Hugging Face's logo

spacy
/
da_core_news_sm

Token Classification
spaCy
Danish
Eval Results
Model card Files Community
adrianeboyd commited on
Commit
c6d5b0f
•
1 Parent(s): 193ce8e

Update spaCy pipeline

Browse files
Files changed (14) hide show
  1. README.md +28 -28
  2. accuracy.json +164 -164
  3. attribute_ruler/patterns +0 -0
  4. config.cfg +5 -4
  5. da_core_news_sm-any-py3-none-any.whl +2 -2
  6. lemmatizer/model +1 -1
  7. meta.json +167 -167
  8. morphologizer/model +1 -1
  9. ner/model +2 -2
  10. parser/model +1 -1
  11. senter/model +1 -1
  12. tok2vec/model +2 -2
  13. tokenizer +0 -0
  14. vocab/strings.json +2 -2
README.md CHANGED
@@ -14,62 +14,62 @@ model-index:
14
  metrics:
15
  - name: NER Precision
16
  type: precision
17
- value: 0.7570498915
18
  - name: NER Recall
19
  type: recall
20
- value: 0.7270833333
21
  - name: NER F Score
22
  type: f_score
23
- value: 0.7417640808
24
  - task:
25
  name: TAG
26
  type: token-classification
27
  metrics:
28
  - name: TAG (XPOS) Accuracy
29
  type: accuracy
30
- value: 0.9498765073
31
  - task:
32
  name: POS
33
  type: token-classification
34
  metrics:
35
  - name: POS (UPOS) Accuracy
36
  type: accuracy
37
- value: 0.9498765073
38
  - task:
39
  name: MORPH
40
  type: token-classification
41
  metrics:
42
  - name: Morph (UFeats) Accuracy
43
  type: accuracy
44
- value: 0.9343341404
45
  - task:
46
  name: LEMMA
47
  type: token-classification
48
  metrics:
49
  - name: Lemma Accuracy
50
  type: accuracy
51
- value: 0.9449878935
52
  - task:
53
  name: UNLABELED_DEPENDENCIES
54
  type: token-classification
55
  metrics:
56
  - name: Unlabeled Attachment Score (UAS)
57
  type: f_score
58
- value: 0.7988826816
59
  - task:
60
  name: LABELED_DEPENDENCIES
61
  type: token-classification
62
  metrics:
63
  - name: Labeled Attachment Score (LAS)
64
  type: f_score
65
- value: 0.752849162
66
  - task:
67
  name: SENTS
68
  type: token-classification
69
  metrics:
70
  - name: Sentences F-Score
71
  type: f_score
72
- value: 0.884097035
73
  ---
74
  ### Details: https://spacy.io/models/da#da_core_news_sm
75
 
@@ -78,8 +78,8 @@ Danish pipeline optimized for CPU. Components: tok2vec, morphologizer, parser, l
78
  | Feature | Description |
79
  | --- | --- |
80
  | **Name** | `da_core_news_sm` |
81
- | **Version** | `3.4.0` |
82
- | **spaCy** | `>=3.4.0,<3.5.0` |
83
  | **Default Pipeline** | `tok2vec`, `morphologizer`, `parser`, `lemmatizer`, `attribute_ruler`, `ner` |
84
  | **Components** | `tok2vec`, `morphologizer`, `parser`, `lemmatizer`, `senter`, `attribute_ruler`, `ner` |
85
  | **Vectors** | 0 keys, 0 unique vectors (0 dimensions) |
@@ -105,22 +105,22 @@ Danish pipeline optimized for CPU. Components: tok2vec, morphologizer, parser, l
105
 
106
  | Type | Score |
107
  | --- | --- |
108
- | `TOKEN_ACC` | 99.95 |
109
  | `TOKEN_P` | 99.78 |
110
  | `TOKEN_R` | 99.75 |
111
  | `TOKEN_F` | 99.76 |
112
- | `POS_ACC` | 94.99 |
113
- | `MORPH_ACC` | 93.43 |
114
- | `MORPH_MICRO_P` | 95.72 |
115
- | `MORPH_MICRO_R` | 94.69 |
116
- | `MORPH_MICRO_F` | 95.20 |
117
- | `SENTS_P` | 89.62 |
118
- | `SENTS_R` | 87.23 |
119
- | `SENTS_F` | 88.41 |
120
- | `DEP_UAS` | 79.89 |
121
- | `DEP_LAS` | 75.28 |
122
- | `LEMMA_ACC` | 94.50 |
123
- | `TAG_ACC` | 94.99 |
124
- | `ENTS_P` | 75.70 |
125
- | `ENTS_R` | 72.71 |
126
- | `ENTS_F` | 74.18 |
 
14
  metrics:
15
  - name: NER Precision
16
  type: precision
17
+ value: 0.7359307359
18
  - name: NER Recall
19
  type: recall
20
+ value: 0.7083333333
21
  - name: NER F Score
22
  type: f_score
23
+ value: 0.7218683652
24
  - task:
25
  name: TAG
26
  type: token-classification
27
  metrics:
28
  - name: TAG (XPOS) Accuracy
29
  type: accuracy
30
+ value: 0.9510387912
31
  - task:
32
  name: POS
33
  type: token-classification
34
  metrics:
35
  - name: POS (UPOS) Accuracy
36
  type: accuracy
37
+ value: 0.9510387912
38
  - task:
39
  name: MORPH
40
  type: token-classification
41
  metrics:
42
  - name: Morph (UFeats) Accuracy
43
  type: accuracy
44
+ value: 0.9373819555
45
  - task:
46
  name: LEMMA
47
  type: token-classification
48
  metrics:
49
  - name: Lemma Accuracy
50
  type: accuracy
51
+ value: 0.945472155
52
  - task:
53
  name: UNLABELED_DEPENDENCIES
54
  type: token-classification
55
  metrics:
56
  - name: Unlabeled Attachment Score (UAS)
57
  type: f_score
58
+ value: 0.8102850755
59
  - task:
60
  name: LABELED_DEPENDENCIES
61
  type: token-classification
62
  metrics:
63
  - name: Labeled Attachment Score (LAS)
64
  type: f_score
65
+ value: 0.7631770164
66
  - task:
67
  name: SENTS
68
  type: token-classification
69
  metrics:
70
  - name: Sentences F-Score
71
  type: f_score
72
+ value: 0.8890860692
73
  ---
74
  ### Details: https://spacy.io/models/da#da_core_news_sm
75
 
