README.md

metadata

configs:
  - config_name: mmlu
    data_files:
      - split: test
        path: mmlu/test.csv
      - split: labeled
        path: mmlu/labeled.csv
      - split: unlabeled
        path: mmlu/unlabeled.csv
  - config_name: mmlu_pro
    data_files:
      - split: test
        path: mmlu_pro/test.csv
      - split: labeled
        path: mmlu_pro/labeled.csv
      - split: unlabeled
        path: mmlu_pro/unlabeled.csv
  - config_name: arc
    data_files:
      - split: test
        path: arc/test.csv
      - split: labeled
        path: arc/labeled.csv
      - split: unlabeled
        path: arc/unlabeled.csv
  - config_name: FPB
    data_files:
      - split: test
        path: FPB/test.csv
      - split: labeled
        path: FPB/labeled.csv
      - split: unlabeled
        path: FPB/unlabeled.csv
  - config_name: USMLE
    data_files:
      - split: test
        path: USMLE/test.csv
      - split: labeled
        path: USMLE/labeled.csv
      - split: unlabeled
        path: USMLE/unlabeled.csv
  - config_name: PubMedQA
    data_files:
      - split: test
        path: PubMedQA/test.csv
      - split: labeled
        path: PubMedQA/labeled.csv
      - split: unlabeled
        path: PubMedQA/unlabeled.csv
  - config_name: ConvFinQA
    data_files:
      - split: test
        path: ConvFinQA/test.csv
      - split: labeled
        path: ConvFinQA/labeled.csv
      - split: unlabeled
        path: ConvFinQA/unlabeled.csv
license: cc-by-4.0
task_categories:
  - question-answering
  - multiple-choice
  - text-generation
language:
  - en
tags:
  - finance
  - medical

Dataset Card for Dataset Name

The SemiEvol dataset is part of the broader work on semi-supervised fine-tuning for Large Language Models (LLMs). The dataset includes labeled and unlabeled data splits designed to enhance the reasoning capabilities of LLMs through a bi-level knowledge propagation and selection framework, as proposed in the paper SemiEvol: Semi-supervised Fine-tuning for LLM Adaptation.

Dataset Details

Dataset Sources [optional]

• Repository:
  • https://huggingface.co/datasets/cais/mmlu
  • https://huggingface.co/datasets/TIGER-Lab/MMLU-Pro
  • https://huggingface.co/datasets/cais/mmlu;
  • https://huggingface.co/AdaptLLM;
• Paper [optional]: https://huggingface.co/papers/2410.14745

Uses

Direct Use

Please refer to https://github.com/luo-junyu/SemiEvol for evaluation instruction.

Citation [optional]

BibTeX:

@misc{luo2024semievol,
    title={SemiEvol: Semi-supervised Fine-tuning for LLM Adaptation}, 
    author={Junyu Luo and Xiao Luo and Xiusi Chen and Zhiping Xiao and Wei Ju and Ming Zhang},
    year={2024},
    eprint={2410.14745},
    archivePrefix={arXiv},
    primaryClass={cs.CL},
    url={https://arxiv.org/abs/2410.14745}, 
}

APA:

Luo, J., Luo, X., Chen, X., Xiao, Z., Ju, W., & Zhang, M. (2024). SemiEvol: Semi-supervised Fine-tuning for LLM Adaptation. ArXiv. https://arxiv.org/abs/2410.14745