Hugging Face's logo

Spaces:
kotstantinovskii
/
YSDA_arxiv_classification
Runtime error

App Files Community
kotstantinovskii commited on
Commit
c19e64d
•
1 Parent(s): 5ebeeca

Upload lables.py

Browse files
Files changed (1) hide show
  1. lables.py +465 -0
lables.py ADDED
@@ -0,0 +1,465 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ num_to_classes = {0: 'astro-ph.CO',
2
+ 1: 'astro-ph.EP',
3
+ 2: 'astro-ph.GA',
4
+ 3: 'astro-ph.HE',
5
+ 4: 'astro-ph.IM',
6
+ 5: 'astro-ph.SR',
7
+ 6: 'astro-ph.all',
8
+ 7: 'cond-mat.dis-nn',
9
+ 8: 'cond-mat.mes-hall',
10
+ 9: 'cond-mat.mtrl-sci',
11
+ 10: 'cond-mat.other',
12
+ 11: 'cond-mat.quant-gas',
13
+ 12: 'cond-mat.soft',
14
+ 13: 'cond-mat.stat-mech',
15
+ 14: 'cond-mat.str-el',
16
+ 15: 'cond-mat.supr-con',
17
+ 16: 'cs.AI',
18
+ 17: 'cs.AR',
19
+ 18: 'cs.CC',
20
+ 19: 'cs.CE',
21
+ 20: 'cs.CG',
22
+ 21: 'cs.CL',
23
+ 22: 'cs.CR',
24
+ 23: 'cs.CV',
25
+ 24: 'cs.CY',
26
+ 25: 'cs.DB',
27
+ 26: 'cs.DC',
28
+ 27: 'cs.DL',
29
+ 28: 'cs.DM',
30
+ 29: 'cs.DS',
31
+ 30: 'cs.ET',
32
+ 31: 'cs.FL',
33
+ 32: 'cs.GL',
34
+ 33: 'cs.GR',
35
+ 34: 'cs.GT',
36
+ 35: 'cs.HC',
37
+ 36: 'cs.IR',
38
+ 37: 'cs.IT',
39
+ 38: 'cs.LG',
40
+ 39: 'cs.LO',
41
+ 40: 'cs.MA',
42
+ 41: 'cs.MM',
43
+ 42: 'cs.MS',
44
+ 43: 'cs.NA',
45
+ 44: 'cs.NE',
46
+ 45: 'cs.NI',
47
+ 46: 'cs.OH',
48
+ 47: 'cs.OS',
49
+ 48: 'cs.PF',
50
+ 49: 'cs.PL',
51
+ 50: 'cs.RO',
52
+ 51: 'cs.SC',
53
+ 52: 'cs.SD',
54
+ 53: 'cs.SE',
55
+ 54: 'cs.SI',
56
+ 55: 'cs.SY',
57
+ 56: 'econ.EM',
58
+ 57: 'econ.GN',
59
+ 58: 'econ.TH',
60
+ 59: 'eess.AS',
61
+ 60: 'eess.IV',
62
+ 61: 'eess.SP',
63
+ 62: 'eess.SY',
64
+ 63: 'gr-qc',
65
+ 64: 'hep-ex',
66
+ 65: 'hep-lat',
67
+ 66: 'hep-ph',
68
+ 67: 'hep-th',
69
+ 68: 'math-ph',
70
+ 69: 'math.AC',
71
+ 70: 'math.AG',
72
+ 71: 'math.AP',
73
+ 72: 'math.AT',
74
+ 73: 'math.CA',
75
+ 74: 'math.CO',
76
+ 75: 'math.CT',
77
+ 76: 'math.CV',
78
+ 77: 'math.DG',
79
+ 78: 'math.DS',
80
+ 79: 'math.FA',
81
+ 80: 'math.GM',
82
+ 81: 'math.GN',
83
+ 82: 'math.GR',
84
+ 83: 'math.GT',
85
+ 84: 'math.HO',
86
+ 85: 'math.KT',
87
+ 86: 'math.LO',
88
+ 87: 'math.MG',
89
+ 88: 'math.NA',
90
+ 89: 'math.NT',
91
+ 90: 'math.OA',
92
+ 91: 'math.OC',
93
+ 92: 'math.PR',
94
+ 93: 'math.QA',
95
+ 94: 'math.RA',
96
+ 95: 'math.RT',
97
+ 96: 'math.SG',
98
+ 97: 'math.SP',
99
+ 98: 'math.ST',
