def main(self, conceptMap, symbolMap, symbolDt, dcpts, dpmfs, symbolName, outDir):
conceptMap = LexMap().read(conceptMap)
symbolMap = LexMap().read(symbolMap)
_sink_ = int(symbolMap["_sink_"])
_SINK_ = int(conceptMap["_SINK_"])
conceptCard = len(conceptMap)
symbolCard = len(symbolMap)
jointDtC1C2C3C4 = gmtk.readDt(symbolDt)
jointDcptC1C2C3C4 = gmtk.readDcpt(dcpts, "jointProbC1C2C3C4")
jointProbC1C2C3C4 = gmtk.combineDtDcpt1(jointDtC1C2C3C4, jointDcptC1C2C3C4)
symbolDpmfC1C2C3C4 = gmtk.readDpmf(dpmfs, "%sGivenC1C2C3C4" % symbolName)
symbolGivenC1C2C3C4 = gmtk.combineDtDcpt2(jointDtC1C2C3C4, symbolDpmfC1C2C3C4)
jointProbSymbolC1C2C3C4 = symbolGivenC1C2C3C4.multiple([1, 2, 3, 4], jointProbC1C2C3C4)
gmtk.saveDpmfsProbs(
outDir,
"%sGivenC1C2C3C4" % symbolName,
len(symbolGivenC1C2C3C4.vectSubList([1, 2, 3, 4])),
symbolCard,
symbolGivenC1C2C3C4,
)
jointProbSymbolC1C2C3 = jointProbSymbolC1C2C3C4.marginalize([0, 1, 2, 3])
symbolGivenC1C2C3 = jointProbSymbolC1C2C3.conditionalize([1, 2, 3])
gmtk.saveDpmfsProbs(
outDir,
"%sGivenC1C2C3" % symbolName,
len(symbolGivenC1C2C3.vectSubList([1, 2, 3])),
symbolCard,
symbolGivenC1C2C3,
)
jointProbSymbolC1C2 = jointProbSymbolC1C2C3.marginalize([0, 1, 2])
symbolGivenC1C2 = jointProbSymbolC1C2.conditionalize([1, 2])
gmtk.saveDpmfsProbs(
outDir, "%sGivenC1C2" % symbolName, len(symbolGivenC1C2.vectSubList([1, 2])), symbolCard, symbolGivenC1C2
)
jointProbSymbolC1 = jointProbSymbolC1C2.marginalize([0, 1])
symbolGivenC1 = jointProbSymbolC1.conditionalize([1])
# in case of conditioning by _SINK_ I have to enable to generate _sink_ word only
# otherwise I would see _SINK_ concept in the stack
# TODO: Turn it into validator
symbolGivenC1.setValue([_sink_, _SINK_], 1)
gmtk.saveDcptBigram(outDir, "%sGivenC1" % symbolName, symbolCard, conceptCard, symbolGivenC1)
symbolUnigram = jointProbSymbolC1.marginalize([0])
# I need to enable to decode _unseen_ word only ! so set the probability of
# generating _empty_ to zero
symbolUnigram.setValue([int(symbolMap["_empty_"])], 0)
# normalize sum of probabilities to one
symbolUnigram = symbolUnigram.normJoint()
gmtk.saveDcptUnigram(outDir, "%sUnigram" % symbolName, symbolCard, symbolUnigram)
gmtk.saveDcptUnseen(outDir, "%sZerogram" % symbolName, symbolCard, symbolMap)
concept1GivenC2C3 = jointProbC1C2C3.conditionalize([1, 2])
gmtk.saveDpmfsProbs(dirOut, "concept1GivenC2C3", len(concept1GivenC2C3.vectSubList([1, 2])), conceptCard, concept1GivenC2C3)
##############################################################################
# save bigrams as CPT
jointProbC1C2 = jointProbC1C2C3.marginalize([0, 1])
concept1GivenC2 = jointProbC1C2.conditionalize([1])
# in case of conditioning by _SINK_ I have to enable to generate _SINK_ concept only
# otherwise I would see _SINK_ concept in the stack
concept1GivenC2.setValue([int(conceptMap["_SINK_"]), int(conceptMap["_SINK_"])], 1)
# do the same for _DUMMY_
concept1GivenC2.setValue([int(conceptMap["_SINK_"]), int(conceptMap["_DUMMY_"])], 1)
gmtk.saveDcptBigram(dirOut, "concept1GivenC2", conceptCard, conceptCard, concept1GivenC2)
##############################################################################
# save unigrams as CPT
concept1Unigram = jointProbC1C2.marginalize([0])
