evalscope 0.9.0__py3-none-any.whl → 0.10.1__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Potentially problematic release.
This version of evalscope might be problematic. Click here for more details.
- evalscope/arguments.py +1 -0
- evalscope/benchmarks/arc/arc_adapter.py +3 -5
- evalscope/benchmarks/bbh/bbh_adapter.py +3 -3
- evalscope/benchmarks/benchmark.py +1 -1
- evalscope/benchmarks/ceval/ceval_adapter.py +5 -82
- evalscope/benchmarks/cmmlu/cmmlu_adapter.py +5 -79
- evalscope/benchmarks/competition_math/competition_math_adapter.py +4 -4
- evalscope/benchmarks/data_adapter.py +69 -70
- evalscope/benchmarks/general_qa/general_qa_adapter.py +10 -63
- evalscope/benchmarks/gpqa/__init__.py +0 -0
- evalscope/benchmarks/gpqa/chain_of_thought.txt +81 -0
- evalscope/benchmarks/gpqa/gpqa_adapter.py +103 -0
- evalscope/benchmarks/gsm8k/gsm8k_adapter.py +4 -5
- evalscope/benchmarks/hellaswag/hellaswag_adapter.py +12 -6
- evalscope/benchmarks/humaneval/humaneval_adapter.py +3 -4
- evalscope/benchmarks/ifeval/__init__.py +0 -0
- evalscope/benchmarks/ifeval/ifeval_adapter.py +56 -0
- evalscope/benchmarks/ifeval/instructions.py +1477 -0
- evalscope/benchmarks/ifeval/instructions_registry.py +188 -0
- evalscope/benchmarks/ifeval/instructions_util.py +1670 -0
- evalscope/benchmarks/ifeval/utils.py +134 -0
- evalscope/benchmarks/iquiz/__init__.py +0 -0
- evalscope/benchmarks/iquiz/iquiz_adapter.py +63 -0
- evalscope/benchmarks/mmlu/mmlu_adapter.py +8 -84
- evalscope/benchmarks/mmlu_pro/mmlu_pro_adapter.py +2 -2
- evalscope/benchmarks/race/race_adapter.py +4 -73
- evalscope/benchmarks/trivia_qa/trivia_qa_adapter.py +3 -6
- evalscope/benchmarks/truthful_qa/truthful_qa_adapter.py +8 -57
- evalscope/cli/cli.py +2 -0
- evalscope/cli/start_app.py +30 -0
- evalscope/collections/evaluator.py +82 -62
- evalscope/collections/sampler.py +47 -41
- evalscope/collections/schema.py +14 -10
- evalscope/constants.py +4 -0
- evalscope/evaluator/evaluator.py +22 -13
- evalscope/metrics/__init__.py +2 -5
- evalscope/metrics/metrics.py +11 -2
- evalscope/metrics/named_metrics.py +17 -0
- evalscope/models/chat_adapter.py +2 -0
- evalscope/models/server_adapter.py +11 -4
- evalscope/perf/__init__.py +1 -0
- evalscope/perf/main.py +0 -1
- evalscope/perf/plugin/api/custom_api.py +1 -1
- evalscope/perf/plugin/api/openai_api.py +1 -1
- evalscope/perf/plugin/datasets/flickr8k.py +1 -1
- evalscope/perf/plugin/datasets/longalpaca.py +1 -1
- evalscope/report/__init__.py +5 -0
- evalscope/report/app.py +693 -0
- evalscope/report/combinator.py +73 -0
- evalscope/report/generator.py +80 -0
- evalscope/report/utils.py +133 -0
- evalscope/run.py +16 -11
- evalscope/summarizer.py +1 -1
- evalscope/utils/chat_service.py +1 -1
- evalscope/utils/logger.py +1 -0
- evalscope/utils/model_utils.py +5 -2
- evalscope/version.py +2 -2
- {evalscope-0.9.0.dist-info → evalscope-0.10.1.dist-info}/METADATA +84 -7
- {evalscope-0.9.0.dist-info → evalscope-0.10.1.dist-info}/RECORD +66 -51
- tests/cli/test_collection.py +11 -7
- tests/cli/test_run.py +13 -4
- evalscope/tools/__init__.py +0 -1
- evalscope/tools/combine_reports.py +0 -133
- evalscope/tools/gen_mmlu_subject_mapping.py +0 -90
- /evalscope/{tools/rewrite_eval_results.py → models/custom/dummy_model.py} +0 -0
- {evalscope-0.9.0.dist-info → evalscope-0.10.1.dist-info}/LICENSE +0 -0
- {evalscope-0.9.0.dist-info → evalscope-0.10.1.dist-info}/WHEEL +0 -0
- {evalscope-0.9.0.dist-info → evalscope-0.10.1.dist-info}/entry_points.txt +0 -0
- {evalscope-0.9.0.dist-info → evalscope-0.10.1.dist-info}/top_level.txt +0 -0
|
@@ -0,0 +1,1670 @@
|
|
|
1
|
+
# Copyright 2023 The Google Research Authors.
|
|
2
|
+
#
|
|
3
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
4
|
+
# you may not use this file except in compliance with the License.
|
|
5
|
+
# You may obtain a copy of the License at
|
|
6
|
+
#
|
|
7
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
|
8
|
+
#
|
|
9
|
+
# Unless required by applicable law or agreed to in writing, software
|
|
10
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
11
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
12
|
+
# See the License for the specific language governing permissions and
|
|
13
|
+
# limitations under the License.
|
|
14
|
+
"""Utility library of instructions."""
|
|
15
|
+
|
|
16
|
+
import functools
|
|
17
|
+
import immutabledict
|
|
18
|
+
import nltk
|
|
19
|
+
import os
|
|
20
|
+
import random
|
|
21
|
+
import re
|
|
22
|
+
|
|
23
|
+
RANK = os.environ.get('LOCAL_RANK', '0')
