I’m setting up a conversational labeling pipeline and am having issues viewing the conversations in the Datasets tabs.
In the labeling section, the conversations show up nicely in the UI and are easy to review; however, in the Datasets tab, they appear as raw JSON objects and are difficult to review.
How can I configure the Datasets viewer so they appear as conversations?
We currently do not support to have the same “view” as the editor view in the Catalog but I will take your feedback and share this with our product team.
Thanks for the quick response!
I think this makes it almost impossible to use the dataset viewer. Can we use the custom UI JS tool for this? How do other customers with conversational datasets handle this?
Given we use a proprietary type of asset this would rule out custom JS.
From what I see customer, label their data, review them and then export.
I understand you might want to have a more visual approach via Catalog, within a project the Data rows tab has a great set of filters to help make the curation.
So customers that have conversational datasets, do not use the catalog feature of Labelbox? They use a separate catalog tool and then just send to labelbox what should be labeled and don’t view the dataset in Labelbox at all?