gmtk.saveDcptUnigram(dirOut, "concept1Unigram", conceptCard, concept1Unigram)
################################################################################
### save zerograms as CPT
##gmtk.saveDcptZerogram(dirOut, "concept1Zerogram", conceptCard)
if verbose:
print("-------------------------------------------------")
print("Finish")
# save trigrams
jointProbPC1C2 = jointProbPC1C2C3.marginalize([0, 1, 2])
pushGivenC1C2 = jointProbPC1C2.conditionalize([1, 2])
pushGivenC1C2 = pushGivenC1C2.insertPenalty(0, penalty, pushCard)
gmtk.saveDpmfsProbs(dirOut, "pushGivenC1C2",
len(pushGivenC1C2.vectSubList([1, 2])), pushCard,
pushGivenC1C2, -penalty)
##############################################################################
# save bigrams as CPT
jointProbPC1 = jointProbPC1C2.marginalize([0, 1])
pushGivenC1 = jointProbPC1.conditionalize([1])
pushGivenC1 = pushGivenC1.insertPenalty(0, penalty, pushCard)
# in case of conditioning by _SINK_, I have to enable to delete all stack
pushGivenC1.setValue([4, int(conceptMap["_SINK_"])], 1)
gmtk.saveDcptBigram(dirOut, "pushGivenC1", pushCard, conceptCard, pushGivenC1)
##############################################################################
# save unigrams as CPT
pushUnigram = jointProbPC1.marginalize([0])
pushUnigram = pushUnigram.insertPenalty(0, penalty, pushCard)
gmtk.saveDcptUnigram(dirOut, "pushUnigram", pushCard, pushUnigram)
if verbose:
print("-------------------------------------------------")
print("Finish")
def main(self, conceptMap, symbolMap, symbolDt, dcpts, dpmfs, symbolName,
outDir):
conceptMap = LexMap().read(conceptMap)
symbolMap = LexMap().read(symbolMap)
_sink_ = int(symbolMap['_sink_'])
_SINK_ = int(conceptMap['_SINK_'])
conceptCard = len(conceptMap)
symbolCard = len(symbolMap)
jointDtC1C2C3C4 = gmtk.readDt(symbolDt)
jointDcptC1C2C3C4 = gmtk.readDcpt(dcpts, "jointProbC1C2C3C4")
jointProbC1C2C3C4 = gmtk.combineDtDcpt1(jointDtC1C2C3C4,
jointDcptC1C2C3C4)
symbolDpmfC1C2C3C4 = gmtk.readDpmf(dpmfs,
"%sGivenC1C2C3C4" % symbolName)
symbolGivenC1C2C3C4 = gmtk.combineDtDcpt2(jointDtC1C2C3C4,
symbolDpmfC1C2C3C4)
jointProbSymbolC1C2C3C4 = symbolGivenC1C2C3C4.multiple(
[1, 2, 3, 4], jointProbC1C2C3C4)
gmtk.saveDpmfsProbs(outDir, "%sGivenC1C2C3C4" % symbolName,
len(symbolGivenC1C2C3C4.vectSubList([1, 2, 3, 4])),
symbolCard, symbolGivenC1C2C3C4)
jointProbSymbolC1C2C3 = jointProbSymbolC1C2C3C4.marginalize(
[0, 1, 2, 3])
symbolGivenC1C2C3 = jointProbSymbolC1C2C3.conditionalize([1, 2, 3])
gmtk.saveDpmfsProbs(outDir, "%sGivenC1C2C3" % symbolName,
len(symbolGivenC1C2C3.vectSubList([1, 2, 3])),
symbolCard, symbolGivenC1C2C3)
jointProbSymbolC1C2 = jointProbSymbolC1C2C3.marginalize([0, 1, 2])
symbolGivenC1C2 = jointProbSymbolC1C2.conditionalize([1, 2])
gmtk.saveDpmfsProbs(outDir, "%sGivenC1C2" % symbolName,
len(symbolGivenC1C2.vectSubList([1, 2])),
symbolCard, symbolGivenC1C2)
jointProbSymbolC1 = jointProbSymbolC1C2.marginalize([0, 1])
symbolGivenC1 = jointProbSymbolC1.conditionalize([1])
# in case of conditioning by _SINK_ I have to enable to generate _sink_ word only
# otherwise I would see _SINK_ concept in the stack
# TODO: Turn it into validator
symbolGivenC1.setValue([_sink_, _SINK_], 1)
gmtk.saveDcptBigram(outDir, "%sGivenC1" % symbolName, symbolCard,
conceptCard, symbolGivenC1)
