Jacobo commited on
Commit
525fae0
·
verified ·
1 Parent(s): ad10d91

Update spaCy pipeline

Browse files
README.md CHANGED
The diff for this file is too large to render. See raw diff
 
config.cfg CHANGED
@@ -1,6 +1,6 @@
1
  [paths]
2
- train = "corpus/ud/proiel/train"
3
- dev = "corpus/ud/proiel/dev"
4
  vectors = "vectors/large"
5
  init_tok2vec = "data/pretrained_weights/model32.bin"
6
  raw_text = "raw_text"
@@ -11,7 +11,7 @@ seed = 0
11
 
12
  [nlp]
13
  lang = "grc"
14
- pipeline = ["tok2vec","morphologizer","tagger","parser","lemmatizer","attribute_ruler","ner"]
15
  batch_size = 128
16
  disabled = []
17
  before_creation = null
@@ -180,7 +180,7 @@ accumulate_gradient = 1
180
  patience = 5000
181
  max_epochs = 0
182
  max_steps = 20000
183
- eval_frequency = 200
184
  frozen_components = ["lemmatizer"]
185
  annotating_components = []
186
  before_to_disk = null
 
1
  [paths]
2
+ train = "corpus/parser/proiel/train"
3
+ dev = "corpus/parser/proiel/dev"
4
  vectors = "vectors/large"
5
  init_tok2vec = "data/pretrained_weights/model32.bin"
6
  raw_text = "raw_text"
 
11
 
12
  [nlp]
13
  lang = "grc"
14
+ pipeline = ["tok2vec","morphologizer","tagger","parser","lemmatizer","ner","attribute_ruler"]
15
  batch_size = 128
16
  disabled = []
17
  before_creation = null
 
180
  patience = 5000
181
  max_epochs = 0
182
  max_steps = 20000
183
+ eval_frequency = 1000
184
  frozen_components = ["lemmatizer"]
185
  annotating_components = []
186
  before_to_disk = null
grc_proiel_lg-any-py3-none-any.whl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6809343436727d7763e38d18cc57539b0a7ee861d58008d86d6cef7433c8afc0
3
- size 279721833
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe7cff0391d7573700239218d74a7033bd6517f9c2597c9b667e115cc72efe68
3
+ size 284804128
lemmatizer/cfg CHANGED
The diff for this file is too large to render. See raw diff
 
lemmatizer/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3c508311a7fb1b93127610f56746de2adcedfb3b9778b4f54c5c181934c56591
3
- size 24489602
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d5013a0dc1727b911c7f243338801fdc212f29d54ddfdb88a7abbec84448a54
3
+ size 29442044
lemmatizer/trees CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:087b25d2e16192e3addbfbe3d55ae0b878ee6893648f136313c0d8b21505df7c
3
- size 5318272
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86606a185ea3cbd4c6f071029121d17265dfe9d5a468c6dbeb071a5808cf49f9
3
+ size 6516433
meta.json CHANGED
@@ -1,14 +1,14 @@
1
  {
2
  "lang":"grc",
3
  "name":"proiel_lg",
4
- "version":"3.7",
5
  "description":"",
6
  "author":"",
7
  "email":"",
8
  "url":"",
9
  "license":"",
10
- "spacy_version":">=3.7.2,<3.8.0",
11
- "spacy_git_version":"a89eae928",
12
  "vectors":{
13
  "width":300,
14
  "vectors":200000,
@@ -45,6 +45,7 @@
45
  "Case=Acc|Gender=Fem|Number=Sing|POS=NOUN",
46
  "Aspect=Perf|Mood=Ind|Number=Plur|POS=VERB|Person=3|Tense=Past|VerbForm=Fin|Voice=Act",
47
  "Case=Dat|Gender=Masc|Number=Plur|POS=PRON|PronType=Rcp",
 
48
  "Case=Nom|Definite=Def|Gender=Masc|Number=Plur|POS=DET|PronType=Dem",
49
  "Case=Nom|Gender=Masc|Number=Plur|POS=NOUN",
50
  "Case=Acc|Gender=Masc|Number=Plur|POS=NOUN",
@@ -727,6 +728,7 @@
727
  "Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|POS=VERB|Tense=Past|VerbForm=Part|Voice=Act",
728
  "Aspect=Perf|Mood=Sub|Number=Plur|POS=VERB|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid",
729
  "Case=Dat|Degree=Sup|Gender=Fem|Number=Sing|POS=ADJ",
 
730
  "Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|POS=VERB|Tense=Past|VerbForm=Part|Voice=Pass",
731
  "Case=Acc|Gender=Fem,Masc|Number=Sing|POS=NOUN",
732
  "Case=Dat|Gender=Fem,Masc|Number=Sing|POS=NOUN",
@@ -1073,7 +1075,8 @@
1073
  "Px",
1074
  "R-",
1075
  "S-",
1076
- "V-"
 
1077
  ],
1078
  "parser":[
1079
  "ROOT",
@@ -1105,18 +1108,23 @@
1105
  "obl:agent",
1106
  "orphan",
1107
  "parataxis",
 
1108
  "vocative",
1109
  "xcomp"
1110
- ],
1111
- "attribute_ruler":[
1112
-
1113
  ],
1114
  "ner":[
 
1115
  "GOD",
 
1116
  "LANGUAGE",
1117
  "LOC",
1118
  "NORP",
1119
- "PERSON"
 
 
 
 
 