 
78
  | Feature | Description |
79
  | --- | --- |
80
  | **Name** | `da_core_news_sm` |
81
+ | **Version** | `3.5.0` |
82
+ | **spaCy** | `>=3.5.0,<3.6.0` |
83
  | **Default Pipeline** | `tok2vec`, `morphologizer`, `parser`, `lemmatizer`, `attribute_ruler`, `ner` |
84
  | **Components** | `tok2vec`, `morphologizer`, `parser`, `lemmatizer`, `senter`, `attribute_ruler`, `ner` |
85
  | **Vectors** | 0 keys, 0 unique vectors (0 dimensions) |
 
105
 
106
  | Type | Score |
107
  | --- | --- |
108
+ | `TOKEN_ACC` | 99.89 |
109
  | `TOKEN_P` | 99.78 |
110
  | `TOKEN_R` | 99.75 |
111
  | `TOKEN_F` | 99.76 |
112
+ | `POS_ACC` | 95.10 |
113
+ | `MORPH_ACC` | 93.74 |
114
+ | `MORPH_MICRO_P` | 95.81 |
115
+ | `MORPH_MICRO_R` | 94.92 |
116
+ | `MORPH_MICRO_F` | 95.36 |
117
+ | `SENTS_P` | 88.99 |
118
+ | `SENTS_R` | 88.83 |
119
+ | `SENTS_F` | 88.91 |
120
+ | `DEP_UAS` | 81.03 |
121
+ | `DEP_LAS` | 76.32 |
122
+ | `LEMMA_ACC` | 94.55 |
123
+ | `TAG_ACC` | 95.10 |
124
+ | `ENTS_P` | 73.59 |
125
+ | `ENTS_R` | 70.83 |
126
+ | `ENTS_F` | 72.19 |
accuracy.json CHANGED
@@ -1,83 +1,83 @@
1
  {
2
- "token_acc": 0.9994672349,
3
  "token_p": 0.9977732598,
4
  "token_r": 0.9974835463,
5
  "token_f": 0.997628382,
6
- "pos_acc": 0.9498765073,
7
- "morph_acc": 0.9343341404,
8
- "morph_micro_p": 0.9571895091,
9
- "morph_micro_r": 0.9468981879,
10
- "morph_micro_f": 0.9520160369,
11
  "morph_per_feat": {
12
  "Mood": {
13
- "p": 0.9656160458,
14
- "r": 0.9637750238,
15
- "f": 0.9646946565
16
  },
17
  "Tense": {
18
- "p": 0.9488721805,
19
- "r": 0.9503012048,
20
- "f": 0.949586155
21
  },
22
  "VerbForm": {
23
- "p": 0.9404176904,
24
  "r": 0.9369645043,
25
- "f": 0.9386879215
26
  },
27
  "Voice": {
28
- "p": 0.9699924981,
29
- "r": 0.966367713,
30
- "f": 0.9681767128
31
  },
32
  "Definite": {
33
- "p": 0.9533574588,
34
- "r": 0.9367838799,
35
- "f": 0.9449980072
36
  },
37
  "Gender": {
38
- "p": 0.9351914036,
39
- "r": 0.9255566633,
40
- "f": 0.9303490897
41
  },
42
  "Number": {
43
- "p": 0.9526455026,
44
- "r": 0.9392279604,
45
- "f": 0.9458891516
46
  },
47
  "AdpType": {
48
- "p": 0.9982046679,
49
- "r": 0.9832007073,
50
- "f": 0.9906458797
51
  },
52
  "PartType": {
53
  "p": 1.0,
54
- "r": 1.0,
55
- "f": 1.0
56
  },
57
  "Case": {
58
- "p": 0.9710144928,
59
- "r": 0.9526066351,
60
- "f": 0.961722488
61
  },
62
  "Person": {
63
- "p": 0.9735449735,
64
- "r": 0.9804618117,
65
- "f": 0.9769911504
66
  },
67
  "PronType": {
68
- "p": 0.9810231023,
69
- "r": 0.9777960526,
70
- "f": 0.9794069193
71
  },
72
  "NumType": {
73
- "p": 0.9930555556,
74
- "r": 0.9470198675,
75
- "f": 0.9694915254
76
  },
77
  "Degree": {
78
- "p": 0.94375,
79
- "r": 0.9096385542,
80
- "f": 0.9263803681
81
  },
82
  "Reflex": {
83
  "p": 1.0,
@@ -94,162 +94,162 @@
94
  "r": 1.0,
95
  "f": 1.0
96
  },
97
- "Polite": {
98
- "p": 0.6,
99
- "r": 0.75,
100
- "f": 0.6666666667
101
- },
102
  "Foreign": {
103
- "p": 0.5,
104
- "r": 0.3,
105
- "f": 0.375
106
  },
107
  "Abbr": {
108
- "p": 0.0,
109
- "r": 0.0,
110
- "f": 0.0
111
  },
112
  "Style": {
113
  "p": 1.0,
114
  "r": 1.0,
115
  "f": 1.0
 
 
 
 
 