symbolUnigram = jointProbSymbolC1.marginalize([0])
# I need to enable to decode _unseen_ word only ! so set the probability of
# generating _empty_ to zero
symbolUnigram.setValue([int(symbolMap["_empty_"])], 0)
# normalize sum of probabilities to one
symbolUnigram = symbolUnigram.normJoint()
gmtk.saveDcptUnigram(outDir, "%sUnigram" % symbolName, symbolCard,
symbolUnigram)
gmtk.saveDcptUnseen(outDir, "%sZerogram" % symbolName, symbolCard,
symbolMap)
##############################################################################
# save trigrams
jointProbPC1C2 = jointProbPC1C2C3.marginalize([0, 1, 2])
popGivenC1C2 = jointProbPC1C2.conditionalize([1, 2])
popGivenC1C2 = popGivenC1C2.insertPenalty(0, penalty, popCard)
gmtk.saveDpmfsProbs(dirOut, "popGivenC1C2", len(popGivenC1C2.vectSubList([1, 2])), popCard, popGivenC1C2, -penalty)
##############################################################################
# save bigrams as CPT
jointProbPC1 = jointProbPC1C2.marginalize([0, 1])
popGivenC1 = jointProbPC1.conditionalize([1])
popGivenC1 = popGivenC1.insertPenalty(0, penalty, popCard)
# in case of conditioning by _SINK_, I have to enable to delete all stack
popGivenC1.setValue([4, int(conceptMap["_SINK_"])], 1)
gmtk.saveDcptBigram(dirOut, "popGivenC1", popCard, conceptCard, popGivenC1)
##############################################################################
# save unigrams as CPT
popUnigram = jointProbPC1.marginalize([0])
popUnigram = popUnigram.insertPenalty(0, penalty, popCard)
gmtk.saveDcptUnigram(dirOut, "popUnigram", popCard, popUnigram)
if verbose:
print("-------------------------------------------------")
print("Finish")
# save trigrams
jointProbPC1C2 = jointProbPC1C2C3.marginalize([0, 1, 2])
popGivenC1C2 = jointProbPC1C2.conditionalize([1, 2])
popGivenC1C2 = popGivenC1C2.insertPenalty(0, penalty, popCard)
gmtk.saveDpmfsProbs(dirOut, "popGivenC1C2",
len(popGivenC1C2.vectSubList([1, 2])), popCard,
popGivenC1C2, -penalty)
##############################################################################
# save bigrams as CPT
jointProbPC1 = jointProbPC1C2.marginalize([0, 1])
popGivenC1 = jointProbPC1.conditionalize([1])
popGivenC1 = popGivenC1.insertPenalty(0, penalty, popCard)
# in case of conditioning by _SINK_, I have to enable to delete all stack
popGivenC1.setValue([4, int(conceptMap["_SINK_"])], 1)
gmtk.saveDcptBigram(dirOut, "popGivenC1", popCard, conceptCard, popGivenC1)
##############################################################################
# save unigrams as CPT
popUnigram = jointProbPC1.marginalize([0])
popUnigram = popUnigram.insertPenalty(0, penalty, popCard)
gmtk.saveDcptUnigram(dirOut, "popUnigram", popCard, popUnigram)
if verbose:
print("-------------------------------------------------")
print("Finish")
##############################################################################
# save trigrams
jointProbPC1C2 = jointProbPC1C2C3.marginalize([0, 1, 2])
pushGivenC1C2 = jointProbPC1C2.conditionalize([1, 2])
pushGivenC1C2 = pushGivenC1C2.insertPenalty(0, penalty, pushCard)
gmtk.saveDpmfsProbs(dirOut, "pushGivenC1C2", len(pushGivenC1C2.vectSubList([1, 2])), pushCard, pushGivenC1C2, -penalty)
##############################################################################