1120
  ]
1121
  },
1122
  "pipeline":[
@@ -1125,8 +1133,8 @@
1125
  "tagger",
1126
  "parser",
1127
  "lemmatizer",
1128
- "attribute_ruler",
1129
- "ner"
1130
  ],
1131
  "components":[
1132
  "tok2vec",
@@ -1134,80 +1142,80 @@
1134
  "tagger",
1135
  "parser",
1136
  "lemmatizer",
1137
- "attribute_ruler",
1138
- "ner"
1139
  ],
1140
  "disabled":[
1141
 
1142
  ],
1143
  "performance":{
1144
- "pos_acc":0.9721652505,
1145
- "morph_acc":0.9029446235,
1146
  "morph_per_feat":{
1147
  "Case":{
1148
- "p":0.9687855428,
1149
- "r":0.9674406865,
1150
- "f":0.9681126476
1151
  },
1152
  "Gender":{
1153
- "p":0.9175768352,
1154
- "r":0.9219088937,
1155
- "f":0.9197377634
1156
  },
1157
  "Number":{
1158
- "p":0.9895358475,
1159
- "r":0.9875917692,
1160
- "f":0.9885628526
1161
  },
1162
  "Person":{
1163
- "p":0.9758902077,
1164
- "r":0.9712070875,
1165
- "f":0.9735430157
1166
  },
1167
  "PronType":{
1168
- "p":0.9856020942,
1169
- "r":0.9779220779,
1170
- "f":0.9817470665
1171
  },
1172
  "Polarity":{
1173
  "p":1.0,
1174
- "r":0.9792746114,
1175
- "f":0.9895287958
1176
  },
1177
  "Aspect":{
1178
- "p":0.9659353349,
1179
- "r":0.9494892168,
1180
- "f":0.9576416714
1181
  },
1182
  "Mood":{
1183
- "p":0.9712478436,
1184
- "r":0.966800229,
1185
- "f":0.9690189329
1186
  },
1187
  "Tense":{
1188
- "p":0.9627071823,
1189
- "r":0.9577464789,
1190
- "f":0.9602204236
1191
  },
1192
  "VerbForm":{
1193
- "p":0.9886049724,
1194
- "r":0.9828355647,
1195
- "f":0.9857118265
1196
  },
1197
  "Voice":{
1198
- "p":0.955801105,
1199
- "r":0.9508759876,
1200
- "f":0.9533321853
1201
  },
1202
  "Degree":{
1203
- "p":0.9107142857,
1204
- "r":0.8911290323,
1205
- "f":0.9008152174
1206
  },
1207
  "Definite":{
1208
- "p":0.9908896034,
1209
- "r":0.998919503,
1210
- "f":0.9948883508
1211
  },
1212
  "Reflex":{
1213
  "p":1.0,
@@ -1216,118 +1224,118 @@
1216
  },
1217
  "Poss":{
1218
  "p":1.0,
1219
- "r":0.7894736842,
1220
- "f":0.8823529412
1221
  }
1222
  },
1223
- "tag_acc":0.9762672136,
1224
- "dep_uas":0.8078669792,
1225
- "dep_las":0.7601816584,
1226
  "dep_las_per_type":{
1227
  "nsubj":{
1228
- "p":0.771556551,
1229
- "r":0.7464788732,
1230
- "f":0.7588105727
1231
  },
1232
  "discourse":{
1233
- "p":0.8090659341,
1234
- "r":0.8090659341,
1235
- "f":0.8090659341
1236
  },
1237
  "mark":{
1238
- "p":0.8571428571,
1239
- "r":0.8467741935,
1240
- "f":0.8519269777
1241
  },
1242
  "advmod":{
1243
- "p":0.7295208655,
1244
- "r":0.7013372957,
1245
- "f":0.7151515152
1246
  },
1247
  "advcl":{
1248
- "p":0.6763590392,
1249
- "r":0.7308743169,
1250
- "f":0.7025607354
1251
  },
1252
  "xcomp":{
1253
- "p":0.5555555556,
1254
- "r":0.48,
1255
- "f":0.5150214592
1256
  },
1257
  "cop":{
1258
- "p":0.7336448598,
1259
- "r":0.7336448598,
1260
- "f":0.7336448598
1261
  },
1262
  "root":{
1263
- "p":0.8128078818,
1264
- "r":0.8096172718,
1265
- "f":0.8112094395
1266
  },
1267
  "det":{
1268
- "p":0.9340611354,
1269
- "r":0.9275802255,
1270
- "f":0.9308093995
1271
  },
1272
  "nmod":{
1273
- "p":0.7253521127,
1274
- "r":0.7253521127,
1275
- "f":0.7253521127
1276
  },
1277
  "obj":{
1278
- "p":0.7748792271,
1279
- "r":0.8328141225,
1280
- "f":0.8028028028
1281
  },
1282
  "case":{
1283
- "p":0.9552889858,
1284
- "r":0.9605263158,
1285
- "f":0.9579004921
1286
  },
1287
  "obl":{
1288
- "p":0.728960396,
1289
- "r":0.7191697192,
1290
- "f":0.7240319607
1291
  },
1292
  "cc":{
1293
- "p":0.6473029046,
1294
- "r":0.6419753086,
1295
- "f":0.6446280992
1296
  },
1297
  "conj":{
1298
- "p":0.5976253298,
1299
- "r":0.594488189,
1300
- "f":0.5960526316
1301
  },
1302
  "obl:agent":{
1303
- "p":0.8095238095,
1304
- "r":0.4594594595,
1305
- "f":0.5862068966
1306
  },
1307
  "ccomp":{
1308
- "p":0.5073170732,
1309
- "r":0.5174129353,
1310
- "f":0.5123152709
1311
  },
1312
  "nsubj:pass":{
1313
- "p":0.6517857143,
1314
- "r":0.6822429907,
1315
- "f":0.6666666667
1316
  },
1317
  "amod":{
1318
- "p":0.736318408,
1319
- "r":0.6948356808,
1320
- "f":0.7149758454
1321
  },
1322
  "acl":{
1323
- "p":0.4571428571,
1324
- "r":0.3878787879,
1325
- "f":0.4196721311
1326
  },
1327
  "iobj":{
1328
- "p":0.7433414044,
1329
- "r":0.7090069284,
1330
- "f":0.7257683215
1331
  },
1332
  "dep":{
1333
  "p":0.0,
@@ -1335,44 +1343,44 @@
1335
  "f":0.0
1336
  },
1337
  "nummod":{
1338
- "p":0.9122807018,
1339
- "r":0.7647058824,
1340
- "f":0.832
1341
  },
1342
  "vocative":{
1343
- "p":0.7413793103,
1344
- "r":0.6231884058,
1345
- "f":0.6771653543
1346
  },
1347
  "orphan":{
1348
- "p":0.2272727273,
1349
  "r":0.1162790698,
1350
- "f":0.1538461538
1351
  },
1352
  "appos":{
1353
- "p":0.4528301887,
1354
- "r":0.3404255319,
1355
- "f":0.3886639676
1356
- },
1357
- "parataxis":{
1358
- "p":0.0,
1359
- "r":0.0,
1360
- "f":0.0
1361
  },
1362
  "dislocated":{
1363
- "p":0.4166666667,
1364
- "r":0.1923076923,
1365
- "f":0.2631578947
1366
  },
1367
- "csubj:pass":{
1368
- "p":0.0,
1369
- "r":0.0,
1370
- "f":0.0
1371
  },
1372
  "flat:name":{
1373
- "p":0.8666666667,
1374
  "r":0.5909090909,
1375
- "f":0.7027027027
 
