Table A1. Single-dataset classifiers, no Q.

Instances where training and test data belong to the same language model are highlighted (bold).

Training data	Test data
	s		xl		s-k		xl-k		GPT3		Grover
	Acc.	AUC	Acc.	AUC	Acc.	AUC	Acc.	AUC	Acc.	AUC	Acc.	AUC
s	0.894	0.962	0.729	0.838	0.486	0.312	0.471	0.281	0.512	0.491	0.484	0.451
xl	0.867	0.957	0.777	0.864	0.443	0.311	0.427	0.289	0.410	0.415	0.462	0.449
s-k	0.492	0.275	0.486	0.335	0.917	0.972	0.800	0.903	0.617	0.775	0.574	0.732
xl-k	0.454	0.174	0.457	0.277	0.887	0.959	0.837	0.917	0.622	0.724	0.566	0.684
GPT3	0.445	0.266	0.458	0.350	0.703	0.791	0.624	0.705	0.739	0.828	0.585	0.629
Grover	0.386	0.265	0.444	0.404	0.705	0.755	0.675	0.719	0.537	0.526	0.683	0.760