Raphael Mitsch

> Yes, for the combined results. But without a definite use case it's not really important, so just and idea. Added an export, stored as `evaluation/{dataset name}/compare-{datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}.csv`. Does this make...

> Is this just about the formatting/appearance? I don't really have a strong opinion either way, so I think it's fine as-is. No. If you run `evaluate`, you'll see two...

Closing this due to lack of a response. @Smyja If you are still working on this, feel free to open a new PR - we'd welcome your contribution! :slightly_smiling_face:

@danieldk Does this kind of modification to the tests reflect what you had in mind? If so, I'll go on adjusting the others that are relevant w.r.t. https://github.com/explosion/thinc/pull/599#discussion_r864553335.

> I don't think this is easily possible, but maybe something we want to look into longer term -- ideally, we'd generate examples based on the type signatures. That's interesting...

Looked into this a bit. From Python 3.7 upwards this kind of type resolution would definitely work, 3.6 and below is a bit of a pain. To resolve function return...

@adrianeboyd Are we good with this or do we want another review round?

@adrianeboyd Does the current state of this PR address your concerns? If so, I'll update the docs.

> From a UX perspective I think it would be better to display the table rows as available (so it doesn't look like it's hanging) and then the result at...

> For textcat I think it's a bug in the scorer that the threshold is not set to `0.0` for `multi_label=False`. Do we want to address this in this PR?