Samantha Z.

this error went away, sorry I forgot to write back

I guess it's possible if you're using DictReader rather than e.g pandas, maybe it's so old it can't handle a lot of things that newer libraries like pandas can do easily with reading and writing CSVs.

so metadata is required? there is no metadata for this, there's only 1 sample in the file

Trying again with a smaller file, but that's also hanging. 😞

Is there a size limit on what we can put in a dataset?

It finally gave me an error saying "failed to create dataset", which isn't very informative. How do I know why it failed, or what to do about it?

I'm attempting to create a golden dataset in Phoenix. Took some finagling to create a file in a format that I think should work, but the UI is just hanging now, so that doesn't inspire a lot of confidence. It's a 'CSV' with just 2 columns and 2 rows, each column has 1 very long string in it.

Actually the easiest for us would be if we could upload separate files, 1 for input, and 1 for expected output.

Yes, though at the moment I just want an easy way to iterate in a 'playground' setting where I can quickly change a prompt and re-run with different model versions, but the data we include with our prompt is fairly large, so copy-pasting it doesn't work well. The other eval frameworks I've used all had this 'playground' option where I could choose a dataset from a dropdown and then just iterate on the prompt that went with it.

ok, now I'm remembering why I never made any datasets. thanks.