Table 5.

Comparison of results for the str2 and near-backbone-11 alphabets for two architectures specified in Table 1 trained on multiple alignments created with three different protocols

Alignment	SAM-T2K	SAM-T04	SAM-T06
str2	mean bits saved
IDaaHr	1.07	1.02	1.05
IDGaaH13	1.10	1.09	1.12
near-backbone-11	mean bits saved
IDaaHr	0.51	0.48	0.49
IDGaaH13	0.54	0.52	0.55

The difference between architectures is consistent across the different multiple alignments, with the IDGaaH13 architecture (which uses a guide sequence) consistently outperforming the IDaaHr architecture.