 
 
 
 
1376
  },
1377
  "aux:pass":{
1378
  "p":0.0,
@@ -1380,9 +1388,9 @@
1380
  "f":0.0
1381
  },
1382
  "fixed":{
1383
- "p":0.7777777778,
1384
- "r":0.7,
1385
- "f":0.7368421053
1386
  },
1387
  "aux":{
1388
  "p":0.0,
@@ -1390,40 +1398,60 @@
1390
  "f":0.0
1391
  }
1392
  },
1393
- "sents_p":0.567816092,
1394
- "sents_r":0.6149377593,
1395
- "sents_f":0.590438247,
1396
- "lemma_acc":0.9734104893,
1397
- "ents_f":0.9117647059,
1398
- "ents_p":0.9117647059,
1399
- "ents_r":0.9117647059,
1400
  "ents_per_type":{
 
 
 
 
 
1401
  "NORP":{
1402
- "p":0.902173913,
1403
- "r":0.9651162791,
1404
- "f":0.9325842697
1405
  },
1406
  "PERSON":{
1407
- "p":0.9032258065,
1408
- "r":0.9572649573,
1409
- "f":0.9294605809
1410
  },
1411
- "LOC":{
1412
- "p":0.9615384615,
1413
- "r":0.8064516129,
1414
- "f":0.8771929825
1415
  },
1416
  "GOD":{
1417
- "p":0.75,
1418
- "r":0.4285714286,
1419
- "f":0.5454545455
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1420
  }
1421
  },
1422
- "tok2vec_loss":84341.9538791711,
1423
- "morphologizer_loss":710.5406859517,
1424
- "tagger_loss":199.4221925758,
1425
- "parser_loss":12070.2685545735,
1426
- "ner_loss":14.0204100051
1427
  },
1428
  "requirements":[
1429
 
 
1
  {
2
  "lang":"grc",
3
  "name":"proiel_lg",
4
+ "version":"3.7.5",
5
  "description":"",
6
  "author":"",
7
  "email":"",
8
  "url":"",
9
  "license":"",
10
+ "spacy_version":">=3.7.5,<3.8.0",
11
+ "spacy_git_version":"a6d0fc360",
12
  "vectors":{
13
  "width":300,
14
  "vectors":200000,
 
45
  "Case=Acc|Gender=Fem|Number=Sing|POS=NOUN",
46
  "Aspect=Perf|Mood=Ind|Number=Plur|POS=VERB|Person=3|Tense=Past|VerbForm=Fin|Voice=Act",
47
  "Case=Dat|Gender=Masc|Number=Plur|POS=PRON|PronType=Rcp",
48
+ "POS=PUNCT",
49
  "Case=Nom|Definite=Def|Gender=Masc|Number=Plur|POS=DET|PronType=Dem",
50
  "Case=Nom|Gender=Masc|Number=Plur|POS=NOUN",
51
  "Case=Acc|Gender=Masc|Number=Plur|POS=NOUN",
 
728
  "Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|POS=VERB|Tense=Past|VerbForm=Part|Voice=Act",
729
  "Aspect=Perf|Mood=Sub|Number=Plur|POS=VERB|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid",
730
  "Case=Dat|Degree=Sup|Gender=Fem|Number=Sing|POS=ADJ",
731
+ "POS=AUX",
732
  "Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|POS=VERB|Tense=Past|VerbForm=Part|Voice=Pass",
733
  "Case=Acc|Gender=Fem,Masc|Number=Sing|POS=NOUN",
734
  "Case=Dat|Gender=Fem,Masc|Number=Sing|POS=NOUN",
 