116
  }
117
  },
118
- "sents_p": 0.8961748634,
119
- "sents_r": 0.8723404255,
120
- "sents_f": 0.884097035,
121
- "dep_uas": 0.7988826816,
122
- "dep_las": 0.752849162,
123
  "dep_las_per_type": {
124
  "advmod": {
125
- "p": 0.6567967699,
126
- "r": 0.6892655367,
127
- "f": 0.6726395589
128
  },
129
  "root": {
130
- "p": 0.8094373866,
131
- "r": 0.7907801418,
132
- "f": 0.8
133
  },
134
  "nsubj": {
135
- "p": 0.8292682927,
136
- "r": 0.8248945148,
137
- "f": 0.8270756214
138
  },
139
  "case": {
140
- "p": 0.8843873518,
141
- "r": 0.882642998,
142
- "f": 0.8835143139
143
  },
144
  "obl": {
145
- "p": 0.666136725,
146
- "r": 0.650621118,
147
- "f": 0.6582875098
148
  },
149
  "cc": {
150
- "p": 0.7679083095,
151
- "r": 0.7790697674,
152
- "f": 0.7734487734
153
  },
154
  "conj": {
155
- "p": 0.5661971831,
156
- "r": 0.536,
157
- "f": 0.5506849315
158
  },
159
  "obj": {
160
- "p": 0.7509090909,
161
  "r": 0.8019417476,
162
- "f": 0.7755868545
163
  },
164
  "aux": {
165
- "p": 0.8475073314,
166
- "r": 0.8425655977,
167
- "f": 0.8450292398
168
  },
169
  "acl:relcl": {
170
- "p": 0.5414364641,
171
- "r": 0.5297297297,
172
- "f": 0.5355191257
173
  },
174
  "advmod:lmod": {
175
- "p": 0.676056338,
176
  "r": 0.7164179104,
177
- "f": 0.6956521739
178
  },
179
  "det": {
180
- "p": 0.8937908497,
181
- "r": 0.9011532125,
182
- "f": 0.8974569319
183
  },
184
  "amod": {
185
- "p": 0.7895622896,
186
- "r": 0.8003412969,
187
- "f": 0.7949152542
188
  },
189
  "nmod:poss": {
190
- "p": 0.6060606061,
191
- "r": 0.5940594059,
192
- "f": 0.6
193
  },
194
  "ccomp": {
195
- "p": 0.5588235294,
196
- "r": 0.6129032258,
197
- "f": 0.5846153846
198
  },
199
  "nummod": {
200
- "p": 0.8487394958,
201
- "r": 0.8416666667,
202
- "f": 0.8451882845
203
  },
204
  "flat": {
205
- "p": 0.75625,
206
- "r": 0.8013245033,
207
- "f": 0.7781350482
208
  },
209
  "compound:prt": {
210
- "p": 0.4285714286,
211
- "r": 0.2195121951,
212
- "f": 0.2903225806
213
  },
214
  "advcl": {
215
- "p": 0.5727272727,
216
  "r": 0.5431034483,
217
- "f": 0.5575221239
218
  },
219
  "mark": {
220
- "p": 0.8426501035,
221
- "r": 0.8357289528,
222
- "f": 0.8391752577
223
  },
224
  "cop": {
225
- "p": 0.7526315789,
226
- "r": 0.8171428571,
227
- "f": 0.7835616438
228
  },
229
  "dep": {
230
- "p": 0.1397849462,
231
  "r": 0.2452830189,
232
- "f": 0.1780821918
233
  },
234
  "nmod": {
235
- "p": 0.614,
236
  "r": 0.599609375,
237
- "f": 0.6067193676
238
  },
239
  "iobj": {
240
- "p": 0.75,
241
- "r": 0.4090909091,
242
- "f": 0.5294117647
243
  },
244
  "xcomp": {
245
- "p": 0.5625,
246
  "r": 0.3050847458,
247
- "f": 0.3956043956
248
  },
249
  "list": {
250
- "p": 0.4,
251
  "r": 0.2222222222,
252
- "f": 0.2857142857
253
  },
254
  "vocative": {
255
  "p": 0.0,
@@ -257,24 +257,24 @@
257
  "f": 0.0
258
  },
259
  "fixed": {
260
- "p": 0.8888888889,
261
- "r": 0.7804878049,
262
- "f": 0.8311688312
263
- },
264
- "expl": {
265
- "p": 0.8529411765,
266
- "r": 0.8529411765,
267
- "f": 0.8529411765
268
  },
269
  "appos": {
270
- "p": 0.6153846154,
271
- "r": 0.4848484848,
272
- "f": 0.5423728814
 
 
 
 
 
273
  },
274
  "obl:tmod": {
275
- "p": 0.625,
276
- "r": 0.2777777778,
277
- "f": 0.3846153846
278
  },
279
  "discourse": {
280
  "p": 0.0,
@@ -287,32 +287,32 @@
287
  "f": 0.0
288
  }
289
  },
290
- "lemma_acc": 0.9449878935,
291
- "tag_acc": 0.9498765073,
292
- "ents_p": 0.7570498915,
293
- "ents_r": 0.7270833333,
294
- "ents_f": 0.7417640808,
295
  "ents_per_type": {
296
  "PER": {
297
- "p": 0.8322580645,
298
- "r": 0.7771084337,
299
- "f": 0.8037383178
300
  },
301
  "ORG": {
302
- "p": 0.6630434783,
303
- "r": 0.6777777778,
304
- "f": 0.6703296703
305
  },
306
  "MISC": {
307
- "p": 0.6788990826,
308
  "r": 0.6548672566,
309
- "f": 0.6666666667
310
  },
311
  "LOC": {
312
- "p": 0.8095238095,
313
- "r": 0.7657657658,
314
- "f": 0.787037037
315
  }
316
  },
317
- "speed": 13598.0305859988
318
  }
 
1
  {
2
+ "token_acc": 0.9989350373,
3
  "token_p": 0.9977732598,
4
  "token_r": 0.9974835463,
5
  "token_f": 0.997628382,
6
+ "pos_acc": 0.9510387912,
7
+ "morph_acc": 0.9373819555,
8
+ "morph_micro_p": 0.9581167839,
9
+ "morph_micro_r": 0.9491696532,
10
+ "morph_micro_f": 0.9536222329,
11
  "morph_per_feat": {
12
  "Mood": {
13
+ "p": 0.9685414681,
14
+ "r": 0.9685414681,
15
+ "f": 0.9685414681
16
  },
17
  "Tense": {
18
+ "p": 0.9546485261,
19
+ "r": 0.9510542169,
20
+ "f": 0.9528479819
21
  },
22
  "VerbForm": {
23
+ "p": 0.949751861,
24
  "r": 0.9369645043,
25
+ "f": 0.943314849
26
  },
27
  "Voice": {
28
+ "p": 0.9721175584,
29
+ "r": 0.9641255605,
30
+ "f": 0.9681050657
31
  },
32
  "Definite": {
33
+ "p": 0.9476628046,
34
+ "r": 0.9371789806,
35
+ "f": 0.9423917362
36
  },
37
  "Gender": {
38
+ "p": 0.9298128342,
39
+ "r": 0.9245596544,
40
+ "f": 0.9271788035
41
  },
42
  "Number": {
43
+ "p": 0.9525441603,
44
+ "r": 0.9423578508,
45
+ "f": 0.9474236266
46
  },
47
  "AdpType": {
48
+ "p": 0.9991071429,
49
+ "r": 0.9893899204,
50
+ "f": 0.994224789
51
  },
52
  "PartType": {
53
  "p": 1.0,
54
+ "r": 0.9967532468,
55
+ "f": 0.9983739837
56
  },
57
  "Case": {
58
+ "p": 0.986970684,
59
+ "r": 0.9573459716,
60
+ "f": 0.9719326383
61
  },
62
  "Person": {
63
+ "p": 0.9840425532,
64
+ "r": 0.9857904085,
65
+ "f": 0.9849157054
66
  },
67
  "PronType": {
68
+ "p": 0.9884868421,
69
+ "r": 0.9884868421,
70
+ "f": 0.9884868421
71
  },
72
  "NumType": {
73
+ "p": 0.9793103448,
74
+ "r": 0.940397351,
75
+ "f": 0.9594594595
76
  },
77
  "Degree": {
78
+ "p": 0.9371921182,
79
+ "r": 0.9168674699,
80
+ "f": 0.9269183922
81
  },
82
  "Reflex": {
83
  "p": 1.0,
 
