Datasets:

---

universal-dependencies

/

universal_dependencies

Like 8

Follow

universal-dependencies 10

I would also be happy if we could reorder the columns for the preview. In terms of planning the release itself, there's unfortunately a large margin of error in the estimate of when we'll get access to the data, so I'm afraid we'll have to plan a meeting quite spontaneously.

reordering of columns in the preview could either be:

• waiting for the open issues to get resolved, here: https://github.com/huggingface/dataset-viewer/issues/2639 and the related one: https://github.com/huggingface/dataset-viewer/issues/2925
  or
• changing the generation process, here: https://github.com/bot-zen/ud-hf-parquet-tools/blob/main/src/ud_hf_parquet_tools/generator.py#L317-L355

in either case, what would be the preferred order ... ?

(somewhat unrelated but nice to see: a few models have picked up the recent changes - since the models have recent updates - and the dataset gets used: https://huggingface.co/models?other=dataset:universal-dependencies%2Funiversal_dependencies&sort=modified)

oh that is nice to see! For the columns, I was thinking it would be good to show some first that are the most representative and informative for someone who isn't very familiar with UD. so maybe tokens lemmas upos feats?

I already moved "comments" at the end, here is the PR: https://github.com/bot-zen/ud-hf-parquet-tools/pull/2
This seemed already more readable to me, do we want to change other stuff?