Datasets:

ai-forever

/

spellcheck_punctuation_benchmark

like 2

Indeed, the dataset viewer doesn't support script-based datasets anymore, and the solution to get it back is to convert the dataset to "data-only" (see https://huggingface.co/docs/hub/datasets-manual-configuration)

The simplest way to do it, for now, is to rely on the datasets library's method push_to_hub (see https://huggingface.co/datasets/jmhessel/newyorker_caption_contest/discussions/4 for an example)

We're working on a CLI tool to do it more easily (https://github.com/huggingface/datasets/pull/6795)