94
  "r": 1.0,
95
  "f": 1.0
96
  },
 
 
 
 
 
97
  "Foreign": {
98
+ "p": 0.8,
99
+ "r": 0.4,
100
+ "f": 0.5333333333
101
  },
102
  "Abbr": {
103
+ "p": 1.0,
104
+ "r": 0.2,
105
+ "f": 0.3333333333
106
  },
107
  "Style": {
108
  "p": 1.0,
109
  "r": 1.0,
110
  "f": 1.0
111
+ },
112
+ "Polite": {
113
+ "p": 0.6,
114
+ "r": 0.75,
115
+ "f": 0.6666666667
116
  }
117
  },
118
+ "sents_p": 0.8898756661,
119
+ "sents_r": 0.8882978723,
120
+ "sents_f": 0.8890860692,
121
+ "dep_uas": 0.8102850755,
122
+ "dep_las": 0.7631770164,
123
  "dep_las_per_type": {
124
  "advmod": {
125
+ "p": 0.6776406036,
126
+ "r": 0.697740113,
127
+ "f": 0.6875434934
128
  },
129
  "root": {
130
+ "p": 0.8156028369,
131
+ "r": 0.8156028369,
132
+ "f": 0.8156028369
133
  },
134
  "nsubj": {
135
+ "p": 0.8382978723,
136
+ "r": 0.8312236287,
137
+ "f": 0.8347457627
138
  },
139
  "case": {
140
+ "p": 0.8762278978,
141
+ "r": 0.8796844181,
142
+ "f": 0.8779527559
143
  },
144
  "obl": {
145
+ "p": 0.6805111821,
146
+ "r": 0.6614906832,
147
+ "f": 0.6708661417
148
  },
149
  "cc": {
150
+ "p": 0.7896253602,
151
+ "r": 0.7965116279,
152
+ "f": 0.7930535456
153
  },
154
  "conj": {
155
+ "p": 0.615819209,
156
+ "r": 0.5813333333,
157
+ "f": 0.598079561
158
  },
159
  "obj": {
160
+ "p": 0.7719626168,
161
  "r": 0.8019417476,
162
+ "f": 0.7866666667
163
  },
164
  "aux": {
165
+ "p": 0.8660714286,
166
+ "r": 0.8483965015,
167
+ "f": 0.8571428571
168
  },
169
  "acl:relcl": {
170
+ "p": 0.5833333333,
171
+ "r": 0.5675675676,
172
+ "f": 0.5753424658
173
  },
174
  "advmod:lmod": {
175
+ "p": 0.6857142857,
176
  "r": 0.7164179104,
177
+ "f": 0.700729927
178
  },
179
  "det": {
180
+ "p": 0.9006514658,
181
+ "r": 0.9110378913,
182
+ "f": 0.9058149058
183
  },
184
  "amod": {
185
+ "p": 0.7969283276,
186
+ "r": 0.7969283276,
187
+ "f": 0.7969283276
188
  },
189
  "nmod:poss": {
190
+ "p": 0.6391752577,
191
+ "r": 0.6138613861,
192
+ "f": 0.6262626263
193
  },
194
  "ccomp": {
195
+ "p": 0.606557377,
196
+ "r": 0.5967741935,
197
+ "f": 0.6016260163
198
  },
199
  "nummod": {
200
+ "p": 0.85,
201
+ "r": 0.85,
202
+ "f": 0.85
203
  },
204
  "flat": {
205
+ "p": 0.7543859649,
206
+ "r": 0.8543046358,
207
+ "f": 0.801242236
208
  },
209
  "compound:prt": {
210
+ "p": 0.4166666667,
211
+ "r": 0.243902439,
212
+ "f": 0.3076923077
213
  },
214
  "advcl": {
215
+ "p": 0.5887850467,
216
  "r": 0.5431034483,
217
+ "f": 0.5650224215
218
  },
219
  "mark": {
220
+ "p": 0.8661087866,
221
+ "r": 0.8501026694,
222
+ "f": 0.8580310881
223
  },
224
  "cop": {
225
+ "p": 0.7777777778,
226
+ "r": 0.84,
227
+ "f": 0.8076923077
228
  },
229
  "dep": {
230
+ "p": 0.1382978723,
231
  "r": 0.2452830189,
232
+ "f": 0.1768707483
233
  },
234
  "nmod": {
235
+ "p": 0.6055226824,
236
  "r": 0.599609375,
237
+ "f": 0.6025515211
238
  },
239
  "iobj": {
240
+ "p": 0.7142857143,
241
+ "r": 0.4545454545,
242
+ "f": 0.5555555556
243
  },
244
  "xcomp": {
245
+ "p": 0.4615384615,
246
  "r": 0.3050847458,
247
+ "f": 0.3673469388
248
  },
249
  "list": {
250
+ "p": 0.2666666667,
251
  "r": 0.2222222222,
252
+ "f": 0.2424242424
253
  },
254
  "vocative": {
255
  "p": 0.0,
 
257
  "f": 0.0
258
  },
259
  "fixed": {
260
+ "p": 0.9189189189,
261
+ "r": 0.8292682927,
262
+ "f": 0.8717948718
 
 
 
 
 
