A compressed large language model embedding dataset of ICD 10 CM descriptions

BMC Bioinformatics

Table 1 The autoencoder parameters and performance ordered by increasing validation loss

Embedding dimension	Batch size	Training loss	Validation loss
100	64	0.534	0.339
100	128	0.487	0.381
50	256	0.403	0.392
1000	64	0.542	0.402
100	256	0.556	0.444
1000	128	1.073	0.486
10	256	0.599	0.594
10	128	0.628	0.609
10	64	0.679	0.641
50	64	1.134	0.699
1000	256	30.435	0.803
50	128	1.053	0.894

ISSN: 1471-2105