1075
  "Px",
1076
  "R-",
1077
  "S-",
1078
+ "V-",
1079
+ "Z"
1080
  ],
1081
  "parser":[
1082
  "ROOT",
 
1108
  "obl:agent",
1109
  "orphan",
1110
  "parataxis",
1111
+ "punct",
1112
  "vocative",
1113
  "xcomp"
 
 
 
1114
  ],
1115
  "ner":[
1116
+ "EVENT",
1117
  "GOD",
1118
+ "GPE",
1119
  "LANGUAGE",
1120
  "LOC",
1121
  "NORP",
1122
+ "ORG",
1123
+ "PERSON",
1124
+ "WORK"
1125
+ ],
1126
+ "attribute_ruler":[
1127
+
1128
  ]
1129
  },
1130
  "pipeline":[
 
1133
  "tagger",
1134
  "parser",
1135
  "lemmatizer",
1136
+ "ner",
1137
+ "attribute_ruler"
1138
  ],
1139
  "components":[
1140
  "tok2vec",
 
1142
  "tagger",
1143
  "parser",
1144
  "lemmatizer",
1145
+ "ner",
1146
+ "attribute_ruler"
1147
  ],
1148
  "disabled":[
1149
 
1150
  ],
1151
  "performance":{
1152
+ "pos_acc":0.9753936337,
1153
+ "morph_acc":0.9139117988,
1154
  "morph_per_feat":{
1155
  "Case":{
1156
+ "p":0.9721135647,
1157
+ "r":0.9722362443,
1158
+ "f":0.9721749006
1159
  },
1160
  "Gender":{
1161
+ "p":0.9201014585,
1162
+ "r":0.9257368891,
1163
+ "f":0.9229105712
1164
  },
1165
  "Number":{
1166
+ "p":0.9906851584,
1167
+ "r":0.9897632096,
1168
+ "f":0.9902239694
1169
  },
1170
  "Person":{
1171
+ "p":0.9777200149,
1172
+ "r":0.9719453673,
1173
+ "f":0.9748241392
1174
  },
1175
  "PronType":{
1176
+ "p":0.9875898106,
1177
+ "r":0.9818181818,
1178
+ "f":0.9846955389
1179
  },
1180
  "Polarity":{
1181
  "p":1.0,
1182
+ "r":0.9844559585,
1183
+ "f":0.9921671018
1184
  },
1185
  "Aspect":{
1186
+ "p":0.9674229203,
1187
+ "r":0.9438138479,
1188
+ "f":0.9554725654
1189
  },
1190
  "Mood":{
1191
+ "p":0.9769717904,
1192
+ "r":0.9713795077,
1193
+ "f":0.9741676234
1194
  },
1195
  "Tense":{
1196
+ "p":0.9595015576,
1197
+ "r":0.9522500859,
1198
+ "f":0.955862069
1199
  },
1200
  "VerbForm":{
1201
+ "p":0.9903080651,
1202
+ "r":0.9821489873,
1203
+ "f":0.9862116512
1204
  },
1205
  "Voice":{
1206
+ "p":0.9574247144,
1207
+ "r":0.9501889385,
1208
+ "f":0.9537931034
1209
  },
1210
  "Degree":{
1211
+ "p":0.9219858156,
1212
+ "r":0.873655914,
1213
+ "f":0.8971704624
1214
  },
1215
  "Definite":{
1216
+ "p":0.9919571046,
1217
+ "r":0.9994597515,
1218
+ "f":0.9956942949
1219
  },
1220
  "Reflex":{
1221
  "p":1.0,
 