|
|
24
|
+
|
|
25
|
+
WORD_LIST = [
|
|
26
|
+
'western',
|
|
27
|
+
'sentence',
|
|
28
|
+
'signal',
|
|
29
|
+
'dump',
|
|
30
|
+
'spot',
|
|
31
|
+
'opposite',
|
|
32
|
+
'bottom',
|
|
33
|
+
'potato',
|
|
34
|
+
'administration',
|
|
35
|
+
'working',
|
|
36
|
+
'welcome',
|
|
37
|
+
'morning',
|
|
38
|
+
'good',
|
|
39
|
+
'agency',
|
|
40
|
+
'primary',
|
|
41
|
+
'wish',
|
|
42
|
+
'responsibility',
|
|
43
|
+
'press',
|
|
44
|
+
'problem',
|
|
45
|
+
'president',
|
|
46
|
+
'steal',
|
|
47
|
+
'brush',
|
|
48
|
+
'read',
|
|
49
|
+
'type',
|
|
50
|
+
'beat',
|
|
51
|
+
'trainer',
|
|
52
|
+
'growth',
|
|
53
|
+
'lock',
|
|
54
|
+
'bone',
|
|
55
|
+
'case',
|
|
56
|
+
'equal',
|
|
57
|
+
'comfortable',
|
|
58
|
+
'region',
|
|
59
|
+
'replacement',
|
|
60
|
+
'performance',
|
|
61
|
+
'mate',
|
|
62
|
+
'walk',
|
|
63
|
+
'medicine',
|
|
64
|
+
'film',
|
|
65
|
+
'thing',
|
|
66
|
+
'rock',
|
|
67
|
+
'tap',
|
|
68
|
+
'total',
|
|
69
|
+
'competition',
|
|
70
|
+
'ease',
|
|
71
|
+
'south',
|
|
72
|
+
'establishment',
|
|
73
|
+
'gather',
|
|
74
|
+
'parking',
|
|
75
|
+
'world',
|
|
76
|
+
'plenty',
|
|
77
|
+
'breath',
|
|
78
|
+
'claim',
|
|
79
|
+
'alcohol',
|
|
80
|
+
'trade',
|
|
81
|
+
'dear',
|
|
82
|
+
'highlight',
|
|
83
|
+
'street',
|
|
84
|
+
'matter',
|
|
85
|
+
'decision',
|
|
86
|
+
'mess',
|
|
87
|
+
'agreement',
|
|
88
|
+
'studio',
|
|
89
|
+
'coach',
|
|
90
|
+
'assist',
|
|
91
|
+
'brain',
|
|
92
|
+
'wing',
|
|
93
|
+
'style',
|
|
94
|
+
'private',
|
|
95
|
+
'top',
|
|
96
|
+
'brown',
|
|
97
|
+
'leg',
|
|
98
|
+
'buy',
|
|
99
|
+
'procedure',
|
|
100
|
+
'method',
|
|
101
|
+
'speed',
|
|
102
|
+
'high',
|
|
103
|
+
'company',
|
|
104
|
+
'valuable',
|
|
105
|
+
'pie',
|
|
106
|
+
'analyst',
|
|
107
|
+
'session',
|
|
108
|
+
'pattern',
|
|
109
|
+
'district',
|
|
110
|
+
'pleasure',
|
|
111
|
+
'dinner',
|
|
112
|
+
'swimming',
|
|
113
|
+
'joke',
|
|
114
|
+
'order',
|
|
115
|
+
'plate',
|
|
116
|
+
'department',
|
|
117
|
+
'motor',
|
|
118
|
+
'cell',
|
|
119
|
+
'spend',
|
|
120
|
+
'cabinet',
|
|
121
|
+
'difference',
|
|
122
|
+
'power',
|
|
123
|
+
'examination',
|
|
124
|
+
'engine',
|
|
125
|
+
'horse',
|
|
126
|
+
'dimension',
|
|
127
|
+
'pay',
|
|
128
|
+
'toe',
|
|
129
|
+
'curve',
|
|
130
|
+
'literature',
|
|
131
|
+
'bother',
|
|
132
|
+
'fire',
|
|
133
|
+
'possibility',
|
|
134
|
+
'debate',
|
|
135
|
+
'activity',
|
|
136
|
+
'passage',
|
|
137
|
+
'hello',
|
|
138
|
+
'cycle',
|
|
139
|
+
'background',
|
|
140
|
+
'quiet',
|
|
141
|
+
'author',
|
|
142
|
+
'effect',
|
|
143
|
+
'actor',
|
|
144
|
+
'page',
|
|
145
|
+
'bicycle',
|
|
146
|
+
'error',
|
|
147
|
+
'throat',
|
|
148
|
+
'attack',
|
|
149
|
+
'character',
|
|
150
|
+
'phone',
|
|
151
|
+
'tea',
|
|
152
|
+
'increase',
|
|
153
|
+
'outcome',
|
|
154
|
+
'file',
|
|
155
|
+
'specific',
|
|
156
|
+
'inspector',
|
|
157
|
+
'internal',
|
|
158
|
+
'potential',
|
|
159
|
+
'staff',
|
|
160
|
+
'building',
|
|
161
|
+
'employer',
|
|
162
|
+
'shoe',
|
|
163
|
+
'hand',
|
|
164
|
+
'direction',
|
|
165
|
+
'garden',
|
|
166
|
+
'purchase',
|
|
167
|
+
'interview',
|
|
168
|
+
'study',
|
|
169
|
+
'recognition',
|
|
170
|
+
'member',
|
|
171
|
+
'spiritual',
|
|
172
|
+
'oven',
|
|
173
|
+
'sandwich',
|
|
174
|
+
'weird',
|
|
175
|
+
'passenger',
|
|
176
|
+
'particular',
|
|
177
|
+
'response',
|
|
178
|
+
'reaction',
|
|
179
|
+
'size',
|
|
180
|
+
'variation',
|
|
181
|
+
'a',
|
|
182
|
+
'cancel',
|
|
183
|
+
'candy',
|
|
184
|
+
'exit',
|
|
185
|
+
'guest',
|
|
186
|
+
'condition',
|
|
187
|
+
'fly',
|
|
188
|
+
'price',
|
|
189
|
+
'weakness',
|
|
190
|
+
'convert',
|
|
191
|
+
'hotel',
|
|
192
|
+
'great',
|
|
193
|
+
'mouth',
|
|
194
|
+
'mind',
|
|
195
|
+
'song',
|
|
196
|
+
'sugar',
|
|
197
|
+
'suspect',
|
|
198
|
+
'telephone',
|
|
199
|
+
'ear',
|
|
200
|
+
'roof',
|
|
201
|
+
'paint',
|
|
202
|
+
'refrigerator',
|
|
203
|
+
'organization',
|
|
204
|
+
'jury',
|
|
205
|
+
'reward',
|
|
206
|
+
'engineering',
|
|
207
|
+
'day',
|
|
208
|
+
'possession',
|
|
209
|
+
'crew',
|
|
210
|
+
'bar',
|
|
211
|
+
'road',
|
|
212
|
+
'description',
|
|
213
|
+
'celebration',
|
|
214
|
+
'score',
|
|
215
|
+
'mark',
|
|
216
|
+
'letter',
|
|
217
|
+
'shower',
|
|
218
|
+
'suggestion',
|
|
219
|
+
'sir',
|
|
220
|
+
'luck',
|
|
221
|
+
'national',
|
|
222
|
+
'progress',
|
|
223
|
+
'hall',
|
|
224
|
+
'stroke',
|
|
225
|
+
'theory',
|
|
226
|
+
'offer',
|
|
227
|
+
'story',
|
|
228
|
+
'tax',
|
|
229
|
+
'definition',
|
|
230
|
+
'history',
|
|
231
|
+
'ride',
|
|
232
|
+
'medium',
|
|
233
|
+
'opening',
|
|
234
|
+
'glass',
|
|
235
|
+
'elevator',
|
|
236
|
+
'stomach',
|
|
237
|
+
'question',
|
|
238
|
+
'ability',
|
|
239
|
+
'leading',
|
|
240
|
+
'village',
|
|
241
|
+
'computer',
|
|
242
|
+
'city',
|
|
243
|
+
'grand',
|
|
244
|
+
'confidence',
|
|
245
|
+
'candle',
|
|
246
|
+
'priest',
|
|
247
|
+
'recommendation',
|
|
248
|
+
'point',
|
|
249
|
+
'necessary',
|
|
250
|
+
'body',
|
|
251
|
+
'desk',
|
|
252
|
+
'secret',
|
|
253
|
+
'horror',
|
|
254
|
+
'noise',
|
|
255
|
+
'culture',
|
|
256
|
+
'warning',
|
|
257
|
+
'water',
|
|
258
|
+
'round',
|
|
259
|
+
'diet',
|
|
260
|
+
'flower',
|
|
261
|
+
'bus',
|
|
262
|
+
'tough',
|
|
263
|
+
'permission',
|
|
264
|
+
'week',
|
|
265
|
+
'prompt',
|
|
266
|
+
'connection',
|
|
267
|
+
'abuse',
|
|
268
|
+
'height',
|
|
269
|
+
'save',
|
|
270
|
+
'corner',
|
|
271
|
+
'border',
|
|
272
|
+
'stress',
|
|
273
|
+
'drive',
|
|
274
|
+
'stop',
|
|
275
|
+
'rip',
|
|
276
|
+
'meal',
|
|
