Neomi26/cc12m-images-007
收藏资源简介:
cc12m-images-007是CC12M数据集的一个子集,包含90,000张图像,这些图像被重新托管为单独的JPEG文件,以便于通过浏览器直接访问。图像URL遵循特定模式:`https://huggingface.co/datasets/Neomi26/cc12m-images-007/resolve/main/{folder}/{key}.jpg`。数据集提供清单文件`manifest.parquet`,包含以下列:`key`(图像键)、`image_url`(图像URL)、`source_url`(源URL)、`caption`(标题)、`width`(宽度)、`height`(高度)、`original_width`(原始宽度)、`original_height`(原始高度)、`shard`(分片)、`repo`(仓库)、`path`(路径)。该数据集是Neomi26/cc12m-images-NNN系列的一部分,该系列共有约131个仓库,全局索引可通过Neomi26/cc12m-images-index访问。原始数据集为pixparse/cc12m-wds,主要用于图像到文本任务,标签包括cc12m和images,规模类别为10K到100K之间。
cc12m-images-007 is a subset of the CC12M dataset, re-hosted as individual JPEG files for browser access. It contains 90,000 images with a URL pattern: `https://huggingface.co/datasets/Neomi26/cc12m-images-007/resolve/main/{folder}/{key}.jpg`. The manifest file `manifest.parquet` includes columns: `key`, `image_url`, `source_url`, `caption`, `width`, `height`, `original_width`, `original_height`, `shard`, `repo`, `path`. It is part of the `Neomi26/cc12m-images-NNN` series (approximately 131 repos in total), with a global index available at `Neomi26/cc12m-images-index`. The original dataset is pixparse/cc12m-wds, categorized under image-to-text tasks, with tags such as cc12m and images, and a size category of 10K to 100K.




