lstmpars.mem14.ws9.txt

来自「C-package of "Long Short-Term Memory" fo」· 文本 代码 · 共 108 行

number memory cell blocks: 14
###here definitions for each block###
**block number: 1**
block size: 1
initial input bias: -2.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 2**
block size: 1
initial input bias: -2.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
**block number: 3**
block size: 1
initial input bias: -2.5
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 4**
block size: 1
initial input bias: -2.5
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
**block number: 5**
block size: 1
initial input bias: -3.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 6**
block size: 1
initial input bias: -3.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
**block number: 7**
block size: 1
initial input bias: -3.5
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 8**
block size: 1
initial input bias: -3.5
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
**block number: 9**
block size: 1
initial input bias: -4.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 10**
block size: 1
initial input bias: -4.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
**block number: 11**
block size: 1
initial input bias: -4.5
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 12**
block size: 1
initial input bias: -4.5
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
**block number: 13**
block size: 1
initial input bias: -5.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: 1.0
**block number: 14**
block size: 1
initial input bias: -5.0
initial input gate bias: -1.0
initial output gate bias: -1.0
initial output weight: -1.0
###end definitions for each block###
windowsize: 9
outputbias: -3.0
inputdata training positive: datasets/pos-train.fasta
inputdata test positive: datasets/pos-test.fasta
inputdata training negative: datasets/neg-train.fasta
inputdata test negative: datasets/neg-test.fasta
targetvalue0: 0.8
targetvalue1: 0.2
learning rate: 0.05
half interval length for intialization: 0.1
performing test after ? epochs: 5
write weight after ? epochs: 5
initialization of random generator: 0
reset the net after each sequence?: 1
weight update after sequence or epoch?: 1
stop learning after n epochs: 100
size of training set: 50000
maxlength of training set: 1200
size of test set: 5000
maxlength of test set: 1200