Convert task category to indicate modality #2107
Conversation
|
Can we add new field to TaskMetadata to save information about |
A big part of this is accepting that such info can be derived from descriptive stats, as @KennethEnevoldsen elaborated here. The discussion concluded with the proposed changes here. |
|
Yes, but can we add it in this PR since we're changing |
|
This PR is about the breaking change. That's why the target is not main. A warning has been added in the linked PR. Would prefer to keep this unchanged. |
|
I see, but we totally don't want to save any information about |
Wouldn't that just be in the descriptive stats? |
|
We have only statistics for this, but no indication for it. We can add property for it |
|
We could do some sort of cutoff or a quick way to access relevant descriptive stats. I could see that it would be nice but I agree with @isaac-chung that it is not for this PR. |
|
Hmm the datasets exist but tests fail 🤔 |
|
Yes, on |
| @@ -14,7 +14,7 @@ class RestaurantReviewSentimentClassification(AbsTaskClassification): | |||
| description="Dataset of 8364 restaurant reviews from qaym.com in Arabic for sentiment analysis", | |||
| reference="https://link.springer.com/chapter/10.1007/978-3-319-18117-2_2", | |||
| type="Classification", | |||
| category="s2s", | |||
| category="t2t", | |||
There was a problem hiding this comment.
I'm not sure that classification and clustering tasks should be considered t2t, as they don't fully fit the definition, but this is a minor
Code Quality
make lintto maintain consistent style.Documentation
Testing
make test-with-coverage.make testormake test-with-coverageto ensure no existing functionality is broken.