Many classification datasets contain a tweet id and one or more class labels. These could get easily converted to something like:
{ "id": "903232323", "class1": "this", "class2": "that" }
It would be very useful if the output would just contain all the non-id, non-entity fields because those jsonl files could then get directly
used or converted for the classification task.
Many classification datasets contain a tweet id and one or more class labels. These could get easily converted to something like:
It would be very useful if the output would just contain all the non-id, non-entity fields because those jsonl files could then get directly
used or converted for the classification task.