277
|
+
'listen',
|
|
278
|
+
'confusion',
|
|
279
|
+
'girlfriend',
|
|
280
|
+
'living',
|
|
281
|
+
'relation',
|
|
282
|
+
'significance',
|
|
283
|
+
'plan',
|
|
284
|
+
'creative',
|
|
285
|
+
'atmosphere',
|
|
286
|
+
'blame',
|
|
287
|
+
'invite',
|
|
288
|
+
'housing',
|
|
289
|
+
'paper',
|
|
290
|
+
'drink',
|
|
291
|
+
'roll',
|
|
292
|
+
'silver',
|
|
293
|
+
'drunk',
|
|
294
|
+
'age',
|
|
295
|
+
'damage',
|
|
296
|
+
'smoke',
|
|
297
|
+
'environment',
|
|
298
|
+
'pack',
|
|
299
|
+
'savings',
|
|
300
|
+
'influence',
|
|
301
|
+
'tourist',
|
|
302
|
+
'rain',
|
|
303
|
+
'post',
|
|
304
|
+
'sign',
|
|
305
|
+
'grandmother',
|
|
306
|
+
'run',
|
|
307
|
+
'profit',
|
|
308
|
+
'push',
|
|
309
|
+
'clerk',
|
|
310
|
+
'final',
|
|
311
|
+
'wine',
|
|
312
|
+
'swim',
|
|
313
|
+
'pause',
|
|
314
|
+
'stuff',
|
|
315
|
+
'singer',
|
|
316
|
+
'funeral',
|
|
317
|
+
'average',
|
|
318
|
+
'source',
|
|
319
|
+
'scene',
|
|
320
|
+
'tradition',
|
|
321
|
+
'personal',
|
|
322
|
+
'snow',
|
|
323
|
+
'nobody',
|
|
324
|
+
'distance',
|
|
325
|
+
'sort',
|
|
326
|
+
'sensitive',
|
|
327
|
+
'animal',
|
|
328
|
+
'major',
|
|
329
|
+
'negotiation',
|
|
330
|
+
'click',
|
|
331
|
+
'mood',
|
|
332
|
+
'period',
|
|
333
|
+
'arrival',
|
|
334
|
+
'expression',
|
|
335
|
+
'holiday',
|
|
336
|
+
'repeat',
|
|
337
|
+
'dust',
|
|
338
|
+
'closet',
|
|
339
|
+
'gold',
|
|
340
|
+
'bad',
|
|
341
|
+
'sail',
|
|
342
|
+
'combination',
|
|
343
|
+
'clothes',
|
|
344
|
+
'emphasis',
|
|
345
|
+
'duty',
|
|
346
|
+
'black',
|
|
347
|
+
'step',
|
|
348
|
+
'school',
|
|
349
|
+
'jump',
|
|
350
|
+
'document',
|
|
351
|
+
'professional',
|
|
352
|
+
'lip',
|
|
353
|
+
'chemical',
|
|
354
|
+
'front',
|
|
355
|
+
'wake',
|
|
356
|
+
'while',
|
|
357
|
+
'inside',
|
|
358
|
+
'watch',
|
|
359
|
+
'row',
|
|
360
|
+
'subject',
|
|
361
|
+
'penalty',
|
|
362
|
+
'balance',
|
|
363
|
+
'possible',
|
|
364
|
+
'adult',
|
|
365
|
+
'aside',
|
|
366
|
+
'sample',
|
|
367
|
+
'appeal',
|
|
368
|
+
'wedding',
|
|
369
|
+
'depth',
|
|
370
|
+
'king',
|
|
371
|
+
'award',
|
|
372
|
+
'wife',
|
|
373
|
+
'blow',
|
|
374
|
+
'site',
|
|
375
|
+
'camp',
|
|
376
|
+
'music',
|
|
377
|
+
'safe',
|
|
378
|
+
'gift',
|
|
379
|
+
'fault',
|
|
380
|
+
'guess',
|
|
381
|
+
'act',
|
|
382
|
+
'shame',
|
|
383
|
+
'drama',
|
|
384
|
+
'capital',
|
|
385
|
+
'exam',
|
|
386
|
+
'stupid',
|
|
387
|
+
'record',
|
|
388
|
+
'sound',
|
|
389
|
+
'swing',
|
|
390
|
+
'novel',
|
|
391
|
+
'minimum',
|
|
392
|
+
'ratio',
|
|
393
|
+
'machine',
|
|
394
|
+
'shape',
|
|
395
|
+
'lead',
|
|
396
|
+
'operation',
|
|
397
|
+
'salary',
|
|
398
|
+
'cloud',
|
|
399
|
+
'affair',
|
|
400
|
+
'hit',
|
|
401
|
+
'chapter',
|
|
402
|
+
'stage',
|
|
403
|
+
'quantity',
|
|
404
|
+
'access',
|
|
405
|
+
'army',
|
|
406
|
+
'chain',
|
|
407
|
+
'traffic',
|
|
408
|
+
'kick',
|
|
409
|
+
'analysis',
|
|
410
|
+
'airport',
|
|
411
|
+
'time',
|
|
412
|
+
'vacation',
|
|
413
|
+
'philosophy',
|
|
414
|
+
'ball',
|
|
415
|
+
'chest',
|
|
416
|
+
'thanks',
|
|
417
|
+
'place',
|
|
418
|
+
'mountain',
|
|
419
|
+
'advertising',
|
|
420
|
+
'red',
|
|
421
|
+
'past',
|
|
422
|
+
'rent',
|
|
423
|
+
'return',
|
|
424
|
+
'tour',
|
|
425
|
+
'house',
|
|
426
|
+
'construction',
|
|
427
|
+
'net',
|
|
428
|
+
'native',
|
|
429
|
+
'war',
|
|
430
|
+
'figure',
|
|
431
|
+
'fee',
|
|
432
|
+
'spray',
|
|
433
|
+
'user',
|
|
434
|
+
'dirt',
|
|
435
|
+
'shot',
|
|
436
|
+
'task',
|
|
437
|
+
'stick',
|
|
438
|
+
'friend',
|
|
439
|
+
'software',
|
|
440
|
+
'promotion',
|
|
441
|
+
'interaction',
|
|
442
|
+
'surround',
|
|
443
|
+
'block',
|
|
444
|
+
'purpose',
|
|
445
|
+
'practice',
|
|
446
|
+
'conflict',
|
|
447
|
+
'routine',
|
|
448
|
+
'requirement',
|
|
449
|
+
'bonus',
|
|
450
|
+
'hole',
|
|
451
|
+
'state',
|
|
452
|
+
'junior',
|
|
453
|
+
'sweet',
|
|
454
|
+
'catch',
|
|
455
|
+
'tear',
|
|
456
|
+
'fold',
|
|
457
|
+
'wall',
|
|
458
|
+
'editor',
|
|
459
|
+
'life',
|
|
460
|
+
'position',
|
|
461
|
+
'pound',
|
|
462
|
+
'respect',
|
|
463
|
+
'bathroom',
|
|
464
|
+
'coat',
|
|
465
|
+
'script',
|
|
466
|
+
'job',
|
|
467
|
+
'teach',
|
|
468
|
+
'birth',
|
|
469
|
+
'view',
|
|
470
|
+
'resolve',
|
|
471
|
+
'theme',
|
|
472
|
+
'employee',
|
|
473
|
+
'doubt',
|
|
474
|
+
'market',
|
|
475
|
+
'education',
|
|
476
|
+
'serve',
|
|
477
|
+
'recover',
|
|
478
|
+
'tone',
|
|
479
|
+
'harm',
|
|
480
|
+
'miss',
|
|
481
|
+
'union',
|
|
482
|
+
'understanding',
|
|
483
|
+
'cow',
|
|
484
|
+
'river',
|
|
485
|
+
'association',
|
|
486
|
+
'concept',
|
|
487
|
+
'training',
|
|
488
|
+
'recipe',
|
|
489
|
+
'relationship',
|
|
490
|
+
'reserve',
|
|
491
|
+
'depression',
|
|
492
|
+
'proof',
|
|
493
|
+
'hair',
|
|
494
|
+
'revenue',
|
|
495
|
+
'independent',
|
|
496
|
+
'lift',
|
|
497
|
+
'assignment',
|
|
498
|
+
'temporary',
|
|
499
|
+
'amount',
|
|
500
|
+
'loss',
|
|
501
|
+
'edge',
|
|
502
|
+
'track',
|
|
503
|
+
'check',
|
|
504
|
+
'rope',
|
|
505
|
+
'estimate',
|
|
506
|
+
'pollution',
|
|
507
|
+
'stable',
|
|
508
|
+
'message',
|
|
509
|
+
'delivery',
|
|
510
|
+
'perspective',
|
|
511
|
+
'mirror',
|
|
512
|
+
'assistant',
|
|
513
|
+
'representative',
|
|
514
|
+
'witness',
|
|
515
|
+
'nature',
|
|
516
|
+
'judge',
|
|
517
|
+
'fruit',
|
|
518
|
+
'tip',
|
|
519
|
+
'devil',
|
|
520
|
+
'town',
|
|
521
|
+
'emergency',
|
|
522
|
+
'upper',
|
|
523
|
+
'drop',
|
|
524
|
+
'stay',
|
|
525
|
+
'human',
|
|
526
|
+
'neck',
|
|
527
|
+
'speaker',
|
|
528
|
+
'network',
|
|
529
