Table 6.

Training Time and Hyperparameters in TLOS

Token length	Train epochs	Batch size	Optimizer	Learning rate	Training time per epoch (Min)
128	10	16	Adam	2e-5	12±0.24
256	10	16	Adam	2e-5	23±0.70
328	10	16	Adam	2e-5	31±0.47
468	10	16	Adam	2e-5	39±1.21
512	10	16	Adam	2e-5	43±0.62