相关数据集
LOKI
LOKI数据集由中山大学和上海人工智能实验室等机构联合创建，旨在评估大型多模态模型在检测合成数据方面的能力。该数据集包含视频、图像、3D、文本和音频五种模态，共计18,000条问题，覆盖26个详细子类别。数据集的创建过程包括使用多种合成模型生成高质量数据，并通过精细的异常标注进行分级。LOKI数据集主要应用于合成数据检测领域，旨在解决未来互联网中合成数据泛滥带来的真实性鉴别难题。
arXiv2024-10-13 更新3960
Synthetic data generated in Unreal Engine 4
Dataset generated with Unreal Engine 4 and Nvidia NDDS. Contains 1500 images of each object: Forklift, pallet, shipping container, barrel, human, paper box, crate and fence.
DataCite Commons2022-08-12 更新110
GenerateNumbersAccToEquation
This was generated as part of DSCI560 HW2 Task1b. This dataset contains the y=3*x+2 relationship of 1000 numbers generated randomly in the range of 0-100. The random numbers originally generated ca
Figshare2020-09-24 更新60
dralsarrani/prompt_safety
Prompt Safety Aggregation Dataset是一个经过策划的提示安全示例聚合数据集，收集自三个高质量来源：SG-Bench、do_not_answer_en.csv和SalKhan12/prompt-safety-dataset。该数据集旨在支持安全感知语言模型的发展和评估。它包含英语语言的双标签安全注释（例如，安全、不安全），并适用于文本分类、提示过滤、安全评估和越狱检测等
Hugging Face2025-11-03 更新120
Supplementary information files for A genetically-optimised artificial life algorithm for complexity-based synthetic dataset generation
Supplementary files for article A genetically-optimised artificial life algorithm for complexity-based synthetic dataset generation Algorithmic evaluation is a vital step in developing new approa
DataCite Commons2023-03-29 更新80
