84 lines
4.1 KiB
Makefile
84 lines
4.1 KiB
Makefile
# TODO: is there some way to make all of the targets use the same command?
|
|
|
|
ARABIC_TEST = format=TREES,/u/nlp/data/lexparser/trees/Arabic/2-Unvoc-Test.utf8.txt
|
|
|
|
CHINESE_TEST = format=TREES,/u/nlp/data/chinese/ctb7/test.mrg
|
|
|
|
# Use dev set for regression testing
|
|
ENGLISH_TEST = /u/nlp/data/pos-tagger/english/test-wsj-19-21
|
|
|
|
FRENCH_TEST = format=TREES,/u/nlp/data/lexparser/trees/FrenchCC/FTB-Test.utf8.txt
|
|
|
|
GERMAN_TEST = format=TREES,trf=edu.stanford.nlp.trees.international.negra.NegraPennTreeReaderFactory,/u/nlp/data/GermanACL08/negra/negra_3.mrg
|
|
|
|
SPANISH_TEST = format=TREES,/u/nlp/data/spanish/ancora/ancora.test,/u/nlp/data/spanish/ldc/ldc-NW.test,/u/nlp/data/spanish/ldc/ldc-DF.test
|
|
|
|
.SECONDEXPANSION:
|
|
|
|
all: arabic chinese english french german spanish testing wsj
|
|
.PHONY: all arabic chinese english french german spanish testing wsj
|
|
|
|
arabic: arabic.tagger arabic-train.tagger
|
|
|
|
# we release an arabic model trained on everything, with a
|
|
# corresponding model on train only for testing purposes
|
|
arabic.tagger arabic-train.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(ARABIC_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(ARABIC_TEST) -verboseResults false >> $@.out 2>&1
|
|
|
|
chinese: chinese-distsim.tagger chinese-nodistsim.tagger
|
|
|
|
chinese-nodistsim.tagger chinese-distsim.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(CHINESE_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(CHINESE_TEST) -verboseResults false >> $@.out 2>&1
|
|
|
|
english: english-bidirectional-distsim.tagger english-caseless-left3words-distsim.tagger english-left3words-distsim.tagger
|
|
|
|
english-bidirectional-distsim.tagger english-caseless-left3words-distsim.tagger english-left3words-distsim.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(ENGLISH_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
java -mx2g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(ENGLISH_TEST) -verboseResults false >> $@.out 2>&1
|
|
|
|
french: french.tagger
|
|
|
|
french.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(FRENCH_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(FRENCH_TEST) -verboseResults false >> $@.out 2>&1
|
|
|
|
german: german-dewac.tagger german-fast.tagger german-fast-caseless.tagger german-hgc.tagger
|
|
|
|
german-dewac.tagger german-fast.tagger german-fast-caseless.tagger german-hgc.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(GERMAN_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(GERMAN_TEST) -verboseResults false >> $@.out 2>&1
|
|
|
|
spanish: spanish.tagger spanish-distsim.tagger
|
|
|
|
spanish.tagger spanish-distsim.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(SPANISH_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
# java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(SPANISH_TEST) -verboseResults false >> $@.out 2>&1
|
|
|
|
testing: testing.tagger
|
|
|
|
testing.tagger:
|
|
@echo Training $@
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
|
|
wsj: wsj-0-18-bidirectional-distsim.tagger wsj-0-18-bidirectional-nodistsim.tagger wsj-0-18-caseless-left3words-distsim.tagger wsj-0-18-left3words-distsim.tagger wsj-0-18-left3words-nodistsim.tagger
|
|
|
|
wsj-0-18-bidirectional-distsim.tagger wsj-0-18-bidirectional-nodistsim.tagger wsj-0-18-caseless-left3words-distsim.tagger wsj-0-18-left3words-distsim.tagger wsj-0-18-left3words-nodistsim.tagger: $$@.props
|
|
@echo Training $@
|
|
@echo Will test on $(ENGLISH_TEST)
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -props $@.props > $@.out 2>&1
|
|
java -mx6g edu.stanford.nlp.tagger.maxent.MaxentTagger -model $@ -testFile $(ENGLISH_TEST) -verboseResults false >> $@.out 2>&1
|