1224
  },
1225
  "Poss":{
1226
  "p":1.0,
1227
+ "r":0.8421052632,
1228
+ "f":0.9142857143
1229
  }
1230
  },
1231
+ "tag_acc":0.9769613523,
1232
+ "dep_uas":0.8360032185,
1233
+ "dep_las":0.7900665643,
1234
  "dep_las_per_type":{
1235
  "nsubj":{
1236
+ "p":0.7834051724,
1237
+ "r":0.7876489707,
1238
+ "f":0.7855213398
1239
  },
1240
  "discourse":{
1241
+ "p":0.8315068493,
1242
+ "r":0.8337912088,
1243
+ "f":0.8326474623
1244
  },
1245
  "mark":{
1246
+ "p":0.8495934959,
1247
+ "r":0.8427419355,
1248
+ "f":0.8461538462
1249
  },
1250
  "advmod":{
1251
+ "p":0.7535321821,
1252
+ "r":0.7132243685,
1253
+ "f":0.7328244275
1254
  },
1255
  "advcl":{
1256
+ "p":0.7212449256,
1257
+ "r":0.7281420765,
1258
+ "f":0.7246770904
1259
  },
1260
  "xcomp":{
1261
+ "p":0.5726495726,
1262
+ "r":0.536,
1263
+ "f":0.5537190083
1264
  },
1265
  "cop":{
1266
+ "p":0.7627906977,
1267
+ "r":0.7663551402,
1268
+ "f":0.7645687646
1269
  },
1270
  "root":{
1271
+ "p":0.9332023576,
1272
+ "r":0.9322865554,
1273
+ "f":0.9327442317
1274
  },
1275
  "det":{
1276
+ "p":0.9344978166,
1277
+ "r":0.9280138768,
1278
+ "f":0.9312445605
1279
  },
1280
  "nmod":{
1281
+ "p":0.7352415027,
1282
+ "r":0.7235915493,
1283
+ "f":0.7293700089
1284
  },
1285
  "obj":{
1286
+ "p":0.8082051282,
1287
+ "r":0.8182762201,
1288
+ "f":0.8132094943
1289
  },
1290
  "case":{
1291
+ "p":0.9574700109,
1292
+ "r":0.9627192982,
1293
+ "f":0.9600874795
1294
  },
1295
  "obl":{
1296
+ "p":0.7216117216,
1297
+ "r":0.7216117216,
1298
+ "f":0.7216117216
1299
  },
1300
  "cc":{
1301
+ "p":0.7589013225,
1302
+ "r":0.7674897119,
1303
+ "f":0.7631713555
1304
  },
1305
  "conj":{
1306
+ "p":0.702247191,
1307
+ "r":0.656167979,
1308
+ "f":0.6784260516
1309
  },
1310
  "obl:agent":{
1311
+ "p":0.6153846154,
1312
+ "r":0.4324324324,
1313
+ "f":0.5079365079
1314
  },
1315
  "ccomp":{
1316
+ "p":0.5662100457,
1317
+ "r":0.6169154229,
1318
+ "f":0.5904761905
1319
  },
1320
  "nsubj:pass":{
1321
+ "p":0.6542056075,
1322
+ "r":0.6542056075,
1323
+ "f":0.6542056075
1324
  },
1325
  "amod":{
1326
+ "p":0.711627907,
1327
+ "r":0.7183098592,
1328
+ "f":0.714953271
1329
  },
1330
  "acl":{
1331
+ "p":0.5255474453,
1332
+ "r":0.4363636364,
1333
+ "f":0.4768211921
1334
  },
1335
  "iobj":{
1336
+ "p":0.7505938242,
1337
+ "r":0.7297921478,
1338
+ "f":0.7400468384
1339
  },
1340
  "dep":{
1341
  "p":0.0,
 
1343
  "f":0.0
1344
  },
1345
  "nummod":{
1346
+ "p":0.9152542373,
1347
+ "r":0.7941176471,
1348
+ "f":0.8503937008
1349
  },
1350
  "vocative":{
1351
+ "p":0.8181818182,
1352
+ "r":0.7826086957,
1353
+ "f":0.8
1354
  },
1355
  "orphan":{
1356
+ "p":0.3571428571,
1357
  "r":0.1162790698,
1358
+ "f":0.1754385965
1359
  },
1360
  "appos":{
1361
+ "p":0.4351851852,
1362
+ "r":0.3333333333,
1363
+ "f":0.3775100402
 
 
 
 
 
1364
  },
1365
  "dislocated":{
1366
+ "p":0.0833333333,
1367
+ "r":0.2692307692,
1368
+ "f":0.1272727273
1369
  },
1370
+ "parataxis":{
1371
+ "p":0.1739130435,
1372
+ "r":0.2,
1373
+ "f":0.1860465116
1374
  },
1375
  "flat:name":{
1376
+ "p":0.8125,
1377
  "r":0.5909090909,
1378
+ "f":0.6842105263
1379
+ },
1380
+ "csubj:pass":{
1381
+ "p":0.4,
1382
+ "r":0.4,
1383
+ "f":0.4
1384
  },
1385
  "aux:pass":{
1386
  "p":0.0,
 
1388
  "f":0.0
1389
  },
1390
  "fixed":{
1391
+ "p":0.75,
1392
+ "r":0.6,
1393
+ "f":0.6666666667
1394
  },
1395
  "aux":{
1396
  "p":0.0,
 
1398
  "f":0.0
1399
  }
1400
  },
1401
+ "sents_p":0.7440553394,
1402
+ "sents_r":0.8776134625,
1403
+ "sents_f":0.8053345812,
1404
+ "lemma_acc":0.9747801786,
1405
+ "ents_f":0.8515151515,
1406
+ "ents_p":0.8541033435,
1407
+ "ents_r":0.8489425982,
1408
  "ents_per_type":{
1409
+ "LOC":{
1410
+ "p":0.8418079096,
1411
+ "r":0.6711711712,
1412
+ "f":0.7468671679
1413
+ },
1414
  "NORP":{
1415
+ "p":0.8620689655,
1416
+ "r":0.9014423077,
1417
+ "f":0.8813160987
1418
  },
1419
  "PERSON":{
1420
+ "p":0.8734567901,
1421
+ "r":0.9099678457,
1422
+ "f":0.8913385827
1423
  },
1424
+ "GPE":{
1425
+ "p":0.4,
1426
+ "r":0.5,
1427
+ "f":0.4444444444
1428
  },
1429
  "GOD":{
1430
+ "p":0.6808510638,
1431
+ "r":0.6037735849,
1432
+ "f":0.64
1433
+ },
1434
+ "WORK":{
1435
+ "p":0.0,
1436
+ "r":0.0,
1437
+ "f":0.0
1438
+ },
1439
+ "ORG":{
1440
+ "p":0.0,
1441
+ "r":0.0,
1442
+ "f":0.0
1443
+ },
1444
+ "LANGUAGE":{
1445
+ "p":0.0,
1446
+ "r":0.0,
1447
+ "f":0.0
1448
  }
1449
  },
1450
+ "tok2vec_loss":476279.6321204497,
1451
+ "morphologizer_loss":3954.7004825324,
1452
+ "tagger_loss":1067.753337322,
1453
+ "parser_loss":43531.8663925835,
1454
+ "ner_loss":396.6605884583
1455
  },
1456
  "requirements":[
1457
 
morphologizer/cfg CHANGED
@@ -27,6 +27,7 @@
27
  "Case=Acc|Gender=Fem|Number=Sing|POS=NOUN":"Case=Acc|Gender=Fem|Number=Sing",
28
  "Aspect=Perf|Mood=Ind|Number=Plur|POS=VERB|Person=3|Tense=Past|VerbForm=Fin|Voice=Act":"Aspect=Perf|Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin|Voice=Act",
29
  "Case=Dat|Gender=Masc|Number=Plur|POS=PRON|PronType=Rcp":"Case=Dat|Gender=Masc|Number=Plur|PronType=Rcp",
 
