67 lines
1.4 KiB
INI
67 lines
1.4 KiB
INI
# Configuration file for Simplified Chinese Desktop recognizer
|
|
|
|
# Configuration
|
|
Locale=CHS
|
|
Config=label_GB1
|
|
|
|
# The list of catagories to select from the .loc file for this configuration.
|
|
CatagoryList=GB1
|
|
|
|
# The file containing the list of FFF file to train from.
|
|
TrainSets="//tpg/reco/TextInk/chs/gb1-train-list.txt"
|
|
|
|
# Free input training set
|
|
FreeTrainSet=""
|
|
|
|
# Tuning sets
|
|
NatTuneSet="//tpg/reco/test/chs/tune-rare000.tfl"
|
|
UniTuneSet=$NatTuneSet
|
|
FreeTuneSet=$NatTuneSet
|
|
|
|
# Unicode text file to generate unigrams and bigrams from.
|
|
TextFile="//tpg/reco/corpora/chs/all.utf"
|
|
|
|
# List of tests to run on recognizer.
|
|
TestList="GB1L"
|
|
|
|
# Target size to shrink zilla to.
|
|
ZillaSize=10000000
|
|
|
|
# Select the recognizer for 1 and 2 stroke characters
|
|
UseFugu=0
|
|
UseOtter=1
|
|
UseSole=0
|
|
UseJaws=0
|
|
|
|
# Select the postprocessor
|
|
UseCrane=1
|
|
UseHawk=0
|
|
|
|
# Shrink parameters for otter.
|
|
OtterStrongCluster=10
|
|
OtterWeakCluster=1
|
|
OtterSize=1000000
|
|
|
|
# Otter limit on number of samples per (folded) code point
|
|
OtterSampleLimit=25000
|
|
|
|
# Number of classes to use for Centipede
|
|
CentipedeClasses=32
|
|
|
|
# Weight of natural frequency vs. uniform frequency in training set.
|
|
NatuFreqWeight=1.0
|
|
|
|
# Flag for WinCE configuration.
|
|
# Set to 1 if true, 0 if false.
|
|
IsWinCE=0
|
|
|
|
# Character detector settings
|
|
UseCharDet=0
|
|
|
|
# Break net and segmentation net settings
|
|
UseBrkNet=0
|
|
UseSegNet=0
|
|
|
|
# Whether to do IFELang3 tuning and testing
|
|
UseIFELang3=0
|