I got an error with a stacktrace when trying to save annotations for a new dataset. The last few lines are
File "C:\NLP\venv\lib\site-packages\prodigy\recipes\ner.py", line 266, in
doc.char_span(span["start"], span["end"], label=span[label])
TypeError: unhashable type: 'list'
The UI has a popup:
When I killed the process, I did get the message Saved 8 annotations to database SQLite
However, when I tried a db-out for the dataset, no file was created, and the message said there were no annotations.
The error message was not helpful. I think the error was because I had the --update flag on. Removing it seemed to resolve the error. Shouldn't the dataset be created if it didn't exist?
Thanks for the report! It looks like this should be label=span["label"] instead. I've already fixed this for the next release. In the meantime, you can also add this in yourself by editing the line in recipes/ner.py (you can run prodigy stats to find the location of your Prodigy installation).
Edit: Just released a new nightly that fixes the problem:
Saved 8 annotations to database SQLite