263
  },
264
  "appos": {
265
+ "p": 0.4516129032,
266
+ "r": 0.4242424242,
267
+ "f": 0.4375
268
+ },
269
+ "expl": {
270
+ "p": 0.8484848485,
271
+ "r": 0.8235294118,
272
+ "f": 0.8358208955
273
  },
274
  "obl:tmod": {
275
+ "p": 0.75,
276
+ "r": 0.3333333333,
277
+ "f": 0.4615384615
278
  },
279
  "discourse": {
280
  "p": 0.0,
 
287
  "f": 0.0
288
  }
289
  },
290
+ "lemma_acc": 0.945472155,
291
+ "tag_acc": 0.9510387912,
292
+ "ents_p": 0.7359307359,
293
+ "ents_r": 0.7083333333,
294
+ "ents_f": 0.7218683652,
295
  "ents_per_type": {
296
  "PER": {
297
+ "p": 0.7898089172,
298
+ "r": 0.7469879518,
299
+ "f": 0.7678018576
300
  },
301
  "ORG": {
302
+ "p": 0.6413043478,
303
+ "r": 0.6555555556,
304
+ "f": 0.6483516484
305
  },
306
  "MISC": {
307
+ "p": 0.6434782609,
308
  "r": 0.6548672566,
309
+ "f": 0.649122807
310
  },
311
  "LOC": {
312
+ "p": 0.8469387755,
313
+ "r": 0.7477477477,
314
+ "f": 0.7942583732
315
  }
316
  },
317
+ "speed": 12639.5619213188
318
  }
attribute_ruler/patterns CHANGED
Binary files a/attribute_ruler/patterns and b/attribute_ruler/patterns differ
 
config.cfg CHANGED
@@ -81,8 +81,8 @@ nO = null
81
  [components.ner.model.tok2vec.embed]
82
  @architectures = "spacy.MultiHashEmbed.v2"
83
  width = 96
84
- attrs = ["NORM","PREFIX","SUFFIX","SHAPE","SPACY"]
85
- rows = [5000,1000,2500,2500,50]
86
  include_static_vectors = false
87
 
88
  [components.ner.model.tok2vec.encode]
@@ -150,8 +150,8 @@ factory = "tok2vec"
150
  [components.tok2vec.model.embed]
151
  @architectures = "spacy.MultiHashEmbed.v2"
152
  width = ${components.tok2vec.model.encode:width}
153
- attrs = ["NORM","PREFIX","SUFFIX","SHAPE","SPACY"]
154
- rows = [5000,1000,2500,2500,50]
155
  include_static_vectors = false
156
 
157
  [components.tok2vec.model.encode]
@@ -193,6 +193,7 @@ eval_frequency = 1000
193
  frozen_components = []
194
  before_to_disk = null
195
  annotating_components = []
 
196
 
197
  [training.batcher]
198
  @batchers = "spacy.batch_by_words.v1"
 
81
  [components.ner.model.tok2vec.embed]
82
  @architectures = "spacy.MultiHashEmbed.v2"
83
  width = 96
84
+ attrs = ["NORM","PREFIX","SUFFIX","SHAPE"]
85
+ rows = [5000,1000,2500,2500]
86
  include_static_vectors = false
87
 
88
  [components.ner.model.tok2vec.encode]
 
150
  [components.tok2vec.model.embed]
151
  @architectures = "spacy.MultiHashEmbed.v2"
152
  width = ${components.tok2vec.model.encode:width}
153
+ attrs = ["NORM","PREFIX","SUFFIX","SHAPE","SPACY","IS_SPACE"]
154
+ rows = [5000,1000,2500,2500,50,50]
155
  include_static_vectors = false
156
 
157
  [components.tok2vec.model.encode]
 
193
  frozen_components = []
194
  before_to_disk = null
195
  annotating_components = []
196
+ before_update = null
197
 
198
  [training.batcher]
199
  @batchers = "spacy.batch_by_words.v1"
da_core_news_sm-any-py3-none-any.whl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:124e1def77953fb5d5b9c917e38e4fed57febc09604aed213097218864c6aeb9
3
- size 12365943
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8711ed2d45dc1c50046b151286334927502d240125fe7b1e6b8361ee3a97f20f
3
+ size 12388647
lemmatizer/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:47a2197d6bc08e018a3c18c7096368b942cf9e636b4b7ce431544b9c4d1e8c89
3
  size 175818
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc6455a53758bb69d4af35e188f013fbc91acbafd9147c6abcabb4847fe9d0a0
3
  size 175818
meta.json CHANGED
@@ -1,14 +1,14 @@
1
  {
2
  "lang":"da",
3
  "name":"core_news_sm",
4
- "version":"3.4.0",
5
  "description":"Danish pipeline optimized for CPU. Components: tok2vec, morphologizer, parser, lemmatizer (trainable_lemmatizer), senter, ner, attribute_ruler.",
6
  "author":"Explosion",
7
  "email":"[email protected]",
8
  "url":"https://explosion.ai",
9
  "license":"CC BY-SA 4.0",
10
- "spacy_version":">=3.4.0,<3.5.0",
11
- "spacy_git_version":"dd038b536",
12
  "vectors":{
13
  "width":0,
14
  "vectors":0,
@@ -244,85 +244,85 @@
244
  "senter"
245
  ],
246
  "performance":{
247
- "token_acc":0.9994672349,
248
  "token_p":0.9977732598,
249
  "token_r":0.9974835463,
250
  "token_f":0.997628382,
251
- "pos_acc":0.9498765073,
252
- "morph_acc":0.9343341404,
253
- "morph_micro_p":0.9571895091,
254
- "morph_micro_r":0.9468981879,
255
- "morph_micro_f":0.9520160369,
256
  "morph_per_feat":{
257
  "Mood":{
258
- "p":0.9656160458,
259
- "r":0.9637750238,
260
- "f":0.9646946565
261
  },
262
  "Tense":{
263
- "p":0.9488721805,
264
- "r":0.9503012048,
265
- "f":0.949586155
266
  },
267
  "VerbForm":{
268
- "p":0.9404176904,
269
  "r":0.9369645043,
270
- "f":0.9386879215
271
  },
272
  "Voice":{
273
- "p":0.9699924981,
274
- "r":0.966367713,
275
- "f":0.9681767128
276
  },
277
  "Definite":{
278
- "p":0.9533574588,
279
- "r":0.9367838799,
280
- "f":0.9449980072
281
  },
282
  "Gender":{
283
- "p":0.9351914036,
284
- "r":0.9255566633,
285
- "f":0.9303490897
286
  },
287
  "Number":{
288
- "p":0.9526455026,
289
- "r":0.9392279604,
290
- "f":0.9458891516
291
  },
292
  "AdpType":{
293
- "p":0.9982046679,
294
- "r":0.9832007073,
295
- "f":0.9906458797
296
  },
297
  "PartType":{
298
  "p":1.0,
299
- "r":1.0,
300
- "f":1.0
301
  },
302
  "Case":{
303
- "p":0.9710144928,
304
- "r":0.9526066351,
305
- "f":0.961722488
306
  },
307
  "Person":{
308
- "p":0.9735449735,
309
- "r":0.9804618117,
310
- "f":0.9769911504
311
  },
312
  "PronType":{
313
- "p":0.9810231023,
314
- "r":0.9777960526,
315
- "f":0.9794069193
316
  },
317
  "NumType":{
318
- "p":0.9930555556,
319
- "r":0.9470198675,
320
- "f":0.9694915254
321
  },
322
  "Degree":{
323
- "p":0.94375,
324
- "r":0.9096385542,
325
- "f":0.9263803681
326
  },
327
  "Reflex":{
328
  "p":1.0,
@@ -339,162 +339,162 @@
339
  "r":1.0,
340
  "f":1.0
341
  },
342
- "Polite":{
343
- "p":0.6,
344
- "r":0.75,
345
- "f":0.6666666667
346
- },
347
  "Foreign":{
348
- "p":0.5,
349
- "r":0.3,
350
- "f":0.375
351
  },
352
  "Abbr":{
353
- "p":0.0,
354
- "r":0.0,
355
- "f":0.0
356
  },
357
  "Style":{
358
  "p":1.0,
359
  "r":1.0,
360
  "f":1.0
 