# save bigrams as CPT
jointProbPC1 = jointProbPC1C2.marginalize([0, 1])
pushGivenC1 = jointProbPC1.conditionalize([1])
pushGivenC1 = pushGivenC1.insertPenalty(0, penalty, pushCard)
# in case of conditioning by _SINK_, I have to enable to delete all stack
pushGivenC1.setValue([4, int(conceptMap["_SINK_"])], 1)
gmtk.saveDcptBigram(dirOut, "pushGivenC1", pushCard, conceptCard, pushGivenC1)
##############################################################################
# save unigrams as CPT
pushUnigram = jointProbPC1.marginalize([0])
pushUnigram = pushUnigram.insertPenalty(0, penalty, pushCard)
gmtk.saveDcptUnigram(dirOut, "pushUnigram", pushCard, pushUnigram)
if verbose:
print("-------------------------------------------------")
print("Finish")
##
##gmtk.saveDpmfsProbs(dirOut, "concept1GivenC2C3C4", len(concept1GivenC2C3C4.vectSubList([1, 2, 3])), conceptCard, concept1GivenC2C3C4)
##gmtk.saveDpmfsProbs(dirOut, "concept1GivenC2C3C4X", len(concept1GivenC2C3C4X.vectSubList([1, 2, 3])), conceptCard, concept1GivenC2C3C4X)
##############################################################################
# save bigrams as CPT
jointProbC2C3 = jointProbC2C3C4.marginalize([0, 1])
concept2GivenC3 = jointProbC2C3.conditionalize([1])
# in case of conditioning by _SINK_ I have to enable to generate _SINK_ concept only
# otherwise I would see _SINK_ concept in the stack
concept2GivenC3.setValue([int(conceptMap["_SINK_"]), int(conceptMap["_SINK_"])], 1)
# do the same for _DUMMY_
concept2GivenC3.setValue([int(conceptMap["_SINK_"]), int(conceptMap["_DUMMY_"])], 1)
gmtk.saveDcptBigram(dirOut, "concept2GivenC3", conceptCard, conceptCard, concept2GivenC3)
##############################################################################
# save unigrams as CPT
concept2Unigram = jointProbC2C3.marginalize([0])
gmtk.saveDcptUnigram(dirOut, "concept2Unigram", conceptCard, concept2Unigram)
################################################################################
### save zerograms as CPT
##gmtk.saveDcptZerogram(dirOut, "concept1Zerogram", conceptCard)
if verbose:
print("-------------------------------------------------")
print("Finish")
##############################################################################
# save bigrams as CPT
jointProbC2C3 = jointProbC2C3C4.marginalize([0, 1])
concept2GivenC3 = jointProbC2C3.conditionalize([1])
# in case of conditioning by _SINK_ I have to enable to generate _SINK_ concept only
# otherwise I would see _SINK_ concept in the stack
concept2GivenC3.setValue(
[int(conceptMap["_SINK_"]),
int(conceptMap["_SINK_"])], 1)
# do the same for _DUMMY_
concept2GivenC3.setValue(
[int(conceptMap["_SINK_"]),
int(conceptMap["_DUMMY_"])], 1)
gmtk.saveDcptBigram(dirOut, "concept2GivenC3", conceptCard, conceptCard,
concept2GivenC3)
##############################################################################
# save unigrams as CPT
concept2Unigram = jointProbC2C3.marginalize([0])
gmtk.saveDcptUnigram(dirOut, "concept2Unigram", conceptCard, concept2Unigram)
################################################################################
### save zerograms as CPT
##gmtk.saveDcptZerogram(dirOut, "concept1Zerogram", conceptCard)
if verbose:
print("-------------------------------------------------")
print("Finish")