|
+
'sing',
|
|
530
|
+
'resist',
|
|
531
|
+
'league',
|
|
532
|
+
'trip',
|
|
533
|
+
'signature',
|
|
534
|
+
'lawyer',
|
|
535
|
+
'importance',
|
|
536
|
+
'gas',
|
|
537
|
+
'choice',
|
|
538
|
+
'engineer',
|
|
539
|
+
'success',
|
|
540
|
+
'part',
|
|
541
|
+
'external',
|
|
542
|
+
'worker',
|
|
543
|
+
'simple',
|
|
544
|
+
'quarter',
|
|
545
|
+
'student',
|
|
546
|
+
'heart',
|
|
547
|
+
'pass',
|
|
548
|
+
'spite',
|
|
549
|
+
'shift',
|
|
550
|
+
'rough',
|
|
551
|
+
'lady',
|
|
552
|
+
'grass',
|
|
553
|
+
'community',
|
|
554
|
+
'garage',
|
|
555
|
+
'youth',
|
|
556
|
+
'standard',
|
|
557
|
+
'skirt',
|
|
558
|
+
'promise',
|
|
559
|
+
'blind',
|
|
560
|
+
'television',
|
|
561
|
+
'disease',
|
|
562
|
+
'commission',
|
|
563
|
+
'positive',
|
|
564
|
+
'energy',
|
|
565
|
+
'calm',
|
|
566
|
+
'presence',
|
|
567
|
+
'tune',
|
|
568
|
+
'basis',
|
|
569
|
+
'preference',
|
|
570
|
+
'head',
|
|
571
|
+
'common',
|
|
572
|
+
'cut',
|
|
573
|
+
'somewhere',
|
|
574
|
+
'presentation',
|
|
575
|
+
'current',
|
|
576
|
+
'thought',
|
|
577
|
+
'revolution',
|
|
578
|
+
'effort',
|
|
579
|
+
'master',
|
|
580
|
+
'implement',
|
|
581
|
+
'republic',
|
|
582
|
+
'floor',
|
|
583
|
+
'principle',
|
|
584
|
+
'stranger',
|
|
585
|
+
'shoulder',
|
|
586
|
+
'grade',
|
|
587
|
+
'button',
|
|
588
|
+
'tennis',
|
|
589
|
+
'police',
|
|
590
|
+
'collection',
|
|
591
|
+
'account',
|
|
592
|
+
'register',
|
|
593
|
+
'glove',
|
|
594
|
+
'divide',
|
|
595
|
+
'professor',
|
|
596
|
+
'chair',
|
|
597
|
+
'priority',
|
|
598
|
+
'combine',
|
|
599
|
+
'peace',
|
|
600
|
+
'extension',
|
|
601
|
+
'maybe',
|
|
602
|
+
'evening',
|
|
603
|
+
'frame',
|
|
604
|
+
'sister',
|
|
605
|
+
'wave',
|
|
606
|
+
'code',
|
|
607
|
+
'application',
|
|
608
|
+
'mouse',
|
|
609
|
+
'match',
|
|
610
|
+
'counter',
|
|
611
|
+
'bottle',
|
|
612
|
+
'half',
|
|
613
|
+
'cheek',
|
|
614
|
+
'resolution',
|
|
615
|
+
'back',
|
|
616
|
+
'knowledge',
|
|
617
|
+
'make',
|
|
618
|
+
'discussion',
|
|
619
|
+
'screw',
|
|
620
|
+
'length',
|
|
621
|
+
'accident',
|
|
622
|
+
'battle',
|
|
623
|
+
'dress',
|
|
624
|
+
'knee',
|
|
625
|
+
'log',
|
|
626
|
+
'package',
|
|
627
|
+
'it',
|
|
628
|
+
'turn',
|
|
629
|
+
'hearing',
|
|
630
|
+
'newspaper',
|
|
631
|
+
'layer',
|
|
632
|
+
'wealth',
|
|
633
|
+
'profile',
|
|
634
|
+
'imagination',
|
|
635
|
+
'answer',
|
|
636
|
+
'weekend',
|
|
637
|
+
'teacher',
|
|
638
|
+
'appearance',
|
|
639
|
+
'meet',
|
|
640
|
+
'bike',
|
|
641
|
+
'rise',
|
|
642
|
+
'belt',
|
|
643
|
+
'crash',
|
|
644
|
+
'bowl',
|
|
645
|
+
'equivalent',
|
|
646
|
+
'support',
|
|
647
|
+
'image',
|
|
648
|
+
'poem',
|
|
649
|
+
'risk',
|
|
650
|
+
'excitement',
|
|
651
|
+
'remote',
|
|
652
|
+
'secretary',
|
|
653
|
+
'public',
|
|
654
|
+
'produce',
|
|
655
|
+
'plane',
|
|
656
|
+
'display',
|
|
657
|
+
'money',
|
|
658
|
+
'sand',
|
|
659
|
+
'situation',
|
|
660
|
+
'punch',
|
|
661
|
+
'customer',
|
|
662
|
+
'title',
|
|
663
|
+
'shake',
|
|
664
|
+
'mortgage',
|
|
665
|
+
'option',
|
|
666
|
+
'number',
|
|
667
|
+
'pop',
|
|
668
|
+
'window',
|
|
669
|
+
'extent',
|
|
670
|
+
'nothing',
|
|
671
|
+
'experience',
|
|
672
|
+
'opinion',
|
|
673
|
+
'departure',
|
|
674
|
+
'dance',
|
|
675
|
+
'indication',
|
|
676
|
+
'boy',
|
|
677
|
+
'material',
|
|
678
|
+
'band',
|
|
679
|
+
'leader',
|
|
680
|
+
'sun',
|
|
681
|
+
'beautiful',
|
|
682
|
+
'muscle',
|
|
683
|
+
'farmer',
|
|
684
|
+
'variety',
|
|
685
|
+
'fat',
|
|
686
|
+
'handle',
|
|
687
|
+
'director',
|
|
688
|
+
'opportunity',
|
|
689
|
+
'calendar',
|
|
690
|
+
'outside',
|
|
691
|
+
'pace',
|
|
692
|
+
'bath',
|
|
693
|
+
'fish',
|
|
694
|
+
'consequence',
|
|
695
|
+
'put',
|
|
696
|
+
'owner',
|
|
697
|
+
'go',
|
|
698
|
+
'doctor',
|
|
699
|
+
'information',
|
|
700
|
+
'share',
|
|
701
|
+
'hurt',
|
|
702
|
+
'protection',
|
|
703
|
+
'career',
|
|
704
|
+
'finance',
|
|
705
|
+
'force',
|
|
706
|
+
'golf',
|
|
707
|
+
'garbage',
|
|
708
|
+
'aspect',
|
|
709
|
+
'kid',
|
|
710
|
+
'food',
|
|
711
|
+
'boot',
|
|
712
|
+
'milk',
|
|
713
|
+
'respond',
|
|
714
|
+
'objective',
|
|
715
|
+
'reality',
|
|
716
|
+
'raw',
|
|
717
|
+
'ring',
|
|
718
|
+
'mall',
|
|
719
|
+
'one',
|
|
720
|
+
'impact',
|
|
721
|
+
'area',
|
|
722
|
+
'news',
|
|
723
|
+
'international',
|
|
724
|
+
'series',
|
|
725
|
+
'impress',
|
|
726
|
+
'mother',
|
|
727
|
+
'shelter',
|
|
728
|
+
'strike',
|
|
729
|
+
'loan',
|
|
730
|
+
'month',
|
|
731
|
+
'seat',
|
|
732
|
+
'anything',
|
|
733
|
+
'entertainment',
|
|
734
|
+
'familiar',
|
|
735
|
+
'clue',
|
|
736
|
+
'year',
|
|
737
|
+
'glad',
|
|
738
|
+
'supermarket',
|
|
739
|
+
'natural',
|
|
740
|
+
'god',
|
|
741
|
+
'cost',
|
|
742
|
+
'conversation',
|
|
743
|
+
'tie',
|
|
744
|
+
'ruin',
|
|
745
|
+
'comfort',
|
|
746
|
+
'earth',
|
|
747
|
+
'storm',
|
|
748
|
+
'percentage',
|
|
749
|
+
'assistance',
|
|
750
|
+
'budget',
|
|
751
|
+
'strength',
|
|
752
|
+
'beginning',
|
|
753
|
+
'sleep',
|
|
754
|
+
'other',
|
|
755
|
+
'young',
|
|
756
|
+
'unit',
|
|
757
|
+
'fill',
|
|
758
|
+
'store',
|
|
759
|
+
'desire',
|
|
760
|
+
'hide',
|
|
761
|
+
'value',
|
|
762
|
+
'cup',
|
|
763
|
+
'maintenance',
|
|
764
|
+
'nurse',
|
|
765
|
+
'function',
|
|
766
|
+
'tower',
|
|
767
|
+
'role',
|
|
768
|
+
'class',
|
|
769
|
+
'camera',
|
|
770
|
+
'database',
|
|
771
|
+
'panic',
|
|
772
|
+
'nation',
|
|
773
|
+
'basket',
|
|
774
|
+
'ice',
|
|
775
|
+
'art',
|
|
776
|
+
'spirit',
|
|
777
|
+
'chart',
|
|
778
|
+
'exchange',
|
|
779
