- h
SAS-Bench
- huggingface.co
Updated Jun 17, 2025
Share
Facebook
Twitter
EmailClick to copy linkLink copiedCitePeichao Lai (2025). SAS-Bench [Dataset]. https://huggingface.co/datasets/aleversn/SAS-BenchDataset updatedJun 17, 2025AuthorsPeichao LaiLicenseApache License, v2.0https://www.apache.org/licenses/LICENSE-2.0
License information was derived automaticallyDescriptionSAS-Bench: A Fine-Grained Benchmark for Evaluating Short Answer Scoring with Large Language Models
Dataset | δΈζ | Paper | Code
π OverviewSAS-Bench represents the first specialized benchmark for evaluating Large Language Models (LLMs) on Short Answer Scoring (SAS) tasks. Utilizing authentic questions from China's National College Entrance Examination (Gaokao)β¦ See the full description on the dataset page: https://huggingface.co/datasets/aleversn/SAS-Bench.
- P
SAS-Bench Dataset
- paperswithcode.com
Updated May 11, 2025ShareFacebookTwitterEmailClick to copy linkLink copiedCitePeichao Lai; Kexuan Zhang; Yi Lin; Linyihan Zhang; Feiyang Ye; Jinhao Yan; Yanwei Xu; Conghui He; Yilei Wang; Wentao Zhang; Bin Cui (2025). SAS-Bench Dataset [Dataset]. https://paperswithcode.com/dataset/sas-benchDataset updatedMay 11, 2025AuthorsPeichao Lai; Kexuan Zhang; Yi Lin; Linyihan Zhang; Feiyang Ye; Jinhao Yan; Yanwei Xu; Conghui He; Yilei Wang; Wentao Zhang; Bin CuiDescriptionSAS-Bench represents the first specialized benchmark for evaluating Large Language Models (LLMs) on Short Answer Scoring (SAS) tasks. Utilizing authentic questions from China's National College Entrance Examination (Gaokao), our benchmark offers:
1,030 questions spanning 9 academic disciplines 4,109 expert-annotated student responses Step-wise scoring with Step-wise error analysis Multi-dimensional evaluation (holistic scoring, step-wise scoring, and error diagnosis consistency)
Not seeing a result you expected?
Learn how you can add new datasets to our index.
FacebookTwitterEmailApache License, v2.0https://www.apache.org/licenses/LICENSE-2.0
License information was derived automatically
SAS-Bench: A Fine-Grained Benchmark for Evaluating Short Answer Scoring with Large Language Models
Dataset | δΈζ | Paper | Code
π Overview
SAS-Bench represents the first specialized benchmark for evaluating Large Language Models (LLMs) on Short Answer Scoring (SAS) tasks. Utilizing authentic questions from China's National College Entrance Examination (Gaokao)β¦ See the full description on the dataset page: https://huggingface.co/datasets/aleversn/SAS-Bench.