100
+ 99: 'nlin.AO',
101
+ 100: 'nlin.CD',
102
+ 101: 'nlin.CG',
103
+ 102: 'nlin.PS',
104
+ 103: 'nlin.SI',
105
+ 104: 'nucl-ex',
106
+ 105: 'nucl-th',
107
+ 106: 'physics.acc-ph',
108
+ 107: 'physics.ao-ph',
109
+ 108: 'physics.app-ph',
110
+ 109: 'physics.atm-clus',
111
+ 110: 'physics.atom-ph',
112
+ 111: 'physics.bio-ph',
113
+ 112: 'physics.chem-ph',
114
+ 113: 'physics.class-ph',
115
+ 114: 'physics.comp-ph',
116
+ 115: 'physics.data-an',
117
+ 116: 'physics.ed-ph',
118
+ 117: 'physics.flu-dyn',
119
+ 118: 'physics.gen-ph',
120
+ 119: 'physics.geo-ph',
121
+ 120: 'physics.hist-ph',
122
+ 121: 'physics.ins-det',
123
+ 122: 'physics.med-ph',
124
+ 123: 'physics.optics',
125
+ 124: 'physics.plasm-ph',
126
+ 125: 'physics.pop-ph',
127
+ 126: 'physics.soc-ph',
128
+ 127: 'physics.space-ph',
129
+ 128: 'q-bio.BM',
130
+ 129: 'q-bio.CB',
131
+ 130: 'q-bio.GN',
132
+ 131: 'q-bio.MN',
133
+ 132: 'q-bio.NC',
134
+ 133: 'q-bio.OT',
135
+ 134: 'q-bio.PE',
136
+ 135: 'q-bio.QM',
137
+ 136: 'q-bio.SC',
138
+ 137: 'q-bio.TO',
139
+ 138: 'q-fin.CP',
140
+ 139: 'q-fin.EC',
141
+ 140: 'q-fin.GN',
142
+ 141: 'q-fin.MF',
143
+ 142: 'q-fin.PM',
144
+ 143: 'q-fin.PR',
145
+ 144: 'q-fin.RM',
146
+ 145: 'q-fin.ST',
147
+ 146: 'q-fin.TR',
148
+ 147: 'quant-ph',
149
+ 148: 'stat.AP',
150
+ 149: 'stat.CO',
151
+ 150: 'stat.ME',
152
+ 151: 'stat.ML',
153
+ 152: 'stat.OT'}
154
+
155
+ classes_to_num = {'astro-ph.CO': 0,
156
+ 'astro-ph.EP': 1,
157
+ 'astro-ph.GA': 2,
158
+ 'astro-ph.HE': 3,
159
+ 'astro-ph.IM': 4,
160
+ 'astro-ph.SR': 5,
161
+ 'astro-ph.all': 6,
162
+ 'cond-mat.dis-nn': 7,
163
+ 'cond-mat.mes-hall': 8,
164
+ 'cond-mat.mtrl-sci': 9,
165
+ 'cond-mat.other': 10,
166
+ 'cond-mat.quant-gas': 11,
167
+ 'cond-mat.soft': 12,
168
+ 'cond-mat.stat-mech': 13,
169
+ 'cond-mat.str-el': 14,
170
+ 'cond-mat.supr-con': 15,
171
+ 'cs.AI': 16,
172
+ 'cs.AR': 17,
173
+ 'cs.CC': 18,
174
+ 'cs.CE': 19,
175
+ 'cs.CG': 20,
176
+ 'cs.CL': 21,
177
+ 'cs.CR': 22,
178
+ 'cs.CV': 23,
179
+ 'cs.CY': 24,
180
+ 'cs.DB': 25,
181
+ 'cs.DC': 26,
182
+ 'cs.DL': 27,
183
+ 'cs.DM': 28,
184
+ 'cs.DS': 29,
185
+ 'cs.ET': 30,
186
+ 'cs.FL': 31,
187
+ 'cs.GL': 32,
188
+ 'cs.GR': 33,
189
+ 'cs.GT': 34,
190
+ 'cs.HC': 35,
191
+ 'cs.IR': 36,
192
+ 'cs.IT': 37,
193
+ 'cs.LG': 38,
194
+ 'cs.LO': 39,
195
+ 'cs.MA': 40,
196
+ 'cs.MM': 41,
