Itandy/raat_ir0.7_d5k_0.5mix1.0

raat-ir0.7-d5k-0.5mix1.0

This model has been fine-tuned using the Divide-Then-Align (DTA) method, as proposed in the paper Divide-Then-Align: Honest Alignment based on the Knowledge Boundary of RAG. The training was conducted on RAAT, a RAFT-based model built upon Llama2-7B-base.

Training Configuration

• training datasize: 5k
• idk_ratio: 0.7
• coe_cls: 0.5
• coe_sft: 1.0

Evaluation

For your reference, we include the evaluation results of this checkpoint on the same test set as in the original paper. Minor differences in metrics were observed.

Model	OQ	AQ			RH		AbQ
	Acc	Rec	Prec	F1	DR	CUR	ARec	APrec	AF1
raat-ir0.7-d5k-0.5mix1.0	63.5	70.1	60.4	64.9	75.8	60.8	50.7	73.4	60.0

Citation

@misc{sun2025dividethenalignhonestalignmentbased,
      title={Divide-Then-Align: Honest Alignment based on the Knowledge Boundary of RAG}, 
      author={Xin Sun and Jianan Xie and Zhongqi Chen and Qiang Liu and Shu Wu and Yuehe Chen and Bowen Song and Weiqiang Wang and Zilei Wang and Liang Wang},
      year={2025},
      eprint={2505.20871},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2505.20871}, 
}