 
 
 
 
361
  }
362
  },
363
- "sents_p":0.8961748634,
364
- "sents_r":0.8723404255,
365
- "sents_f":0.884097035,
366
- "dep_uas":0.7988826816,
367
- "dep_las":0.752849162,
368
  "dep_las_per_type":{
369
  "advmod":{
370
- "p":0.6567967699,
371
- "r":0.6892655367,
372
- "f":0.6726395589
373
  },
374
  "root":{
375
- "p":0.8094373866,
376
- "r":0.7907801418,
377
- "f":0.8
378
  },
379
  "nsubj":{
380
- "p":0.8292682927,
381
- "r":0.8248945148,
382
- "f":0.8270756214
383
  },
384
  "case":{
385
- "p":0.8843873518,
386
- "r":0.882642998,
387
- "f":0.8835143139
388
  },
389
  "obl":{
390
- "p":0.666136725,
391
- "r":0.650621118,
392
- "f":0.6582875098
393
  },
394
  "cc":{
395
- "p":0.7679083095,
396
- "r":0.7790697674,
397
- "f":0.7734487734
398
  },
399
  "conj":{
400
- "p":0.5661971831,
401
- "r":0.536,
402
- "f":0.5506849315
403
  },
404
  "obj":{
405
- "p":0.7509090909,
406
  "r":0.8019417476,
407
- "f":0.7755868545
408
  },
409
  "aux":{
410
- "p":0.8475073314,
411
- "r":0.8425655977,
412
- "f":0.8450292398
413
  },
414
  "acl:relcl":{
415
- "p":0.5414364641,
416
- "r":0.5297297297,
417
- "f":0.5355191257
418
  },
419
  "advmod:lmod":{
420
- "p":0.676056338,
421
  "r":0.7164179104,
422
- "f":0.6956521739
423
  },
424
  "det":{
425
- "p":0.8937908497,
426
- "r":0.9011532125,
427
- "f":0.8974569319
428
  },
429
  "amod":{
430
- "p":0.7895622896,
431
- "r":0.8003412969,
432
- "f":0.7949152542
433
  },
434
  "nmod:poss":{
435
- "p":0.6060606061,
436
- "r":0.5940594059,
437
- "f":0.6
438
  },
439
  "ccomp":{
440
- "p":0.5588235294,
441
- "r":0.6129032258,
442
- "f":0.5846153846
443
  },
444
  "nummod":{
445
- "p":0.8487394958,
446
- "r":0.8416666667,
447
- "f":0.8451882845
448
  },
449
  "flat":{
450
- "p":0.75625,
451
- "r":0.8013245033,
452
- "f":0.7781350482
453
  },
454
  "compound:prt":{
455
- "p":0.4285714286,
456
- "r":0.2195121951,
457
- "f":0.2903225806
458
  },
459
  "advcl":{
460
- "p":0.5727272727,
461
  "r":0.5431034483,
462
- "f":0.5575221239
463
  },
464
  "mark":{
465
- "p":0.8426501035,
466
- "r":0.8357289528,
467
- "f":0.8391752577
468
  },
469
  "cop":{
470
- "p":0.7526315789,
471
- "r":0.8171428571,
472
- "f":0.7835616438
473
  },
474
  "dep":{
475
- "p":0.1397849462,
476
  "r":0.2452830189,
477
- "f":0.1780821918
478
  },
479
  "nmod":{
480
- "p":0.614,
481
  "r":0.599609375,
482
- "f":0.6067193676
483
  },
484
  "iobj":{
485
- "p":0.75,
486
- "r":0.4090909091,
487
- "f":0.5294117647
488
  },
489
  "xcomp":{
490
- "p":0.5625,
491
  "r":0.3050847458,
492
- "f":0.3956043956
493
  },
494
  "list":{
495
- "p":0.4,
496
  "r":0.2222222222,
497
- "f":0.2857142857
498
  },
499
  "vocative":{
500
  "p":0.0,
@@ -502,24 +502,24 @@
502
  "f":0.0
503
  },
504
  "fixed":{
505
- "p":0.8888888889,
506
- "r":0.7804878049,
507
- "f":0.8311688312
508
- },
509
- "expl":{
510
- "p":0.8529411765,
511
- "r":0.8529411765,
512
- "f":0.8529411765
513
  },
514
  "appos":{
515
- "p":0.6153846154,
516
- "r":0.4848484848,
517
- "f":0.5423728814
 