|
+
'feedback',
|
|
780
|
+
'statement',
|
|
781
|
+
'reputation',
|
|
782
|
+
'search',
|
|
783
|
+
'hunt',
|
|
784
|
+
'exercise',
|
|
785
|
+
'nasty',
|
|
786
|
+
'notice',
|
|
787
|
+
'male',
|
|
788
|
+
'yard',
|
|
789
|
+
'annual',
|
|
790
|
+
'collar',
|
|
791
|
+
'date',
|
|
792
|
+
'platform',
|
|
793
|
+
'plant',
|
|
794
|
+
'fortune',
|
|
795
|
+
'passion',
|
|
796
|
+
'friendship',
|
|
797
|
+
'spread',
|
|
798
|
+
'cancer',
|
|
799
|
+
'ticket',
|
|
800
|
+
'attitude',
|
|
801
|
+
'island',
|
|
802
|
+
'active',
|
|
803
|
+
'object',
|
|
804
|
+
'service',
|
|
805
|
+
'buyer',
|
|
806
|
+
'bite',
|
|
807
|
+
'card',
|
|
808
|
+
'face',
|
|
809
|
+
'steak',
|
|
810
|
+
'proposal',
|
|
811
|
+
'patient',
|
|
812
|
+
'heat',
|
|
813
|
+
'rule',
|
|
814
|
+
'resident',
|
|
815
|
+
'broad',
|
|
816
|
+
'politics',
|
|
817
|
+
'west',
|
|
818
|
+
'knife',
|
|
819
|
+
'expert',
|
|
820
|
+
'girl',
|
|
821
|
+
'design',
|
|
822
|
+
'salt',
|
|
823
|
+
'baseball',
|
|
824
|
+
'grab',
|
|
825
|
+
'inspection',
|
|
826
|
+
'cousin',
|
|
827
|
+
'couple',
|
|
828
|
+
'magazine',
|
|
829
|
+
'cook',
|
|
830
|
+
'dependent',
|
|
831
|
+
'security',
|
|
832
|
+
'chicken',
|
|
833
|
+
'version',
|
|
834
|
+
'currency',
|
|
835
|
+
'ladder',
|
|
836
|
+
'scheme',
|
|
837
|
+
'kitchen',
|
|
838
|
+
'employment',
|
|
839
|
+
'local',
|
|
840
|
+
'attention',
|
|
841
|
+
'manager',
|
|
842
|
+
'fact',
|
|
843
|
+
'cover',
|
|
844
|
+
'sad',
|
|
845
|
+
'guard',
|
|
846
|
+
'relative',
|
|
847
|
+
'county',
|
|
848
|
+
'rate',
|
|
849
|
+
'lunch',
|
|
850
|
+
'program',
|
|
851
|
+
'initiative',
|
|
852
|
+
'gear',
|
|
853
|
+
'bridge',
|
|
854
|
+
'breast',
|
|
855
|
+
'talk',
|
|
856
|
+
'dish',
|
|
857
|
+
'guarantee',
|
|
858
|
+
'beer',
|
|
859
|
+
'vehicle',
|
|
860
|
+
'reception',
|
|
861
|
+
'woman',
|
|
862
|
+
'substance',
|
|
863
|
+
'copy',
|
|
864
|
+
'lecture',
|
|
865
|
+
'advantage',
|
|
866
|
+
'park',
|
|
867
|
+
'cold',
|
|
868
|
+
'death',
|
|
869
|
+
'mix',
|
|
870
|
+
'hold',
|
|
871
|
+
'scale',
|
|
872
|
+
'tomorrow',
|
|
873
|
+
'blood',
|
|
874
|
+
'request',
|
|
875
|
+
'green',
|
|
876
|
+
'cookie',
|
|
877
|
+
'church',
|
|
878
|
+
'strip',
|
|
879
|
+
'forever',
|
|
880
|
+
'beyond',
|
|
881
|
+
'debt',
|
|
882
|
+
'tackle',
|
|
883
|
+
'wash',
|
|
884
|
+
'following',
|
|
885
|
+
'feel',
|
|
886
|
+
'maximum',
|
|
887
|
+
'sector',
|
|
888
|
+
'sea',
|
|
889
|
+
'property',
|
|
890
|
+
'economics',
|
|
891
|
+
'menu',
|
|
892
|
+
'bench',
|
|
893
|
+
'try',
|
|
894
|
+
'language',
|
|
895
|
+
'start',
|
|
896
|
+
'call',
|
|
897
|
+
'solid',
|
|
898
|
+
'address',
|
|
899
|
+
'income',
|
|
900
|
+
'foot',
|
|
901
|
+
'senior',
|
|
902
|
+
'honey',
|
|
903
|
+
'few',
|
|
904
|
+
'mixture',
|
|
905
|
+
'cash',
|
|
906
|
+
'grocery',
|
|
907
|
+
'link',
|
|
908
|
+
'map',
|
|
909
|
+
'form',
|
|
910
|
+
'factor',
|
|
911
|
+
'pot',
|
|
912
|
+
'model',
|
|
913
|
+
'writer',
|
|
914
|
+
'farm',
|
|
915
|
+
'winter',
|
|
916
|
+
'skill',
|
|
917
|
+
'anywhere',
|
|
918
|
+
'birthday',
|
|
919
|
+
'policy',
|
|
920
|
+
'release',
|
|
921
|
+
'husband',
|
|
922
|
+
'lab',
|
|
923
|
+
'hurry',
|
|
924
|
+
'mail',
|
|
925
|
+
'equipment',
|
|
926
|
+
'sink',
|
|
927
|
+
'pair',
|
|
928
|
+
'driver',
|
|
929
|
+
'consideration',
|
|
930
|
+
'leather',
|
|
931
|
+
'skin',
|
|
932
|
+
'blue',
|
|
933
|
+
'boat',
|
|
934
|
+
'sale',
|
|
935
|
+
'brick',
|
|
936
|
+
'two',
|
|
937
|
+
'feed',
|
|
938
|
+
'square',
|
|
939
|
+
'dot',
|
|
940
|
+
'rush',
|
|
941
|
+
'dream',
|
|
942
|
+
'location',
|
|
943
|
+
'afternoon',
|
|
944
|
+
'manufacturer',
|
|
945
|
+
'control',
|
|
946
|
+
'occasion',
|
|
947
|
+
'trouble',
|
|
948
|
+
'introduction',
|
|
949
|
+
'advice',
|
|
950
|
+
'bet',
|
|
951
|
+
'eat',
|
|
952
|
+
'kill',
|
|
953
|
+
'category',
|
|
954
|
+
'manner',
|
|
955
|
+
'office',
|
|
956
|
+
'estate',
|
|
957
|
+
'pride',
|
|
958
|
+
'awareness',
|
|
959
|
+
'slip',
|
|
960
|
+
'crack',
|
|
961
|
+
'client',
|
|
962
|
+
'nail',
|
|
963
|
+
'shoot',
|
|
964
|
+
'membership',
|
|
965
|
+
'soft',
|
|
966
|
+
'anybody',
|
|
967
|
+
'web',
|
|
968
|
+
'official',
|
|
969
|
+
'individual',
|
|
970
|
+
'pizza',
|
|
971
|
+
'interest',
|
|
972
|
+
'bag',
|
|
973
|
+
'spell',
|
|
974
|
+
'profession',
|
|
975
|
+
'queen',
|
|
976
|
+
'deal',
|
|
977
|
+
'resource',
|
|
978
|
+
'ship',
|
|
979
|
+
'guy',
|
|
980
|
+
'chocolate',
|
|
981
|
+
'joint',
|
|
982
|
+
'formal',
|
|
983
|
+
'upstairs',
|
|
984
|
+
'car',
|
|
985
|
+
'resort',
|
|
986
|
+
'abroad',
|
|
987
|
+
'dealer',
|
|
988
|
+
'associate',
|
|
989
|
+
'finger',
|
|
990
|
+
'surgery',
|
|
991
|
+
'comment',
|
|
992
|
+
'team',
|
|
993
|
+
'detail',
|
|
994
|
+
'crazy',
|
|
995
|
+
'path',
|
|
996
|
+
'tale',
|
|
997
|
+
'initial',
|
|
998
|
+
'arm',
|
|
999
|
+
'radio',
|
|
1000
|
+
'demand',
|
|
1001
|
+
'single',
|
|
1002
|
+
'draw',
|
|
1003
|
+
'yellow',
|
|
1004
|
+
'contest',
|
|
1005
|
+
'piece',
|
|
1006
|
+
'quote',
|
|
1007
|
+
'pull',
|
|
1008
|
+
'commercial',
|
|
1009
|
+
'shirt',
|
|
1010
|
+
'contribution',
|
|
1011
|
+
'cream',
|
|
1012
|
+
'channel',
|
|
1013
|
+
'suit',
|
|
1014
|
+
'discipline',
|
|
1015
|
+
'instruction',
|
|
1016
|
+
'concert',
|
|
1017
|
+
'speech',
|
|
1018
|
+
'low',
|
|
1019
|
+
'effective',
|
|
1020
|
+
'hang',
|
|
1021
|
+
'scratch',
|
|
1022
|
+
'industry',
|
|
1023
|
+
'breakfast',
|
|
1024
|
+
'lay',
|
|
1025
|
+
'join',
|
|
1026
|
+
'metal',
|
|
1027
|
+
'bedroom',
|
|
1028
|
+
'minute',
|
|
1029
|
+
'product',
|
|
1030
|
+
'rest',
