TAFM: A Recommendation Algorithm Based on Text-Attention Factorization Mechanism

Table 2

Effect of different learning rates and batch sizes on the model.

Learning rate	Batch size	Accuracy	AUC	Log_loss
0.0001	2048	0.825604	0.697758	0.4256
0.0005	2048	0.825103	0.697491	0.4263
0.001	16	0.821059	0.673495	0.4414
0.001	32	0.825450	0.681666	0.4342
0.001	64	0.824624	0.717330	0.4216
0.001	128	0.826032	0.724808	0.4166
0.001	256	0.826041	0.726161	0.4162
0.001	512	0.825287	0.695343	0.4265
0.001	1024	0.826938	0.730182	0.4144
0.001	2048	0.826891	0.728792	0.4141
0.005	2048	0.825827	0.697286	0.4257
0.01	2048	0.823770	0.682339	0.4326
0.1	2048	0.822422	0.5	17.7577