 
 
 
 
518
  },
519
  "obl:tmod":{
520
- "p":0.625,
521
- "r":0.2777777778,
522
- "f":0.3846153846
523
  },
524
  "discourse":{
525
  "p":0.0,
@@ -532,34 +532,34 @@
532
  "f":0.0
533
  }
534
  },
535
- "lemma_acc":0.9449878935,
536
- "tag_acc":0.9498765073,
537
- "ents_p":0.7570498915,
538
- "ents_r":0.7270833333,
539
- "ents_f":0.7417640808,
540
  "ents_per_type":{
541
  "PER":{
542
- "p":0.8322580645,
543
- "r":0.7771084337,
544
- "f":0.8037383178
545
  },
546
  "ORG":{
547
- "p":0.6630434783,
548
- "r":0.6777777778,
549
- "f":0.6703296703
550
  },
551
  "MISC":{
552
- "p":0.6788990826,
553
  "r":0.6548672566,
554
- "f":0.6666666667
555
  },
556
  "LOC":{
557
- "p":0.8095238095,
558
- "r":0.7657657658,
559
- "f":0.787037037
560
  }
561
  },
562
- "speed":13598.0305859988
563
  },
564
  "sources":[
565
  {
 
1
  {
2
  "lang":"da",
3
  "name":"core_news_sm",
4
+ "version":"3.5.0",
5
  "description":"Danish pipeline optimized for CPU. Components: tok2vec, morphologizer, parser, lemmatizer (trainable_lemmatizer), senter, ner, attribute_ruler.",
6
  "author":"Explosion",
7
  "email":"[email protected]",
8
  "url":"https://explosion.ai",
9
  "license":"CC BY-SA 4.0",
10
+ "spacy_version":">=3.5.0,<3.6.0",
11
+ "spacy_git_version":"9e0322de1",
12
  "vectors":{
13
  "width":0,
14
  "vectors":0,
 
244
  "senter"
245
  ],
246
  "performance":{
247
+ "token_acc":0.9989350373,
248
  "token_p":0.9977732598,
249
  "token_r":0.9974835463,
250
  "token_f":0.997628382,
251
+ "pos_acc":0.9510387912,
252
+ "morph_acc":0.9373819555,
253
+ "morph_micro_p":0.9581167839,
254
+ "morph_micro_r":0.9491696532,
255
+ "morph_micro_f":0.9536222329,
256
  "morph_per_feat":{
257
  "Mood":{
258
+ "p":0.9685414681,
259
+ "r":0.9685414681,
260
+ "f":0.9685414681
261
  },
262
  "Tense":{
263
+ "p":0.9546485261,
264
+ "r":0.9510542169,
265
+ "f":0.9528479819
266
  },
267
  "VerbForm":{
268
+ "p":0.949751861,
269
  "r":0.9369645043,
270
+ "f":0.943314849
271
  },
272
  "Voice":{
273
+ "p":0.9721175584,
274
+ "r":0.9641255605,
275
+ "f":0.9681050657
276
  },
277
  "Definite":{
278
+ "p":0.9476628046,
279
+ "r":0.9371789806,
280
+ "f":0.9423917362
281
  },
282
  "Gender":{
283
+ "p":0.9298128342,
284
+ "r":0.9245596544,
285
+ "f":0.9271788035
286
  },
287
  "Number":{
288
+ "p":0.9525441603,
289
+ "r":0.9423578508,
290
+ "f":0.9474236266
291
  },
292
  "AdpType":{
293
+ "p":0.9991071429,
294
+ "r":0.9893899204,
295
+ "f":0.994224789
296
  },
297
  "PartType":{
298
  "p":1.0,
299
+ "r":0.9967532468,
300
+ "f":0.9983739837
301
  },
302
  "Case":{
303
+ "p":0.986970684,
304
+ "r":0.9573459716,
305
+ "f":0.9719326383
306
  },
307
  "Person":{
308
+ "p":0.9840425532,
309
+ "r":0.9857904085,
310
+ "f":0.9849157054
311
  },
312
  "PronType":{
313
+ "p":0.9884868421,
314
+ "r":0.9884868421,
315
+ "f":0.9884868421
316
  },
317
  "NumType":{
318
+ "p":0.9793103448,
319
+ "r":0.940397351,
320
+ "f":0.9594594595
321
  },
322
  "Degree":{
323
+ "p":0.9371921182,
324
+ "r":0.9168674699,
325
+ "f":0.9269183922
326
  },
327
  "Reflex":{
328
  "p":1.0,
 
339
  "r":1.0,
340
  "f":1.0
341
  },
 
 
 
 
 
