- Notifications
You must be signed in to change notification settings - Fork328
-
Thanks for the amazing work! I am trying to build a prof of concept using huggingface models that might be replaced with custom build once later on.eleldar/theme-classification supports the arguments multi_class and candidate_labels but I am currently running into problems applying them correctly. topics JSONB GENERATED ALWAYSAS (pgml.transform( task=>'{ "task": "text-classification", "model": "eleldar/theme-classification" }'::JSONB, args=>'{ "multi_class": true, "candidate_labels": "[a, b,c]" }'::JSONB, inputs=> ARRAY[content])) STORED This results in:
What am i missing? |
BetaWas this translation helpful?Give feedback.
All reactions
Replies: 1 comment 1 reply
-
To accept the multi_class + candidate_labels arguments, we'll need to add support for the zero shot classification pipeline.
|
BetaWas this translation helpful?Give feedback.
All reactions
1 reply
-
Issue#1519 |
BetaWas this translation helpful?Give feedback.
All reactions
❤️ 1
Sign up for freeto join this conversation on GitHub. Already have an account?Sign in to comment