five

artefactory/abstention-reranking-benchmark

收藏
Hugging Face2024-10-02 更新2026-05-10 收录
下载链接:
https://hf-mirror.com/datasets/artefactory/abstention-reranking-benchmark
下载链接
链接失效反馈
官方服务:
资源简介:
--- dataset_info: features: - name: model_name dtype: string - name: dataset_path dtype: string - name: num_docs_pr dtype: int64 - name: random_seed dtype: int64 - name: queries sequence: string - name: positives sequence: sequence: string - name: negatives sequence: sequence: string - name: scores sequence: sequence: float64 - name: targets sequence: sequence: float64 splits: - name: test num_bytes: 2218701030 num_examples: 132 download_size: 1121022682 dataset_size: 2218701030 configs: - config_name: default data_files: - split: test path: data/test-* --- # Towards Trustworthy Reranking: A Simple yet Effective Abstention Mechanism ## Dataset This dataset provides the evaluation benchmark used in the paper "Towards Trustworthy Reranking: A Simple yet Effective Abstention Mechanism", accepted at TMLR (September 2024). ## Paper For more details on this work, you can read the full paper [here](https://arxiv.org/pdf/2402.12997). **Citation:** ``` @article{gisserot2024towards, title={Towards Trustworthy Reranking: A Simple yet Effective Abstention Mechanism}, author={Gisserot-Boukhlef, Hippolyte and Faysse, Manuel and Malherbe, Emmanuel and Hudelot, C{\'e}line and Colombo, Pierre}, journal={arXiv preprint arXiv:2402.12997}, year={2024} } ```

dataset_info: 数据集元信息: 特征: - 名称:model_name 数据类型:字符串(string) - 名称:dataset_path 数据类型:字符串(string) - 名称:num_docs_pr 数据类型:64位整数(int64) - 名称:random_seed 数据类型:64位整数(int64) - 名称:queries 数据类型:字符串序列(sequence) - 名称:positives 数据类型:字符串序列的序列(sequence of sequence) - 名称:negatives 数据类型:字符串序列的序列(sequence of sequence) - 名称:scores 数据类型:64位浮点数序列的序列(sequence of sequence of float64) - 名称:targets 数据类型:64位浮点数序列的序列(sequence of sequence of float64) 数据集划分: - 名称:test 数据字节数:2218701030 样本数量:132 下载总大小:1121022682 数据集总大小:2218701030 配置: - 配置名称:default 数据文件: - 划分集:test 路径:data/test-* --- # 迈向可信赖重排序:一种简洁高效的弃权机制 ## 数据集 本数据集为收录于TMLR(2024年9月)的论文《迈向可信赖重排序:一种简洁高效的弃权机制》中所使用的评估基准。 ## 论文 如需了解此项研究的更多细节,可点击[此处](https://arxiv.org/pdf/2402.12997)查阅完整论文。 **引用格式:** @article{gisserot2024towards, title={迈向可信赖重排序:一种简洁高效的弃权机制}, author={Gisserot-Boukhlef, Hippolyte and Faysse, Manuel and Malherbe, Emmanuel and Hudelot, Céline and Colombo, Pierre}, journal={arXiv preprint arXiv:2402.12997}, year={2024} }
提供机构:
artefactory
二维码
社区交流群
二维码
科研交流群
商业服务