30
  "Case=Nom|Definite=Def|Gender=Masc|Number=Plur|POS=DET|PronType=Dem":"Case=Nom|Definite=Def|Gender=Masc|Number=Plur|PronType=Dem",
31
  "Case=Nom|Gender=Masc|Number=Plur|POS=NOUN":"Case=Nom|Gender=Masc|Number=Plur",
32
  "Case=Acc|Gender=Masc|Number=Plur|POS=NOUN":"Case=Acc|Gender=Masc|Number=Plur",
@@ -709,6 +710,7 @@
709
  "Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|POS=VERB|Tense=Past|VerbForm=Part|Voice=Act":"Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|Tense=Past|VerbForm=Part|Voice=Act",
710
  "Aspect=Perf|Mood=Sub|Number=Plur|POS=VERB|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid":"Aspect=Perf|Mood=Sub|Number=Plur|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid",
711
  "Case=Dat|Degree=Sup|Gender=Fem|Number=Sing|POS=ADJ":"Case=Dat|Degree=Sup|Gender=Fem|Number=Sing",
 
712
  "Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|POS=VERB|Tense=Past|VerbForm=Part|Voice=Pass":"Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass",
713
  "Case=Acc|Gender=Fem,Masc|Number=Sing|POS=NOUN":"Case=Acc|Gender=Fem,Masc|Number=Sing",
714
  "Case=Dat|Gender=Fem,Masc|Number=Sing|POS=NOUN":"Case=Dat|Gender=Fem,Masc|Number=Sing",
@@ -1058,6 +1060,7 @@
1058
  "Case=Acc|Gender=Fem|Number=Sing|POS=NOUN":92,
1059
  "Aspect=Perf|Mood=Ind|Number=Plur|POS=VERB|Person=3|Tense=Past|VerbForm=Fin|Voice=Act":100,
1060
  "Case=Dat|Gender=Masc|Number=Plur|POS=PRON|PronType=Rcp":95,
 
1061
  "Case=Nom|Definite=Def|Gender=Masc|Number=Plur|POS=DET|PronType=Dem":90,
1062
  "Case=Nom|Gender=Masc|Number=Plur|POS=NOUN":92,
1063
  "Case=Acc|Gender=Masc|Number=Plur|POS=NOUN":92,
@@ -1740,6 +1743,7 @@
1740
  "Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|POS=VERB|Tense=Past|VerbForm=Part|Voice=Act":100,
1741
  "Aspect=Perf|Mood=Sub|Number=Plur|POS=VERB|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid":100,
1742
  "Case=Dat|Degree=Sup|Gender=Fem|Number=Sing|POS=ADJ":84,
 
1743
  "Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|POS=VERB|Tense=Past|VerbForm=Part|Voice=Pass":100,
1744
  "Case=Acc|Gender=Fem,Masc|Number=Sing|POS=NOUN":92,
1745
  "Case=Dat|Gender=Fem,Masc|Number=Sing|POS=NOUN":92,
 
27
  "Case=Acc|Gender=Fem|Number=Sing|POS=NOUN":"Case=Acc|Gender=Fem|Number=Sing",
28
  "Aspect=Perf|Mood=Ind|Number=Plur|POS=VERB|Person=3|Tense=Past|VerbForm=Fin|Voice=Act":"Aspect=Perf|Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin|Voice=Act",
29
  "Case=Dat|Gender=Masc|Number=Plur|POS=PRON|PronType=Rcp":"Case=Dat|Gender=Masc|Number=Plur|PronType=Rcp",
30
+ "POS=PUNCT":"",
31
  "Case=Nom|Definite=Def|Gender=Masc|Number=Plur|POS=DET|PronType=Dem":"Case=Nom|Definite=Def|Gender=Masc|Number=Plur|PronType=Dem",
32
  "Case=Nom|Gender=Masc|Number=Plur|POS=NOUN":"Case=Nom|Gender=Masc|Number=Plur",
33
  "Case=Acc|Gender=Masc|Number=Plur|POS=NOUN":"Case=Acc|Gender=Masc|Number=Plur",
 
710
  "Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|POS=VERB|Tense=Past|VerbForm=Part|Voice=Act":"Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|Tense=Past|VerbForm=Part|Voice=Act",
711
  "Aspect=Perf|Mood=Sub|Number=Plur|POS=VERB|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid":"Aspect=Perf|Mood=Sub|Number=Plur|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid",
712
  "Case=Dat|Degree=Sup|Gender=Fem|Number=Sing|POS=ADJ":"Case=Dat|Degree=Sup|Gender=Fem|Number=Sing",
713
+ "POS=AUX":"",
714
  "Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|POS=VERB|Tense=Past|VerbForm=Part|Voice=Pass":"Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|Tense=Past|VerbForm=Part|Voice=Pass",
715
  "Case=Acc|Gender=Fem,Masc|Number=Sing|POS=NOUN":"Case=Acc|Gender=Fem,Masc|Number=Sing",
716
  "Case=Dat|Gender=Fem,Masc|Number=Sing|POS=NOUN":"Case=Dat|Gender=Fem,Masc|Number=Sing",
 