197
+ 'cs.MS': 42,
198
+ 'cs.NA': 43,
199
+ 'cs.NE': 44,
200
+ 'cs.NI': 45,
201
+ 'cs.OH': 46,
202
+ 'cs.OS': 47,
203
+ 'cs.PF': 48,
204
+ 'cs.PL': 49,
205
+ 'cs.RO': 50,
206
+ 'cs.SC': 51,
207
+ 'cs.SD': 52,
208
+ 'cs.SE': 53,
209
+ 'cs.SI': 54,
210
+ 'cs.SY': 55,
211
+ 'econ.EM': 56,
212
+ 'econ.GN': 57,
213
+ 'econ.TH': 58,
214
+ 'eess.AS': 59,
215
+ 'eess.IV': 60,
216
+ 'eess.SP': 61,
217
+ 'eess.SY': 62,
218
+ 'gr-qc': 63,
219
+ 'hep-ex': 64,
220
+ 'hep-lat': 65,
221
+ 'hep-ph': 66,
222
+ 'hep-th': 67,
223
+ 'math-ph': 68,
224
+ 'math.AC': 69,
225
+ 'math.AG': 70,
226
+ 'math.AP': 71,
227
+ 'math.AT': 72,
228
+ 'math.CA': 73,
229
+ 'math.CO': 74,
230
+ 'math.CT': 75,
231
+ 'math.CV': 76,
232
+ 'math.DG': 77,
233
+ 'math.DS': 78,
234
+ 'math.FA': 79,
235
+ 'math.GM': 80,
236
+ 'math.GN': 81,
237
+ 'math.GR': 82,
238
+ 'math.GT': 83,
239
+ 'math.HO': 84,
240
+ 'math.KT': 85,
241
+ 'math.LO': 86,
242
+ 'math.MG': 87,
243
+ 'math.NA': 88,
244
+ 'math.NT': 89,
245
+ 'math.OA': 90,
246
+ 'math.OC': 91,
247
+ 'math.PR': 92,
248
+ 'math.QA': 93,
249
+ 'math.RA': 94,
250
+ 'math.RT': 95,
251
+ 'math.SG': 96,
252
+ 'math.SP': 97,
253
+ 'math.ST': 98,
254
+ 'nlin.AO': 99,
255
+ 'nlin.CD': 100,
256
+ 'nlin.CG': 101,
257
+ 'nlin.PS': 102,
258
+ 'nlin.SI': 103,
259
+ 'nucl-ex': 104,
260
+ 'nucl-th': 105,
261
+ 'physics.acc-ph': 106,
262
+ 'physics.ao-ph': 107,
263
+ 'physics.app-ph': 108,
264
+ 'physics.atm-clus': 109,
265
+ 'physics.atom-ph': 110,
266
+ 'physics.bio-ph': 111,
267
+ 'physics.chem-ph': 112,
268
+ 'physics.class-ph': 113,
269
+ 'physics.comp-ph': 114,
270
+ 'physics.data-an': 115,
271
+ 'physics.ed-ph': 116,
272
+ 'physics.flu-dyn': 117,
273
+ 'physics.gen-ph': 118,
274
+ 'physics.geo-ph': 119,
275
+ 'physics.hist-ph': 120,
276
+ 'physics.ins-det': 121,
277
+ 'physics.med-ph': 122,
278
+ 'physics.optics': 123,
279
+ 'physics.plasm-ph': 124,
280
+ 'physics.pop-ph': 125,
281
+ 'physics.soc-ph': 126,
282
+ 'physics.space-ph': 127,
283
+ 'q-bio.BM': 128,
284
+ 'q-bio.CB': 129,
285
+ 'q-bio.GN': 130,
286
+ 'q-bio.MN': 131,
287
+ 'q-bio.NC': 132,
288
+ 'q-bio.OT': 133,
289
+ 'q-bio.PE': 134,
290
+ 'q-bio.QM': 135,
291
+ 'q-bio.SC': 136,
292
+ 'q-bio.TO': 137,
293
+ 'q-fin.CP': 138,
294
+ 'q-fin.EC': 139,
295
+ 'q-fin.GN': 140,
296
+ 'q-fin.MF': 141,
297
+ 'q-fin.PM': 142,
298
+ 'q-fin.PR': 143,