|
|
1031
|
+
'temperature',
|
|
1032
|
+
'many',
|
|
1033
|
+
'give',
|
|
1034
|
+
'argument',
|
|
1035
|
+
'print',
|
|
1036
|
+
'purple',
|
|
1037
|
+
'laugh',
|
|
1038
|
+
'health',
|
|
1039
|
+
'credit',
|
|
1040
|
+
'investment',
|
|
1041
|
+
'sell',
|
|
1042
|
+
'setting',
|
|
1043
|
+
'lesson',
|
|
1044
|
+
'egg',
|
|
1045
|
+
'middle',
|
|
1046
|
+
'marriage',
|
|
1047
|
+
'level',
|
|
1048
|
+
'evidence',
|
|
1049
|
+
'phrase',
|
|
1050
|
+
'love',
|
|
1051
|
+
'self',
|
|
1052
|
+
'benefit',
|
|
1053
|
+
'guidance',
|
|
1054
|
+
'affect',
|
|
1055
|
+
'you',
|
|
1056
|
+
'dad',
|
|
1057
|
+
'anxiety',
|
|
1058
|
+
'special',
|
|
1059
|
+
'boyfriend',
|
|
1060
|
+
'test',
|
|
1061
|
+
'blank',
|
|
1062
|
+
'payment',
|
|
1063
|
+
'soup',
|
|
1064
|
+
'obligation',
|
|
1065
|
+
'reply',
|
|
1066
|
+
'smile',
|
|
1067
|
+
'deep',
|
|
1068
|
+
'complaint',
|
|
1069
|
+
'addition',
|
|
1070
|
+
'review',
|
|
1071
|
+
'box',
|
|
1072
|
+
'towel',
|
|
1073
|
+
'minor',
|
|
1074
|
+
'fun',
|
|
1075
|
+
'soil',
|
|
1076
|
+
'issue',
|
|
1077
|
+
'cigarette',
|
|
1078
|
+
'internet',
|
|
1079
|
+
'gain',
|
|
1080
|
+
'tell',
|
|
1081
|
+
'entry',
|
|
1082
|
+
'spare',
|
|
1083
|
+
'incident',
|
|
1084
|
+
'family',
|
|
1085
|
+
'refuse',
|
|
1086
|
+
'branch',
|
|
1087
|
+
'can',
|
|
1088
|
+
'pen',
|
|
1089
|
+
'grandfather',
|
|
1090
|
+
'constant',
|
|
1091
|
+
'tank',
|
|
1092
|
+
'uncle',
|
|
1093
|
+
'climate',
|
|
1094
|
+
'ground',
|
|
1095
|
+
'volume',
|
|
1096
|
+
'communication',
|
|
1097
|
+
'kind',
|
|
1098
|
+
'poet',
|
|
1099
|
+
'child',
|
|
1100
|
+
'screen',
|
|
1101
|
+
'mine',
|
|
1102
|
+
'quit',
|
|
1103
|
+
'gene',
|
|
1104
|
+
'lack',
|
|
1105
|
+
'charity',
|
|
1106
|
+
'memory',
|
|
1107
|
+
'tooth',
|
|
1108
|
+
'fear',
|
|
1109
|
+
'mention',
|
|
1110
|
+
'marketing',
|
|
1111
|
+
'reveal',
|
|
1112
|
+
'reason',
|
|
1113
|
+
'court',
|
|
1114
|
+
'season',
|
|
1115
|
+
'freedom',
|
|
1116
|
+
'land',
|
|
1117
|
+
'sport',
|
|
1118
|
+
'audience',
|
|
1119
|
+
'classroom',
|
|
1120
|
+
'law',
|
|
1121
|
+
'hook',
|
|
1122
|
+
'win',
|
|
1123
|
+
'carry',
|
|
1124
|
+
'eye',
|
|
1125
|
+
'smell',
|
|
1126
|
+
'distribution',
|
|
1127
|
+
'research',
|
|
1128
|
+
'country',
|
|
1129
|
+
'dare',
|
|
1130
|
+
'hope',
|
|
1131
|
+
'whereas',
|
|
1132
|
+
'stretch',
|
|
1133
|
+
'library',
|
|
1134
|
+
'if',
|
|
1135
|
+
'delay',
|
|
1136
|
+
'college',
|
|
1137
|
+
'plastic',
|
|
1138
|
+
'book',
|
|
1139
|
+
'present',
|
|
1140
|
+
'use',
|
|
1141
|
+
'worry',
|
|
1142
|
+
'champion',
|
|
1143
|
+
'goal',
|
|
1144
|
+
'economy',
|
|
1145
|
+
'march',
|
|
1146
|
+
'election',
|
|
1147
|
+
'reflection',
|
|
1148
|
+
'midnight',
|
|
1149
|
+
'slide',
|
|
1150
|
+
'inflation',
|
|
1151
|
+
'action',
|
|
1152
|
+
'challenge',
|
|
1153
|
+
'guitar',
|
|
1154
|
+
'coast',
|
|
1155
|
+
'apple',
|
|
1156
|
+
'campaign',
|
|
1157
|
+
'field',
|
|
1158
|
+
'jacket',
|
|
1159
|
+
'sense',
|
|
1160
|
+
'way',
|
|
1161
|
+
'visual',
|
|
1162
|
+
'remove',
|
|
1163
|
+
'weather',
|
|
1164
|
+
'trash',
|
|
1165
|
+
'cable',
|
|
1166
|
+
'regret',
|
|
1167
|
+
'buddy',
|
|
1168
|
+
'beach',
|
|
1169
|
+
'historian',
|
|
1170
|
+
'courage',
|
|
1171
|
+
'sympathy',
|
|
1172
|
+
'truck',
|
|
1173
|
+
'tension',
|
|
1174
|
+
'permit',
|
|
1175
|
+
'nose',
|
|
1176
|
+
'bed',
|
|
1177
|
+
'son',
|
|
1178
|
+
'person',
|
|
1179
|
+
'base',
|
|
1180
|
+
'meat',
|
|
1181
|
+
'usual',
|
|
1182
|
+
'air',
|
|
1183
|
+
'meeting',
|
|
1184
|
+
'worth',
|
|
1185
|
+
'game',
|
|
1186
|
+
'independence',
|
|
1187
|
+
'physical',
|
|
1188
|
+
'brief',
|
|
1189
|
+
'play',
|
|
1190
|
+
'raise',
|
|
1191
|
+
'board',
|
|
1192
|
+
'she',
|
|
1193
|
+
'key',
|
|
1194
|
+
'writing',
|
|
1195
|
+
'pick',
|
|
1196
|
+
'command',
|
|
1197
|
+
'party',
|
|
1198
|
+
'yesterday',
|
|
1199
|
+
'spring',
|
|
1200
|
+
'candidate',
|
|
1201
|
+
'physics',
|
|
1202
|
+
'university',
|
|
1203
|
+
'concern',
|
|
1204
|
+
'development',
|
|
1205
|
+
'change',
|
|
1206
|
+
'string',
|
|
1207
|
+
'target',
|
|
1208
|
+
'instance',
|
|
1209
|
+
'room',
|
|
1210
|
+
'bitter',
|
|
1211
|
+
'bird',
|
|
1212
|
+
'football',
|
|
1213
|
+
'normal',
|
|
1214
|
+
'split',
|
|
1215
|
+
'impression',
|
|
1216
|
+
'wood',
|
|
1217
|
+
'long',
|
|
1218
|
+
'meaning',
|
|
1219
|
+
'stock',
|
|
1220
|
+
'cap',
|
|
1221
|
+
'leadership',
|
|
1222
|
+
'media',
|
|
1223
|
+
'ambition',
|
|
1224
|
+
'fishing',
|
|
1225
|
+
'essay',
|
|
1226
|
+
'salad',
|
|
1227
|
+
'repair',
|
|
1228
|
+
'today',
|
|
1229
|
+
'designer',
|
|
1230
|
+
'night',
|
|
1231
|
+
'bank',
|
|
1232
|
+
'drawing',
|
|
1233
|
+
'inevitable',
|
|
1234
|
+
'phase',
|
|
1235
|
+
'vast',
|
|
1236
|
+
'chip',
|
|
1237
|
+
'anger',
|
|
1238
|
+
'switch',
|
|
1239
|
+
'cry',
|
|
1240
|
+
'twist',
|
|
1241
|
+
'personality',
|
|
1242
|
+
'attempt',
|
|
1243
|
+
'storage',
|
|
1244
|
+
'being',
|
|
1245
|
+
'preparation',
|
|
1246
|
+
'bat',
|
|
1247
|
+
'selection',
|
|
1248
|
+
'white',
|
|
1249
|
+
'technology',
|
|
1250
|
+
'contract',
|
|
1251
|
+
'side',
|
|
1252
|
+
'section',
|
|
1253
|
+
'station',
|
|
1254
|
+
'till',
|
|
1255
|
+
'structure',
|
|
1256
|
+
'tongue',
|
|
1257
|
+
'taste',
|
|
1258
|
+
'truth',
|
|
1259
|
+
'difficulty',
|
|
1260
|
+
'group',
|
|
1261
|
+
'limit',
|
|
1262
|
+
'main',
|
|
1263
|
+
'move',
|
|
1264
|
+
'feeling',
|
|
1265
|
+
'light',
|
|
1266
|
+
'example',
|
|
1267
|
+
'mission',
|
|
1268
|
+
'might',
|
|
1269
|
+
'wait',
|
|
1270
|
+
'wheel',
|
|
1271
|
+
'shop',
|
|
1272
|
+
'host',
