
All training and evaluation datasets used in this study are available at the provided https://drive.google.com/drive/folders/1jzM1tbPbiWVWtlErZfffIZGXyTB_MiV1?usp=sharing. 


The directory is organized as follows:

test/
├── out_of_domain/
│   ├── finefact500.csv
│   ├── fever.csv
│   └── fact_score.csv
│
├── in_domain/
│   ├── ragtruth.csv
│   ├── halu_eval_summ.csv
│   ├── halu_eval_qa.csv
│   ├── covid_qa.csv
│   └── anah.csv

train/
├── evidence_extraction/
│   ├── sft_25k.csv
│   └── grpo_71k.csv
│
├── claim_verification/
│   ├── sft_15k.csv
│   └── grpo_50k.csv
│
└── claim_extraction/
    ├── sft_5k.csv
    └── dpo_13k.csv