1060
  "Case=Acc|Gender=Fem|Number=Sing|POS=NOUN":92,
1061
  "Aspect=Perf|Mood=Ind|Number=Plur|POS=VERB|Person=3|Tense=Past|VerbForm=Fin|Voice=Act":100,
1062
  "Case=Dat|Gender=Masc|Number=Plur|POS=PRON|PronType=Rcp":95,
1063
+ "POS=PUNCT":97,
1064
  "Case=Nom|Definite=Def|Gender=Masc|Number=Plur|POS=DET|PronType=Dem":90,
1065
  "Case=Nom|Gender=Masc|Number=Plur|POS=NOUN":92,
1066
  "Case=Acc|Gender=Masc|Number=Plur|POS=NOUN":92,
 
1743
  "Aspect=Perf|Case=Nom|Gender=Neut|Number=Plur|POS=VERB|Tense=Past|VerbForm=Part|Voice=Act":100,
1744
  "Aspect=Perf|Mood=Sub|Number=Plur|POS=VERB|Person=2|Tense=Past|VerbForm=Fin|Voice=Mid":100,
1745
  "Case=Dat|Degree=Sup|Gender=Fem|Number=Sing|POS=ADJ":84,
1746
+ "POS=AUX":87,
1747
  "Aspect=Perf|Case=Dat|Gender=Neut|Number=Sing|POS=VERB|Tense=Past|VerbForm=Part|Voice=Pass":100,
1748
  "Case=Acc|Gender=Fem,Masc|Number=Sing|POS=NOUN":92,
1749
  "Case=Dat|Gender=Fem,Masc|Number=Sing|POS=NOUN":92,
morphologizer/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:091c6ad23ec87beefdb468cf9d2fe63901de31826f7053fba6d7f195246b8650
3
- size 1058262
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:16e20faa17eb8af32d746936646dc788df3bb674c322f66fbf5434bf9db901b1
3
+ size 1060318
ner/model CHANGED
Binary files a/ner/model and b/ner/model differ
 
ner/moves CHANGED
@@ -1 +1 @@
1
- ��moves�{"0":{},"1":{"PERSON":626,"NORP":479,"LOC":257,"GOD":44,"LANGUAGE":4},"2":{"PERSON":626,"NORP":479,"LOC":257,"GOD":44,"LANGUAGE":4},"3":{"PERSON":626,"NORP":479,"LOC":257,"GOD":44,"LANGUAGE":4},"4":{"PERSON":626,"NORP":479,"LOC":257,"GOD":44,"LANGUAGE":4,"":1},"5":{"":1}}�cfg��neg_key�
 
1
+ ��moves��{"0":{},"1":{"PERSON":2290,"NORP":1501,"LOC":790,"GOD":250,"GPE":35,"LANGUAGE":12,"ORG":5,"WORK":2,"EVENT":1},"2":{"PERSON":2290,"NORP":1501,"LOC":790,"GOD":250,"GPE":35,"LANGUAGE":12,"ORG":5,"WORK":2,"EVENT":1},"3":{"PERSON":2290,"NORP":1501,"LOC":790,"GOD":250,"GPE":35,"LANGUAGE":12,"ORG":5,"WORK":2,"EVENT":1},"4":{"PERSON":2290,"NORP":1501,"LOC":790,"GOD":250,"GPE":35,"LANGUAGE":12,"ORG":5,"WORK":2,"EVENT":1,"":1},"5":{"":1}}�cfg��neg_key�
parser/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d48352207fe3c10a8cf055fa5bc68e9740c1dd7cb187c0e2f919a7405f689eb7
3
- size 1782525
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4de9436b4ab36bf0da24006628533783de436357923276858ca896ee71345c34
3
+ size 1783041
parser/moves CHANGED
@@ -1 +1 @@
1
- ��moves�:{"0":{"":99376},"1":{"":72645},"2":{"det":25097,"case":13178,"advmod":8281,"nsubj":8155,"discourse":7820,"advcl":5178,"obj":4646,"obl":4348,"mark":3054,"cc":2782,"iobj":1765,"cop":1464,"nmod":1391,"amod":1350,"nsubj:pass":966,"det||nsubj":961,"xcomp":917,"vocative":752,"nummod":549,"mark||advcl":409,"case||obl":395,"obj||advcl":366,"det||obj":353,"dislocated":261,"acl":254,"obl||advcl":245,"orphan":206,"nmod||nsubj":192,"nsubj||advcl":175,"ccomp":161,"nsubj||ccomp":158,"det||nsubj:pass":140,"advmod||advcl":136,"obj||xcomp":132,"obl:agent":126,"cc||advcl":121,"conj||advcl":118,"det||obl":115,"nmod||obj":108,"parataxis":106,"det||iobj":94,"amod||obj":88,"det||nmod":79,"xcomp||advcl":77,"amod||nsubj":75,"obj||ccomp":71,"iobj||advcl":70,"obl||xcomp":64,"iobj||xcomp":64,"advmod||xcomp":55,"advmod||ccomp":49,"appos||nsubj":47,"obl||ccomp":45,"ccomp||advcl":44,"det||advmod":42,"cc||nsubj":42,"nmod||obl":41,"advmod||advmod":39,"nsubj:pass||advcl":34,"iobj||ccomp":34,"amod||obl":30,"dep":0},"3":{"conj":8819,"cc":8497,"obl":7446,"obj":6636,"nmod":5543,"nsubj":3918,"advcl":3876,"det":3859,"iobj":3825,"xcomp":2226,"ccomp":2144,"discourse":2105,"advmod":1915,"appos":1627,"acl":1443,"amod":1361,"cop":1355,"nsubj:pass":710,"orphan":452,"obl:agent":273,"flat:name":248,"vocative":243,"nummod":240,"acl||obj":174,"acl||nsubj":152,"fixed":148,"appos||nsubj":120,"csubj:pass":118,"nmod||obj":116,"conj||nsubj":113,"parataxis":110,"cc||nsubj":110,"nmod||nsubj":102,"conj||obj":101,"cc||obj":92,"appos||obj":84,"amod||obj":79,"case":69,"conj||obl":65,"cc||obl":63,"amod||nsubj":63,"dislocated":59,"det||obj":54,"acl||obl":53,"appos||obl":48,"conj||xcomp":46,"det||nsubj":45,"cop||xcomp":40,"iobj||xcomp":38,"conj||nmod":37,"obl||xcomp":35,"conj||iobj":35,"cc||nmod":35,"cop||ccomp":34,"cc||iobj":33,"cc||xcomp":32,"dep":0},"4":{"":20653,"ROOT":15014}}�cfg��neg_key�
 
