Pixel-Linguist/rendered-sts14

--- dataset_info: features: - name: sentence1 dtype: image - name: sentence2 dtype: image - name: score dtype: float64 splits: - name: test num_bytes: 41188467.5 num_examples: 3750 download_size: 31780216 dataset_size: 41188467.5 configs: - config_name: default data_files: - split: test path: data/test-* --- ### Dataset Summary This dataset is rendered to images from STS-14. We envision the need to assess vision encoders' abilities to understand texts. A natural way will be assessing them with the STS protocols, with texts rendered into images. **Examples of Use** Load test split: ```python from datasets import load_dataset dataset = load_dataset("Pixel-Linguist/rendered-sts14", split="test") ``` ### Languages English-only; for multilingual and cross-lingual datasets, see `Pixel-Linguist/rendered-stsb` and `Pixel-Linguist/rendered-sts17` ### Citation Information ``` @article{xiao2024pixel, title={Pixel Sentence Representation Learning}, author={Xiao, Chenghao and Huang, Zhuoxu and Chen, Danlu and Hudson, G Thomas and Li, Yizhi and Duan, Haoran and Lin, Chenghua and Fu, Jie and Han, Jungong and Moubayed, Noura Al}, journal={arXiv preprint arXiv:2402.08183}, year={2024} } ```数据集信息： 特征： - 特征名：sentence1，数据类型：图像 - 特征名：sentence2，数据类型：图像 - 特征名：score，数据类型：64位浮点数(float64) 数据集划分： - 划分名称：test，字节数：41188467.5，样本数：3750 下载大小：31780216 数据集总大小：41188467.5 配置项： - 配置名称：default，数据文件： - 划分：test，路径：data/test-* ### 数据集概述 本数据集由STS-14渲染为图像而来。我们认为，评估视觉编码器(vision encoder)理解文本的能力是一项实际需求，而采用STS协议、将文本渲染为图像的评估方式，正是契合该需求的自然路径。 ### 使用示例 加载test划分： python from datasets import load_dataset dataset = load_dataset("Pixel-Linguist/rendered-sts14", split="test") ### 语言说明 仅支持英语；如需多语言及跨语言数据集，请参阅`Pixel-Linguist/rendered-stsb`与`Pixel-Linguist/rendered-sts17`。 ### 引用信息 @article{xiao2024pixel, title={Pixel Sentence Representation Learning}, author={Xiao, Chenghao and Huang, Zhuoxu and Chen, Danlu and Hudson, G Thomas and Li, Yizhi and Duan, Haoran and Lin, Chenghua and Fu, Jie and Han, Jungong and Moubayed, Noura Al}, journal={arXiv preprint arXiv:2402.08183}, year={2024} }