Table 2:

The AUC results for binary sequence classification tasks which have multi-species involved, including promoter region prediction (first four rows), human vs worm classification and mouse transcription factor binding site (TFBS) identification. The results for mouse TFBS are averaged over 5 independent datasets focusing on different TFBSs. Using sentence-level summary token pooling method. **DeLong Test significance < 0.01. Bolded value: DeLong Test significance < 0.05.

Data	DNABERT-2	NT-v2	HyenaDNA
Promoter B_amyloliquefaciens	0.856**	0.797	0.688
Promoter R_capsulatus	0.661	0.668	0.602
Promoter Arabidopsis NonTATA	0.891**	0.85	0.814
Promoter Arabidopsis TATA	0.903**	0.855	0.82
Human vs worm	0.946**	0.919	0.837
Mouse TFBS	0.700	0.722	0.624