How can I delete an entry from a dataset once it is already saved in the database?
Thank you.
1 Like
If you want to modify an existing set, you can export the JSONL data, edit it however you like (remove entries, add entries) and then import the modified to a new dataset.
For example:
prodigy db-out old_dataset > dataset.jsonl
# modify dataset.jsonl
prodigy db-in new_dataset dataset.jsonl
In most cases, I’d recommend keeping the original dataset and adding a new one, so you don’t lose any state. Especially if you’ve already been running experiments, you don’t want to lose the ability to re-run them with the previous state and compare the results (or restore the old dataset if you’Ve made a mistake). If you do want to delete an existing dataset, you can use the prodigy drop command.
1 Like