299
+ 'q-fin.RM': 144,
300
+ 'q-fin.ST': 145,
301
+ 'q-fin.TR': 146,
302
+ 'quant-ph': 147,
303
+ 'stat.AP': 148,
304
+ 'stat.CO': 149,
305
+ 'stat.ME': 150,
306
+ 'stat.ML': 151,
307
+ 'stat.OT': 152}
308
+
309
+ taxonomy = {'cs.AI': 'Artificial Intelligence',
310
+ 'cs.AR': 'Hardware Architecture',
311
+ 'cs.CC': 'Computational Complexity',
312
+ 'cs.CE': 'Computational Engineering, Finance, and Science',
313
+ 'cs.CG': 'Computational Geometry',
314
+ 'cs.CL': 'Computation and Language',
315
+ 'cs.CR': 'Cryptography and Security',
316
+ 'cs.CV': 'Computer Vision and Pattern Recognition',
317
+ 'cs.CY': 'Computers and Society',
318
+ 'cs.DB': 'Databases',
319
+ 'cs.DC': 'Distributed, Parallel, and Cluster Computing',
320
+ 'cs.DL': 'Digital Libraries',
321
+ 'cs.DM': 'Discrete Mathematics',
322
+ 'cs.DS': 'Data Structures and Algorithms',
323
+ 'cs.ET': 'Emerging Technologies',
324
+ 'cs.FL': 'Formal Languages and Automata Theory',
325
+ 'cs.GL': 'General Literature',
326
+ 'cs.GR': 'Graphics',
327
+ 'cs.GT': 'Computer Science and Game Theory',
328
+ 'cs.HC': 'Human-Computer Interaction',
329
+ 'cs.IR': 'Information Retrieval',
330
+ 'cs.IT': 'Information Theory',
331
+ 'cs.LG': 'Machine Learning',
332
+ 'cs.LO': 'Logic in Computer Science',
333
+ 'cs.MA': 'Multiagent Systems',
334
+ 'cs.MM': 'Multimedia',
335
+ 'cs.MS': 'Mathematical Software',
336
+ 'cs.NA': 'Numerical Analysis',
337
+ 'cs.NE': 'Neural and Evolutionary Computing',
338
+ 'cs.NI': 'Networking and Internet Architecture',
339
+ 'cs.OH': 'Other Computer Science',
340
+ 'cs.OS': 'Operating Systems',
341
+ 'cs.PF': 'Performance',
342
+ 'cs.PL': 'Programming Languages',
343
+ 'cs.RO': 'Robotics',
344
+ 'cs.SC': 'Symbolic Computation',
345
+ 'cs.SD': 'Sound',
346
+ 'cs.SE': 'Software Engineering',
347
+ 'cs.SI': 'Social and Information Networks',
348
+ 'cs.SY': 'Systems and Control',
349
+ 'econ.EM': 'Econometrics',
350
+ 'econ.GN': 'General Economics',
351
+ 'econ.TH': 'Theoretical Economics',
352
+ 'eess.AS': 'Audio and Speech Processing',
353
+ 'eess.IV': 'Image and Video Processing',
354
+ 'eess.SP': 'Signal Processing',
355
+ 'eess.SY': 'Systems and Control',
356
+ 'math.AC': 'Commutative Algebra',
357
+ 'math.AG': 'Algebraic Geometry',
358
+ 'math.AP': 'Analysis of PDEs',
359
+ 'math.AT': 'Algebraic Topology',
360
