Add OntoNotes to the possibles dataset · pytorch/text#730

(1 comment) (0 reactions) (0 assignees)Python (3,396 stars) (822 forks)batch import

help wanted

説明

It's a great dataset, with 18 different entities which is much more than other available dataset and on 3 major languages. Pitch

Be able to load ontonotes as the other datasets already availables. Alternatives

Let people do their custom dataloader (which is long). This kind of automation will help lot of peoples.

Additional context

Long live pytorch (and torchaudio and torchtext and all the submodules <3)

技術スタック: python
領域: datamachine learning
Issue 種別: feature
難度: 3
推定時間: 1-2 days
活動状況: stale
明確さ: mostly clear
前提条件: PyTorch basicsPythonNLP dataset formats
初心者向け度: 30
調査方針: Examine existing dataset loaders in torchtext, e.g., in torchtext.datasets. Understand the OntoNotes data format (likely CONLL style). Implement a new dataset class following the same pattern. Consider adding tests and documentation.