342
  "Foreign":{
343
+ "p":0.8,
344
+ "r":0.4,
345
+ "f":0.5333333333
346
  },
347
  "Abbr":{
348
+ "p":1.0,
349
+ "r":0.2,
350
+ "f":0.3333333333
351
  },
352
  "Style":{
353
  "p":1.0,
354
  "r":1.0,
355
  "f":1.0
356
+ },
357
+ "Polite":{
358
+ "p":0.6,
359
+ "r":0.75,
360
+ "f":0.6666666667
361
  }
362
  },
363
+ "sents_p":0.8898756661,
364
+ "sents_r":0.8882978723,
365
+ "sents_f":0.8890860692,
366
+ "dep_uas":0.8102850755,
367
+ "dep_las":0.7631770164,
368
  "dep_las_per_type":{
369
  "advmod":{
370
+ "p":0.6776406036,
371
+ "r":0.697740113,
372
+ "f":0.6875434934
373
  },
374
  "root":{
375
+ "p":0.8156028369,
376
+ "r":0.8156028369,
377
+ "f":0.8156028369
378
  },
379
  "nsubj":{
380
+ "p":0.8382978723,
381
+ "r":0.8312236287,
382
+ "f":0.8347457627
383
  },
384
  "case":{
385
+ "p":0.8762278978,
386
+ "r":0.8796844181,
387
+ "f":0.8779527559
388
  },
389
  "obl":{
390
+ "p":0.6805111821,
391
+ "r":0.6614906832,
392
+ "f":0.6708661417
393
  },
394
  "cc":{
395
+ "p":0.7896253602,
396
+ "r":0.7965116279,
397
+ "f":0.7930535456
398
  },
399
  "conj":{
400
+ "p":0.615819209,
401
+ "r":0.5813333333,
402
+ "f":0.598079561
403
  },
404
  "obj":{
405
+ "p":0.7719626168,
406
  "r":0.8019417476,
407
+ "f":0.7866666667
408
  },
409
  "aux":{
410
+ "p":0.8660714286,
411
+ "r":0.8483965015,
412
+ "f":0.8571428571
413
  },
414
  "acl:relcl":{
415
+ "p":0.5833333333,
416
+ "r":0.5675675676,
417
+ "f":0.5753424658
418
  },
419
  "advmod:lmod":{
420
+ "p":0.6857142857,
421
  "r":0.7164179104,
422
+ "f":0.700729927
423
  },
424
  "det":{
425
+ "p":0.9006514658,
426
+ "r":0.9110378913,
427
+ "f":0.9058149058
428
  },
429
  "amod":{
430
+ "p":0.7969283276,
431
+ "r":0.7969283276,
432
+ "f":0.7969283276
433
  },
434
  "nmod:poss":{
435
+ "p":0.6391752577,
436
+ "r":0.6138613861,
437
+ "f":0.6262626263
438
  },
439
  "ccomp":{
440
+ "p":0.606557377,
441
+ "r":0.5967741935,
442
+ "f":0.6016260163
443
  },
444
  "nummod":{
445
+ "p":0.85,
446
+ "r":0.85,
447
+ "f":0.85
448
  },
449
  "flat":{
450
+ "p":0.7543859649,
451
+ "r":0.8543046358,
452
+ "f":0.801242236
453
  },
454
  "compound:prt":{
455
+ "p":0.4166666667,
456
+ "r":0.243902439,
457
+ "f":0.3076923077
458
  },
459
  "advcl":{
460
+ "p":0.5887850467,
461
  "r":0.5431034483,
462
+ "f":0.5650224215
463
  },
464
  "mark":{
465
+ "p":0.8661087866,
466
+ "r":0.8501026694,
467
+ "f":0.8580310881
468
  },
469
  "cop":{
470
+ "p":0.7777777778,
471
+ "r":0.84,
472
+ "f":0.8076923077
473
  },
474
  "dep":{
475
+ "p":0.1382978723,
476
  "r":0.2452830189,
477
+ "f":0.1768707483
478
  },
479
  "nmod":{
480
+ "p":0.6055226824,
481
  "r":0.599609375,
482
+ "f":0.6025515211
483
  },
484
  "iobj":{
485
+ "p":0.7142857143,
486
+ "r":0.4545454545,
487
+ "f":0.5555555556
488
  },
489
  "xcomp":{
490
+ "p":0.4615384615,
491
  "r":0.3050847458,
492
+ "f":0.3673469388
493
  },
494
  "list":{
495
+ "p":0.2666666667,
496
  "r":0.2222222222,
497
+ "f":0.2424242424
498
  },
499
  "vocative":{
500
  "p":0.0,
 
502
  "f":0.0
503
  },
504
  "fixed":{
505
+ "p":0.9189189189,
506
+ "r":0.8292682927,
507
+ "f":0.8717948718
 
 
 
 
 
508
  },
509
  "appos":{
510
+ "p":0.4516129032,
511
+ "r":0.4242424242,
512
+ "f":0.4375
513
+ },
514
+ "expl":{
515
+ "p":0.8484848485,
516
+ "r":0.8235294118,
517
+ "f":0.8358208955
518
  },
519
  "obl:tmod":{
520
+ "p":0.75,
521
+ "r":0.3333333333,
522
+ "f":0.4615384615
523
  },
524
  "discourse":{
525
  "p":0.0,
 
532
  "f":0.0
533
  }
534
  },
535
+ "lemma_acc":0.945472155,
536
+ "tag_acc":0.9510387912,
537
+ "ents_p":0.7359307359,
538
+ "ents_r":0.7083333333,
539
+ "ents_f":0.7218683652,
540
  "ents_per_type":{
541
  "PER":{
542
+ "p":0.7898089172,
543
+ "r":0.7469879518,
544
+ "f":0.7678018576
545
  },
546
  "ORG":{
547
+ "p":0.6413043478,
548
+ "r":0.6555555556,
549
+ "f":0.6483516484
550
  },
551
  "MISC":{
552
+ "p":0.6434782609,
553
  "r":0.6548672566,
554
+ "f":0.649122807
555
  },
556
  "LOC":{
557
+ "p":0.8469387755,
558
+ "r":0.7477477477,
559
+ "f":0.7942583732
560
  }
561
  },
562
+ "speed":12639.5619213188
563
  },
564
  "sources":[
565
  {
morphologizer/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9fe59e1aa1713120c71e87e27b774ffa7ea2f35a5ee7f5cc63ad549400aa63c9
3
  size 61739
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4334d44e36e09cb0e38b598dac43e4dae76a4df281a6eff19c4a61db43ffa22
3
  size 61739
ner/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:916ba80bb110c7ebbd93b16fc25580daa660cfb5d8846077b8862b0f0256fab5
3
- size 6270202
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c2280eaa61482d7a9af86f80191a34d12d0d269470d4ef880315a6affb0e076e
3
+ size 6140040
parser/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0930e3053b5d7012a69992b6e134c9e5b220ba85b70b6bdd061b82a568688720
3
  size 308728
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df298826e39ea8519df27e2aaa6a213b8a3888412531ae76dba295248ce2fcea
3
  size 308728
senter/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d785ccd6985e835627d3cf247689b56238463a7d9856ca62c707e9604b437c30
3
  size 197089
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3fb68bbf49de6b527cc6a72b28c0840a0751070ee0b2c5e6725c781d012de3d1
3
  size 197089
tok2vec/model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f70f39f4c4a4da505eb8145a9ab81687ed971091b0d3415acd723ce7b0cd1b98
3
- size 6139229
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8530f45de9657248f80fd340cb8926ee4fa115ae19b94bbe5473c01ff0e656fb
3
+ size 6269370
tokenizer CHANGED
The diff for this file is too large to render. See raw diff
 
vocab/strings.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4732847e3c978ba9af94043f6f556ee73957028a1d53e6cfb3d9b088078b7a91
3
- size 479242
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01d9f5f86bd25b1f987eaa58a5b80d53c3290341bf0b7068f22af81576573504
3
+ size 482127