+ 'math.CA': 'Classical Analysis and ODEs',
361
+ 'math.CO': 'Combinatorics',
362
+ 'math.CT': 'Category Theory',
363
+ 'math.CV': 'Complex Variables',
364
+ 'math.DG': 'Differential Geometry',
365
+ 'math.DS': 'Dynamical Systems',
366
+ 'math.FA': 'Functional Analysis',
367
+ 'math.GM': 'General Mathematics',
368
+ 'math.GN': 'General Topology',
369
+ 'math.GR': 'Group Theory',
370
+ 'math.GT': 'Geometric Topology',
371
+ 'math.HO': 'History and Overview',
372
+ 'math.IT': 'Information Theory',
373
+ 'math.KT': 'K-Theory and Homology',
374
+ 'math.LO': 'Logic',
375
+ 'math.MG': 'Metric Geometry',
376
+ 'math.MP': 'Mathematical Physics',
377
+ 'math.NA': 'Numerical Analysis',
378
+ 'math.NT': 'Number Theory',
379
+ 'math.OA': 'Operator Algebras',
380
+ 'math.OC': 'Optimization and Control',
381
+ 'math.PR': 'Probability',
382
+ 'math.QA': 'Quantum Algebra',
383
+ 'math.RA': 'Rings and Algebras',
384
+ 'math.RT': 'Representation Theory',
385
+ 'math.SG': 'Symplectic Geometry',
386
+ 'math.SP': 'Spectral Theory',
387
+ 'math.ST': 'Statistics Theory',
388
+ 'astro-ph.all': 'Astrophysics',
389
+ 'astro-ph.CO': 'Cosmology and Nongalactic Astrophysics',
390
+ 'astro-ph.EP': 'Earth and Planetary Astrophysics',
391
+ 'astro-ph.GA': 'Astrophysics of Galaxies',
392
+ 'astro-ph.HE': 'High Energy Astrophysical Phenomena',
393
+ 'astro-ph.IM': 'Instrumentation and Methods for Astrophysics',
394
+ 'astro-ph.SR': 'Solar and Stellar Astrophysics',
395
+ 'cond-mat.dis-nn': 'Disordered Systems and Neural Networks',
396
+ 'cond-mat.mes-hall': 'Mesoscale and Nanoscale Physics',
397
+ 'cond-mat.mtrl-sci': 'Materials Science',
398
+ 'cond-mat.other': 'Other Condensed Matter',
399
+ 'cond-mat.quant-gas': 'Quantum Gases',
400
+ 'cond-mat.soft': 'Soft Condensed Matter',
401
+ 'cond-mat.stat-mech': 'Statistical Mechanics',
402
+ 'cond-mat.str-el': 'Strongly Correlated Electrons',
403
+ 'cond-mat.supr-con': 'Superconductivity',
404
+ 'gr-qc': 'General Relativity and Quantum Cosmology',
405
+ 'hep-ex': 'High Energy Physics - Experiment',
406
+ 'hep-lat': 'High Energy Physics - Lattice',
407
+ 'hep-ph': 'High Energy Physics - Phenomenology',
408
+ 'hep-th': 'High Energy Physics - Theory',
409
+ 'math-ph': 'Mathematical Physics',
410
+ 'nlin.AO': 'Adaptation and Self-Organizing Systems',
411
+ 'nlin.CD': 'Chaotic Dynamics',
412
