parametric Knowledge Participation for Faithful Retrieval-Augmented Generation}, 
      author={Ruochang Li and Pengcheng Huang and Zhenghao Liu and Yukun Yan and Huiyuan Xie and Yu Gu and Ge Yu and Maosong Sun},
      year={2026},
      eprint={2606.27786},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2606.27786}, 
}