Skip to main content
. 2017 Dec 28;10(Suppl 5):73. doi: 10.1186/s12920-017-0316-8

Table 3.

The hyper-parameters and corresponding values

Hyper-parameter Value
Input context window size 13 (NCBI = CDR)
Word-level embedding dimension 200 (NCBI = CDR)
Character-level embedding dimension 20 (NCBI = CDR)
Lexicon feature embedding dimension 5 (NCBI = CDR)
Character-level CNN’s window size 3 (NCBI = CDR)
Character-level CNN’s filters number 20 (NCBI = CDR)
Word-level CNN’s window size 3 (NCBI = CDR)
Word-level CNN’s filters number 100 (NCBI = CDR)
Word-level Convolutional layers size 3 (NCBI); 4 (CDR)