Skip to main content

Table 5 The detailed training settings of ELECTRA-DTA

From: ELECTRA-DTA: a new compound-protein binding affinity prediction model based on the contextualized sequence encoding

Parameter Setting Parameter Setting
CNN kernel size 3 Learning rate (lr) 3e-4
Length of SMILES sequence 100 Length of protein sequence 1000
Vector dimension 256 Number of filters 256;512
Epoch 100 Batchsize 256
hidden neurons 1024; 1024; 512 dropout 0.4