Table 3.

Comparison of the performances of models with different numbers of heads in the self-attention layer.

Number of heads	MAE	RMSE	PCC
1	5.92 ±0.62	7.56 ±0.78	0.44 ±0.11
2	6.89 ± 0.31	8.73 ± 0.51	0.31 ± 0.16
4	6.92 ± 0.48	8.71 ± 0.61	0.34 ± 0.14
8	7.09 ± 0.58	8.96 ± 0.54	0.33 ± 0.11