README.md

metadata

language:
  - en
license: apache-2.0
task_categories:
  - text-generation
tags:
  - rag
  - retrieval-augmented-generation

SPARK Training Data

This repository contains the training data for SPARK, presented in the paper SPARK: Adaptive Parametric Knowledge Participation for Faithful Retrieval-Augmented Generation.

• Repository: https://github.com/OpenBMB/SPARK
• Paper: https://arxiv.org/abs/2606.27786

Dataset Description

SPARK is a lightweight framework for resolving knowledge conflicts in retrieval-augmented generation (RAG). 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. This dataset contains the training data synthesized with FlashRAG using MRQA-Shared-Task-2019 to train the participation policy of SPARK.

Citation

If you find this work useful, please cite our 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}, 
}