+ 'nlin.CG': 'Cellular Automata and Lattice Gases',
413
+ 'nlin.PS': 'Pattern Formation and Solitons',
414
+ 'nlin.SI': 'Exactly Solvable and Integrable Systems',
415
+ 'nucl-ex': 'Nuclear Experiment',
416
+ 'nucl-th': 'Nuclear Theory',
417
+ 'physics.acc-ph': 'Accelerator Physics',
418
+ 'physics.ao-ph': 'Atmospheric and Oceanic Physics',
419
+ 'physics.app-ph': 'Applied Physics',
420
+ 'physics.atm-clus': 'Atomic and Molecular Clusters',
421
+ 'atom-ph.all': 'Atomic Physics',
422
+ 'physics.atom-ph': 'Atomic Physics',
423
+ 'physics.bio-ph': 'Biological Physics',
424
+ 'physics.chem-ph': 'Chemical Physics',
425
+ 'physics.class-ph': 'Classical Physics',
426
+ 'physics.comp-ph': 'Computational Physics',
427
+ 'physics.data-an': 'Data Analysis, Statistics and Probability',
428
+ 'physics.ed-ph': 'Physics Education',
429
+ 'physics.flu-dyn': 'Fluid Dynamics',
430
+ 'physics.gen-ph': 'General Physics',
431
+ 'physics.geo-ph': 'Geophysics',
432
+ 'physics.hist-ph': 'History and Philosophy of Physics',
433
+ 'physics.ins-det': 'Instrumentation and Detectors',
434
+ 'physics.med-ph': 'Medical Physics',
435
+ 'physics.optics': 'Optics',
436
+ 'physics.plasm-ph': 'Plasma Physics',
437
+ 'physics.pop-ph': 'Popular Physics',
438
+ 'physics.soc-ph': 'Physics and Society',
439
+ 'physics.space-ph': 'Space Physics',
440
+ 'quant-ph': 'Quantum Physics',
441
+ 'q-bio.BM': 'Biomolecules',
442
+ 'q-bio.CB': 'Cell Behavior',
443
+ 'q-bio.GN': 'Genomics',
444
+ 'q-bio.MN': 'Molecular Networks',
445
+ 'q-bio.NC': 'Neurons and Cognition',
446
+ 'q-bio.OT': 'Other Quantitative Biology',
447
+ 'q-bio.PE': 'Populations and Evolution',
448
+ 'q-bio.QM': 'Quantitative Methods',
449
+ 'q-bio.SC': 'Subcellular Processes',
450
+ 'q-bio.TO': 'Tissues and Organs',
451
+ 'q-fin.CP': 'Computational Finance',
452
+ 'q-fin.EC': 'Economics',
453
+ 'q-fin.GN': 'General Finance',
454
+ 'q-fin.MF': 'Mathematical Finance',
455
+ 'q-fin.PM': 'Portfolio Management',
456
+ 'q-fin.PR': 'Pricing of Securities',
457
+ 'q-fin.RM': 'Risk Management',
458
+ 'q-fin.ST': 'Statistical Finance',
459
+ 'q-fin.TR': 'Trading and Market Microstructure',
460
+ 'stat.AP': 'Applications',
461
+ 'stat.CO': 'Computation',
462
+ 'stat.ME': 'Methodology',
463
+ 'stat.ML': 'Machine Learning',
464
+ 'stat.OT': 'Other Statistics',
465
+ 'stat.TH': 'Statistics Theory'}