swap-uniba/il_calderone

这是一个意大利语版本的机器翻译数据集，源自The Cauldron。原始数据集包含50个任务，但只有15个任务在机器翻译后仍保持其意义，因此被选中进行翻译。在这15个任务中，选择了前10k行进行翻译，并丢弃了未正确翻译的问答对。图像路径的格式为{task-name}/images/{row_number}_{image_number}，其中{task-name}是原始数据集中的任务名称，{row_number}是原始数据集中的行号，{image_number}是图像的索引。

This is a machine-translated Italian version of the [The Cauldron](https://huggingface.co/datasets/HuggingFaceM4/the_cauldron) dataset. Out of the 50 tasks in the original dataset, only 15 tasks that retain their meaning after translation, particularly those not relying on textual content of images, are considered. From these 15 tasks, the first 10,000 rows were selected for machine translation, and incorrectly translated question-answer pairs were discarded. Image paths are formatted according to a specific strategy to maintain correspondence with the original dataset.