# this file was generated by the following command(s):
#   /Users/moot/Corpus/WSJ/candc-1.00/bin/train_pos --input m3.txt --ifmt %w|%p|%s \n --model all_pos --solver bfgs --niterations 10000 --verbose --comment Training and tagging the POS tagger


# the tag dictionary file path
tagdict = "//tagdict"
# the set of tags for unknown words
unknowns = "//unknowns"
# the minimum frequency cutoff for features
cutoff_default = 1
# the minimum frequency cutoff for word features
cutoff_words = 1
# the word frequency for which rare word features are used
rare_cutoff = 5
# the number of best tags to keep in the beam
beam_width = 5
# the ratio of the worst:best tags in the beam
beam_ratio = 0.005
# the beta*ratio of the worst:best tags in the forward step
forward_beam_ratio = 0.01
# the minimum frequency for adding a word-tag pair to the tag dict
tagdict_min = 5
# the ratio of the min:max frequency of word-tag pairs in the tag dict
tagdict_ratio = 500

# maximum sentence length the tagger will accept
maxwords = 250
# the set of tags for unknown number tokens
number_unknowns = "//number_unknowns"

# maximum entropy model
# an explanatory comment for the model
model-comment = "Training and tagging the POS tagger"
# the path to the original training data file
model-data = "m3.txt"
# the maximum entropy model solver [gis, bfgs]
model-solver = "bfgs"
# the smoothing method [none, gaussian]
model-smoothing = "gaussian"
# the smoothing parameter
model-sigma = 0.707
# the number of interations the solver should perform
model-niterations = 10000
# the weights file path
model-weights = "//weights"