|
|
1273
|
+
'classic',
|
|
1274
|
+
'alternative',
|
|
1275
|
+
'cause',
|
|
1276
|
+
'agent',
|
|
1277
|
+
'consist',
|
|
1278
|
+
'table',
|
|
1279
|
+
'airline',
|
|
1280
|
+
'text',
|
|
1281
|
+
'pool',
|
|
1282
|
+
'craft',
|
|
1283
|
+
'range',
|
|
1284
|
+
'fuel',
|
|
1285
|
+
'tool',
|
|
1286
|
+
'partner',
|
|
1287
|
+
'load',
|
|
1288
|
+
'entrance',
|
|
1289
|
+
'deposit',
|
|
1290
|
+
'hate',
|
|
1291
|
+
'article',
|
|
1292
|
+
'video',
|
|
1293
|
+
'summer',
|
|
1294
|
+
'feature',
|
|
1295
|
+
'extreme',
|
|
1296
|
+
'mobile',
|
|
1297
|
+
'hospital',
|
|
1298
|
+
'flight',
|
|
1299
|
+
'fall',
|
|
1300
|
+
'pension',
|
|
1301
|
+
'piano',
|
|
1302
|
+
'fail',
|
|
1303
|
+
'result',
|
|
1304
|
+
'rub',
|
|
1305
|
+
'gap',
|
|
1306
|
+
'system',
|
|
1307
|
+
'report',
|
|
1308
|
+
'suck',
|
|
1309
|
+
'ordinary',
|
|
1310
|
+
'wind',
|
|
1311
|
+
'nerve',
|
|
1312
|
+
'ask',
|
|
1313
|
+
'shine',
|
|
1314
|
+
'note',
|
|
1315
|
+
'line',
|
|
1316
|
+
'mom',
|
|
1317
|
+
'perception',
|
|
1318
|
+
'brother',
|
|
1319
|
+
'reference',
|
|
1320
|
+
'bend',
|
|
1321
|
+
'charge',
|
|
1322
|
+
'treat',
|
|
1323
|
+
'trick',
|
|
1324
|
+
'term',
|
|
1325
|
+
'homework',
|
|
1326
|
+
'bake',
|
|
1327
|
+
'bid',
|
|
1328
|
+
'status',
|
|
1329
|
+
'project',
|
|
1330
|
+
'strategy',
|
|
1331
|
+
'orange',
|
|
1332
|
+
'let',
|
|
1333
|
+
'enthusiasm',
|
|
1334
|
+
'parent',
|
|
1335
|
+
'concentrate',
|
|
1336
|
+
'device',
|
|
1337
|
+
'travel',
|
|
1338
|
+
'poetry',
|
|
1339
|
+
'business',
|
|
1340
|
+
'society',
|
|
1341
|
+
'kiss',
|
|
1342
|
+
'end',
|
|
1343
|
+
'vegetable',
|
|
1344
|
+
'employ',
|
|
1345
|
+
'schedule',
|
|
1346
|
+
'hour',
|
|
1347
|
+
'brave',
|
|
1348
|
+
'focus',
|
|
1349
|
+
'process',
|
|
1350
|
+
'movie',
|
|
1351
|
+
'illegal',
|
|
1352
|
+
'general',
|
|
1353
|
+
'coffee',
|
|
1354
|
+
'ad',
|
|
1355
|
+
'highway',
|
|
1356
|
+
'chemistry',
|
|
1357
|
+
'psychology',
|
|
1358
|
+
'hire',
|
|
1359
|
+
'bell',
|
|
1360
|
+
'conference',
|
|
1361
|
+
'relief',
|
|
1362
|
+
'show',
|
|
1363
|
+
'neat',
|
|
1364
|
+
'funny',
|
|
1365
|
+
'weight',
|
|
1366
|
+
'quality',
|
|
1367
|
+
'club',
|
|
1368
|
+
'daughter',
|
|
1369
|
+
'zone',
|
|
1370
|
+
'touch',
|
|
1371
|
+
'tonight',
|
|
1372
|
+
'shock',
|
|
1373
|
+
'burn',
|
|
1374
|
+
'excuse',
|
|
1375
|
+
'name',
|
|
1376
|
+
'survey',
|
|
1377
|
+
'landscape',
|
|
1378
|
+
'advance',
|
|
1379
|
+
'satisfaction',
|
|
1380
|
+
'bread',
|
|
1381
|
+
'disaster',
|
|
1382
|
+
'item',
|
|
1383
|
+
'hat',
|
|
1384
|
+
'prior',
|
|
1385
|
+
'shopping',
|
|
1386
|
+
'visit',
|
|
1387
|
+
'east',
|
|
1388
|
+
'photo',
|
|
1389
|
+
'home',
|
|
1390
|
+
'idea',
|
|
1391
|
+
'father',
|
|
1392
|
+
'comparison',
|
|
1393
|
+
'cat',
|
|
1394
|
+
'pipe',
|
|
1395
|
+
'winner',
|
|
1396
|
+
'count',
|
|
1397
|
+
'lake',
|
|
1398
|
+
'fight',
|
|
1399
|
+
'prize',
|
|
1400
|
+
'foundation',
|
|
1401
|
+
'dog',
|
|
1402
|
+
'keep',
|
|
1403
|
+
'ideal',
|
|
1404
|
+
'fan',
|
|
1405
|
+
'struggle',
|
|
1406
|
+
'peak',
|
|
1407
|
+
'safety',
|
|
1408
|
+
'solution',
|
|
1409
|
+
'hell',
|
|
1410
|
+
'conclusion',
|
|
1411
|
+
'population',
|
|
1412
|
+
'strain',
|
|
1413
|
+
'alarm',
|
|
1414
|
+
'measurement',
|
|
1415
|
+
'second',
|
|
1416
|
+
'train',
|
|
1417
|
+
'race',
|
|
1418
|
+
'due',
|
|
1419
|
+
'insurance',
|
|
1420
|
+
'boss',
|
|
1421
|
+
'tree',
|
|
1422
|
+
'monitor',
|
|
1423
|
+
'sick',
|
|
1424
|
+
'course',
|
|
1425
|
+
'drag',
|
|
1426
|
+
'appointment',
|
|
1427
|
+
'slice',
|
|
1428
|
+
'still',
|
|
1429
|
+
'care',
|
|
1430
|
+
'patience',
|
|
1431
|
+
'rich',
|
|
1432
|
+
'escape',
|
|
1433
|
+
'emotion',
|
|
1434
|
+
'royal',
|
|
1435
|
+
'female',
|
|
1436
|
+
'childhood',
|
|
1437
|
+
'government',
|
|
1438
|
+
'picture',
|
|
1439
|
+
'will',
|
|
1440
|
+
'sock',
|
|
1441
|
+
'big',
|
|
1442
|
+
'gate',
|
|
1443
|
+
'oil',
|
|
1444
|
+
'cross',
|
|
1445
|
+
'pin',
|
|
1446
|
+
'improvement',
|
|
1447
|
+
'championship',
|
|
1448
|
+
'silly',
|
|
1449
|
+
'help',
|
|
1450
|
+
'sky',
|
|
1451
|
+
'pitch',
|
|
1452
|
+
'man',
|
|
1453
|
+
'diamond',
|
|
1454
|
+
'most',
|
|
1455
|
+
'transition',
|
|
1456
|
+
'work',
|
|
1457
|
+
'science',
|
|
1458
|
+
'committee',
|
|
1459
|
+
'moment',
|
|
1460
|
+
'fix',
|
|
1461
|
+
'teaching',
|
|
1462
|
+
'dig',
|
|
1463
|
+
'specialist',
|
|
1464
|
+
'complex',
|
|
1465
|
+
'guide',
|
|
1466
|
+
'people',
|
|
1467
|
+
'dead',
|
|
1468
|
+
'voice',
|
|
1469
|
+
'original',
|
|
1470
|
+
'break',
|
|
1471
|
+
'topic',
|
|
1472
|
+
'data',
|
|
1473
|
+
'degree',
|
|
1474
|
+
'reading',
|
|
1475
|
+
'recording',
|
|
1476
|
+
'bunch',
|
|
1477
|
+
'reach',
|
|
1478
|
+
'judgment',
|
|
1479
|
+
'lie',
|
|
1480
|
+
'regular',
|
|
1481
|
+
'set',
|
|
1482
|
+
'painting',
|
|
1483
|
+
'mode',
|
|
1484
|
+
'list',
|
|
1485
|
+
'player',
|
|
1486
|
+
'bear',
|
|
1487
|
+
'north',
|
|
1488
|
+
'wonder',
|
|
1489
|
+
'carpet',
|
|
1490
|
+
'heavy',
|
|
1491
|
+
'officer',
|
|
1492
|
+
'negative',
|
|
1493
|
+
'clock',
|
|
1494
|
+
'unique',
|
|
1495
|
+
'baby',
|
|
1496
|
+
'pain',
|
|
1497
|
+
'assumption',
|
|
1498
|
+
'disk',
|
|
1499
|
+
'iron',
|
|
1500
|
+
'bill',
|
|
1501
|
+
'drawer',
|
|
1502
|
+
'look',
|
|
1503
|
+
'double',
|
|
1504
|
+
'mistake',
|
|
1505
|
+
'finish',
|
|
1506
|
+
'future',
|
|
1507
|
+
'brilliant',
|
|
1508
|
+
'contact',
|
|
1509
|
+
'math',
|
|
1510
|
+
'rice',
|
|
1511
|
+
'leave',
|
|
1512
|
+
'restaurant',
|
|
1513
|
+
'discount',
|
|
1514
|
+
'sex',
|
|
1515
|
+
'virus',
|
|
1516
