Skip to main content
. 2022 Mar 12;14:12. doi: 10.1186/s13321-022-00596-6

Table 1.

Bayesian optimization for hyperparameters in D-MPNN with RDKit descriptors

Hyperparameters Values
Dropout probability [0, 0.4] (Interval: 0.05)
Number of message-passing iterations 2, 3, 4, 5, 6
Number of feed-forward layers 1, 2, 3
Hidden size of D-MPNN [300, 2400] (Interval: 100)