cant get data out for the "ner_news_headlines" tutorial

Getting the error

✘ Can't find 'ner_news_headlines ' in database SQLite

when using the command

prodigy db-out ner_news_headlines

I was following the basic tutorial on prodigy 101, step 2.

Steps:

  • loaded server
  • created some data via webpage (high lighting and hitting accept)
  • hit save button and saw the history disappear.
  • tried to out put and got error above.

Notes:

  • I can see the sqlite file in my prodigy home folder
  • I've not changed any settings
  • using windows
  • only a single file in the folder I'm running the server from which is "news_healines.jsonl"

Let's run some diagnostics. I'm assuming that you ran this command beforehand?

prodigy ner.manual ner_news_headlines blank:en ./news_headlines.jsonl --label PERSON,ORG,PRODUCT,LOCATION

If so, there should be some annotations. But it is possible that you're dealing with a typo, so just to exclude that. What do you see when you run the stats recipe?

prodigy stats -l

This should list all the datasets that have been created. Do you see ner_news_headlines?

You can also see stats for a specific dataset by running:

prodigy stats new_news_headlines

Thank you so much for your quick reply!!!

:frowning: found the issue, im using cygwin for my terminal and it looks like there is confusion in the whitespace. So prodigy is getting an extra space even though there isn't one on the terminal.

I will need to use another terminal I guess.

I've got it working in powershell and cygwin with:

python -m prodigy db-out ner_news_headlines

thank you again... will try and work something out on my end

Happy to hear the issue is now gone.

Maybe a tip:

I'm not familiar with cygwin, I'm a zsh user myself. But with zsh I have gotten into the habbit of adding quotes. For example, this won't run with zsh:

python -m pip install embetter[text]

But this does!

python -m pip install "embetter[text]"

It could be that adding some quotes around some of your commands might also fix it. Might be worth a try.