Table 2.

GPU usage test results.

Model	L	D	Node #	BS	GPU memory
LSTM-Att	500	100	50	128	8,893 MB
AD-ESN	500	100	50	128	6,425 MB
LSTM-Att	1000	10	50	128	6,581 MB
AD-ESN	1000	10	50	128	1,647 MB
LSTM-Att	5000	1	50	128	N/A
LSTM-Att	5000	1	50	↓ 32	7464 MB
AD-ESN	5000	1	50	128	2227 MB

L, Time Series Length; D, Input dimension; BS, Batch size. N/A means the GPU memory is not enough for this setting. The best performance is indicated by the use of bold numbers.