Table 2 The hyperparameters of fine-tuned PEGASUS
From: Logical definition-based identification of potential missing concepts in SNOMED CT
Parameter | Value |
|---|---|
Learning rate | 5e-5 |
Batch size per GPU | 32 |
Maximum epochs | 50 |
Optimizer | AdamW |
Max input tokens | 1024 |
Max target tokens | 128 |
Metric for best model | Loss |