1
+ ��moves�H{"0":{"":99376},"1":{"":87659},"2":{"det":25097,"case":13178,"advmod":8281,"nsubj":8155,"discourse":7820,"advcl":5178,"obj":4646,"obl":4348,"mark":3054,"cc":2782,"iobj":1765,"cop":1464,"nmod":1391,"amod":1350,"nsubj:pass":966,"det||nsubj":961,"xcomp":917,"vocative":752,"nummod":549,"mark||advcl":409,"case||obl":395,"obj||advcl":366,"det||obj":353,"dislocated":261,"acl":254,"obl||advcl":245,"orphan":206,"nmod||nsubj":192,"nsubj||advcl":175,"ccomp":161,"nsubj||ccomp":158,"det||nsubj:pass":140,"advmod||advcl":136,"obj||xcomp":132,"obl:agent":126,"cc||advcl":121,"conj||advcl":118,"det||obl":115,"nmod||obj":108,"parataxis":106,"det||iobj":94,"amod||obj":88,"det||nmod":79,"xcomp||advcl":77,"amod||nsubj":75,"obj||ccomp":71,"iobj||advcl":70,"obl||xcomp":64,"iobj||xcomp":64,"advmod||xcomp":55,"advmod||ccomp":49,"appos||nsubj":47,"obl||ccomp":45,"ccomp||advcl":44,"det||advmod":42,"cc||nsubj":42,"nmod||obl":41,"advmod||advmod":39,"nsubj:pass||advcl":34,"iobj||ccomp":34,"amod||obl":30,"dep":0},"3":{"punct":15014,"conj":8819,"cc":8497,"obl":7446,"obj":6636,"nmod":5543,"nsubj":3918,"advcl":3876,"det":3859,"iobj":3825,"xcomp":2226,"ccomp":2144,"discourse":2105,"advmod":1915,"appos":1627,"acl":1443,"amod":1361,"cop":1355,"nsubj:pass":710,"orphan":452,"obl:agent":273,"flat:name":248,"vocative":243,"nummod":240,"acl||obj":174,"acl||nsubj":152,"fixed":148,"appos||nsubj":120,"csubj:pass":118,"nmod||obj":116,"conj||nsubj":113,"parataxis":110,"cc||nsubj":110,"nmod||nsubj":102,"conj||obj":101,"cc||obj":92,"appos||obj":84,"amod||obj":79,"case":69,"conj||obl":65,"cc||obl":63,"amod||nsubj":63,"dislocated":59,"det||obj":54,"acl||obl":53,"appos||obl":48,"conj||xcomp":46,"det||nsubj":45,"cop||xcomp":40,"iobj||xcomp":38,"conj||nmod":37,"obl||xcomp":35,"conj||iobj":35,"cc||nmod":35,"cop||ccomp":34,"cc||iobj":33,"cc||xcomp":32,"dep":0},"4":{"":67143,"ROOT":15014}}�cfg��neg_key�
tagger/cfg CHANGED
@@ -23,7 +23,8 @@
23
  "Px",
24
  "R-",
25
  "S-",
26
- "V-"
 
27
  ],
28
  "neg_prefix":"!",
29
  "overwrite":false
 
23
  "Px",
24
  "R-",
25
  "S-",
26
+ "V-",
27
+ "Z"
28
  ],
29
  "neg_prefix":"!",
30
  "overwrite":false
tagger/model CHANGED
Binary files a/tagger/model and b/tagger/model differ
 
tok2vec/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:17e9db48d7010b068935322c3b1643d093e949b5185f6a481e709312507ef2e7
3
  size 35970008
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8212fcec1770b5b94659ff3cbb3d3a80a4c12b620489bceaf4fee5f3a929babc
3
  size 35970008
vocab/strings.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6c9c2d8e6fb73c7097f5195f732ace9b799340205e57c60f9c8ba53fe8eed51c
3
- size 19290160
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9dd92795656a608df6b436e44ec4592cfdbcddb2c2b65b2d68ef7a369b132706
3
+ size 22024861