|
+
'bit',
|
|
1517
|
+
'trust',
|
|
1518
|
+
'event',
|
|
1519
|
+
'wear',
|
|
1520
|
+
'juice',
|
|
1521
|
+
'failure',
|
|
1522
|
+
'bug',
|
|
1523
|
+
'context',
|
|
1524
|
+
'mud',
|
|
1525
|
+
'whole',
|
|
1526
|
+
'wrap',
|
|
1527
|
+
'intention',
|
|
1528
|
+
'draft',
|
|
1529
|
+
'pressure',
|
|
1530
|
+
'cake',
|
|
1531
|
+
'dark',
|
|
1532
|
+
'explanation',
|
|
1533
|
+
'space',
|
|
1534
|
+
'angle',
|
|
1535
|
+
'word',
|
|
1536
|
+
'efficiency',
|
|
1537
|
+
'management',
|
|
1538
|
+
'habit',
|
|
1539
|
+
'star',
|
|
1540
|
+
'chance',
|
|
1541
|
+
'finding',
|
|
1542
|
+
'transportation',
|
|
1543
|
+
'stand',
|
|
1544
|
+
'criticism',
|
|
1545
|
+
'flow',
|
|
1546
|
+
'door',
|
|
1547
|
+
'injury',
|
|
1548
|
+
'insect',
|
|
1549
|
+
'surprise',
|
|
1550
|
+
'apartment',
|
|
1551
|
+
] # pylint: disable=line-too-long
|
|
1552
|
+
|
|
1553
|
+
# ISO 639-1 codes to language names.
|
|
1554
|
+
LANGUAGE_CODES = immutabledict.immutabledict({
|
|
1555
|
+
'en': 'English',
|
|
1556
|
+
'es': 'Spanish',
|
|
1557
|
+
'pt': 'Portuguese',
|
|
1558
|
+
'ar': 'Arabic',
|
|
1559
|
+
'hi': 'Hindi',
|
|
1560
|
+
'fr': 'French',
|
|
1561
|
+
'ru': 'Russian',
|
|
1562
|
+
'de': 'German',
|
|
1563
|
+
'ja': 'Japanese',
|
|
1564
|
+
'it': 'Italian',
|
|
1565
|
+
'bn': 'Bengali',
|
|
1566
|
+
'uk': 'Ukrainian',
|
|
1567
|
+
'th': 'Thai',
|
|
1568
|
+
'ur': 'Urdu',
|
|
1569
|
+
'ta': 'Tamil',
|
|
1570
|
+
'te': 'Telugu',
|
|
1571
|
+
'bg': 'Bulgarian',
|
|
1572
|
+
'ko': 'Korean',
|
|
1573
|
+
'pl': 'Polish',
|
|
1574
|
+
'he': 'Hebrew',
|
|
1575
|
+
'fa': 'Persian',
|
|
1576
|
+
'vi': 'Vietnamese',
|
|
1577
|
+
'ne': 'Nepali',
|
|
1578
|
+
'sw': 'Swahili',
|
|
1579
|
+
'kn': 'Kannada',
|
|
1580
|
+
'mr': 'Marathi',
|
|
1581
|
+
'gu': 'Gujarati',
|
|
1582
|
+
'pa': 'Punjabi',
|
|
1583
|
+
'ml': 'Malayalam',
|
|
1584
|
+
'fi': 'Finnish',
|
|
1585
|
+
})
|
|
1586
|
+
|
|
1587
|
+
_ALPHABETS = '([A-Za-z])'
|
|
1588
|
+
_PREFIXES = '(Mr|St|Mrs|Ms|Dr)[.]'
|
|
1589
|
+
_SUFFIXES = '(Inc|Ltd|Jr|Sr|Co)'
|
|
1590
|
+
_STARTERS = r'(Mr|Mrs|Ms|Dr|Prof|Capt|Cpt|Lt|He\s|She\s|It\s|They\s|Their\s|Our\s|We\s|But\s|However\s|That\s|This\s|Wherever)' # noqa: E501
|
|
1591
|
+
_ACRONYMS = '([A-Z][.][A-Z][.](?:[A-Z][.])?)'
|
|
1592
|
+
_WEBSITES = '[.](com|net|org|io|gov|edu|me)'
|
|
1593
|
+
_DIGITS = '([0-9])'
|
|
1594
|
+
_MULTIPLE_DOTS = r'\.{2,}'
|
|
1595
|
+
|
|
1596
|
+
|
|
1597
|
+
def split_into_sentences(text):
|
|
1598
|
+
"""Split the text into sentences.
|
|
1599
|
+
|
|
1600
|
+
Args:
|
|
1601
|
+
text: A string that consists of more than or equal to one sentences.
|
|
1602
|
+
|
|
1603
|
+
Returns:
|
|
1604
|
+
A list of strings where each string is a sentence.
|
|
1605
|
+
"""
|
|
1606
|
+
text = ' ' + text + ' '
|
|
1607
|
+
text = text.replace('\n', ' ')
|
|
1608
|
+
text = re.sub(_PREFIXES, '\\1<prd>', text)
|
|
1609
|
+
text = re.sub(_WEBSITES, '<prd>\\1', text)
|
|
1610
|
+
text = re.sub(_DIGITS + '[.]' + _DIGITS, '\\1<prd>\\2', text)
|
|
1611
|
+
text = re.sub(
|
|
1612
|
+
_MULTIPLE_DOTS,
|
|
1613
|
+
lambda match: '<prd>' * len(match.group(0)) + '<stop>',
|
|
1614
|
+
text,
|
|
1615
|
+
)
|
|
1616
|
+
if 'Ph.D' in text:
|
|
1617
|
+
text = text.replace('Ph.D.', 'Ph<prd>D<prd>')
|
|
1618
|
+
text = re.sub(r'\s' + _ALPHABETS + '[.] ', ' \\1<prd> ', text)
|
|
1619
|
+
text = re.sub(_ACRONYMS + ' ' + _STARTERS, '\\1<stop> \\2', text)
|
|
1620
|
+
text = re.sub(
|
|
1621
|
+
_ALPHABETS + '[.]' + _ALPHABETS + '[.]' + _ALPHABETS + '[.]',
|
|
1622
|
+
'\\1<prd>\\2<prd>\\3<prd>',
|
|
1623
|
+
text,
|
|
1624
|
+
)
|
|
1625
|
+
text = re.sub(_ALPHABETS + '[.]' + _ALPHABETS + '[.]', '\\1<prd>\\2<prd>', text)
|
|
1626
|
+
text = re.sub(' ' + _SUFFIXES + '[.] ' + _STARTERS, ' \\1<stop> \\2', text)
|
|
1627
|
+
text = re.sub(' ' + _SUFFIXES + '[.]', ' \\1<prd>', text)
|
|
1628
|
+
text = re.sub(' ' + _ALPHABETS + '[.]', ' \\1<prd>', text)
|
|
1629
|
+
if '”' in text:
|
|
1630
|
+
text = text.replace('.”', '”.')
|
|
1631
|
+
if '"' in text:
|
|
1632
|
+
text = text.replace('."', '".')
|
|
1633
|
+
if '!' in text:
|
|
1634
|
+
text = text.replace('!"', '"!')
|
|
1635
|
+
if '?' in text:
|
|
1636
|
+
text = text.replace('?"', '"?')
|
|
1637
|
+
text = text.replace('.', '.<stop>')
|
|
1638
|
+
text = text.replace('?', '?<stop>')
|
|
1639
|
+
text = text.replace('!', '!<stop>')
|
|
1640
|
+
text = text.replace('<prd>', '.')
|
|
1641
|
+
sentences = text.split('<stop>')
|
|
1642
|
+
sentences = [s.strip() for s in sentences]
|
|
1643
|
+
if sentences and not sentences[-1]:
|
|
1644
|
+
sentences = sentences[:-1]
|
|
1645
|
+
return sentences
|
|
1646
|
+
|
|
1647
|
+
|
|
1648
|
+
def count_words(text):
|
|
1649
|
+
"""Counts the number of words."""
|
|
1650
|
+
tokenizer = nltk.tokenize.RegexpTokenizer(r'\w+')
|
|
1651
|
+
tokens = tokenizer.tokenize(text)
|
|
1652
|
+
num_words = len(tokens)
|
|
1653
|
+
return num_words
|
|
1654
|
+
|
|
1655
|
+
|
|
1656
|
+
@functools.lru_cache(maxsize=None)
|
|
1657
|
+
def _get_sentence_tokenizer():
|
|
1658
|
+
return nltk.data.load('nltk:tokenizers/punkt/english.pickle')
|
|
1659
|
+
|
|
1660
|
+
|
|
1661
|
+
def count_sentences(text):
|
|
1662
|
+
"""Count the number of sentences."""
|
|
1663
|
+
tokenizer = _get_sentence_tokenizer()
|
|
1664
|
+
tokenized_sentences = tokenizer.tokenize(text)
|
|
1665
|
+
return len(tokenized_sentences)
|
|
1666
|
+
|
|
1667
|
+
|
|
1668
|
+
def generate_keywords(num_keywords):
|
|
1669
|
+
"""Randomly generates a few keywords."""
|
|
1670
|
+
return random.sample(WORD_LIST, k=num_keywords)
|