-
Notifications
You must be signed in to change notification settings - Fork 12
/
Copy pathrun_snli
executable file
·212 lines (194 loc) · 6.23 KB
/
run_snli
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
#!/bin/bash
exec scala -cp qry.jar:test/src/test_client.jar:src/naturalli_client.jar "$0" "$@"
!#
import Qry._
val SNLI_TRAIN = "etc/snli/snli_train.tab"
val SNLI_DEV = "etc/snli/snli_dev.tab"
val SNLI_TEST = "etc/snli/snli_test.tab"
val SICK_TRAIN = "etc/sick/sick_train.tab"
val SICK_SNLI_TRAIN = "etc/sick/sick+snli_train.tab"
val SICK_TEST = "etc/sick/sick_test.tab"
val RTE_TRAIN = "etc/RTE-3/English_dev.tab"
val RTE_SNLI_TRAIN = "etc/sick/rte+snli_train.tab"
val RTE_TEST = "etc/RTE-3/English_dev.tab"
//
// Define the program
//
def program(memoryMB:Int):List[String] = ("java"
-('cp, cp)
// -"Xrunhprof:cpu=samples,depth=25"
-("mx" + memoryMB + "M")
-"Dwordnet.database.dir=etc/WordNet-3.1/dict"
-'server
-'ea
->"edu.stanford.nlp.naturalli.entail.ClassifierTrainer"
-("classifier", "SIMPLE")
-("naturalli_search", "/dev/null")
-("test.errors", touch("errors.txt"))
).toList
using("runs/")
submit(program(24000)
-("train.file", SNLI_TRAIN)
-("train.cache.do", "false")
-("test.file", SNLI_DEV)
-("test.cache.do", "false")
-("train.count", "1000000")
-("train.regularizer", "l1")
-("train.sigma", "1.5" )
-("train.align.maxlength", "4")
-("train.align.mincount", "0")
-("vocab.threshold", "0")
-("model", touch("model.ser.gz"))
// -("features", // all the features implemented
// """[
// BLEU LENGTH_DIFF OVERLAP POS_OVERLAP
// ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ENTAIL_BOTH_GRAM
// KEYWORD_OVERLAP
// DEPTREE_ROOT_STRUCTURE ENTAIL_DEPTREE
// MT_UNIGRAM MT_PHRASE
// ]"""
// )
-("features",
"""[
BLEU OVERLAP POS_OVERLAP
ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ENTAIL_BOTH_GRAM
]"""
)
-("features.nolex", "false")
-("parallel", "true")
// -("threads", "1")
-("log.file", touch("redwood.log"))
)
////
//// Run the learning curve
////
//parallel(8) submit(program(8500)
// -("train.file", SNLI_TRAIN)
// -("train.cache.do", "false")
// -("test.file", SNLI_TEST)
// -("test.cache.do", "false")
//
// -("train.count", "10" & "100" & "1000" & "10000" & "50000" & "100000" & "250000" & "1000000")
// -("model", touch("model.ser.gz"))
// -("features", """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ]""" )
// -("features.nolex", "true" & "false")
//
// -("parallel", "false")
//
// -("log.file", touch("redwood.log"))
//)
//
////
//// Run the unigram learning curve
////
//parallel(8) submit(program(8500)
// -("train.file", SNLI_TRAIN)
// -("train.cache.do", "false")
// -("test.file", SNLI_TEST)
// -("test.cache.do", "false")
//
// -("train.count", "10" & "100" & "1000" & "10000" & "50000" & "100000" & "250000" & "1000000")
// -("model", touch("model.ser.gz"))
// -("features", """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ]""" )
// -("features.nolex", "false")
//
// -("parallel", "false")
//
// -("log.file", touch("redwood.log"))
//)
////
//// Try out other features
////
//parallel(2) submit(program(32000)
// -("train.file", SNLI_TRAIN)
// -("train.cache.do", "false")
// -("test.file", SNLI_TEST)
// -("test.cache.do", "false")
//
// -("train.count", "1000000")
// -("model", touch("model.ser.gz"))
// -("features",
//// """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ENTAIL_BOTH_GRAM ]""" &
//// """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ENTAIL_TRIGRAM ]""" &
//// """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ENTAIL_TRIGRAM ENTAIL_BOTH_GRAM ]""" &
// """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ]"""
// )
// -("features.nolex", "false")
// -("parallel", "false")
// -("log.file", touch("redwood.log"))
//)
////
//// Run ablation studies with keyword features
////
//parallel(1) submit(program(70000)
// -("train.file", SNLI_TRAIN)
// -("train.cache.do", "false")
// -("test.file", SNLI_TEST)
// -("test.cache.do", "false")
//
// -("train.count", "10" & "100" & "1000" & "10000" & "50000" & "100000" & "250000" & "1000000")
// -("model", touch("model.ser.gz"))
// -("features", """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM KEYWORD_OVERLAP ENTAIL_KEYWORD ]""" )
// -("features.nolex", "true" & "false")
//
// -("parallel", "false")
//
// -("log.file", touch("redwood.log"))
//)
//
////
//// Run the SICK data
////
//parallel(8) submit(program(8500)
// -("train.file", SICK_TRAIN & SNLI_TRAIN & SICK_SNLI_TRAIN)
// -("train.cache.do", "false")
// -("test.file", SICK_TEST)
// -("test.cache.do", "false")
//
// -("train.count", "10000")
// -("model", touch("model.ser.gz"))
// -("features", """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ]""" )
// -("features.nolex", "true" & "false")
//
// -("parallel", "false")
//
// -("log.file", touch("redwood.log"))
//)
//
////
//// Run the RTE data
////
//parallel(8) submit(program(8500)
// -("train.file", RTE_TRAIN & SNLI_TRAIN & RTE_SNLI_TRAIN)
// -("train.cache.do", "false")
// -("test.file", SICK_TEST)
// -("test.cache.do", "false")
//
// -("train.count", "10000")
// -("model", touch("model.ser.gz"))
// -("features", """[ BLEU LENGTH_DIFF OVERLAP POS_OVERLAP ENTAIL_UNIGRAM ENTAIL_BIGRAM CONCLUSION_NGRAM ]""" )
// -("features.nolex", "true" & "false")
//
// -("parallel", "false")
//
// -("log.file", touch("redwood.log"))
//)
def cp:String = {
val JAVANLP = List(
System.getenv("JAVANLP_HOME") + "/projects/core/classes",
System.getenv("JAVANLP_HOME") + "/projects/more/classes",
"stanford-corenlp-models-current.jar"
// "/u/nlp/data/StanfordCoreNLPModels/stanford-corenlp-models-current.jar",
// "/u/nlp/data/StanfordCoreNLPModels/stanford-corenlp-caseless-models-current.jar"
).mkString(":")
val CUSTOM = List(
"lib/corenlp-scala.jar",
"lib/trove.jar",
"lib/jaws.jar",
"lib/scripts/sim-1.0.jar",
"lib/gson-2.3.1.jar",
"lib/corenlp/protobuf.jar",
"lib/berkeleyaligner.jar"
).mkString(":")
List("src/naturalli_preprocess.jar", JAVANLP, CUSTOM).mkString(":")
}