adrianeboyd
commited on
Commit
·
6f612fd
1
Parent(s):
be92e91
Update spaCy pipeline
Browse files- README.md +28 -28
- accuracy.json +168 -168
- config.cfg +54 -26
- da_core_news_trf-any-py3-none-any.whl +2 -2
- lemmatizer/cfg +172 -178
- lemmatizer/model +2 -2
- lemmatizer/trees +0 -0
- meta.json +173 -173
- morphologizer/model +2 -2
- ner/model +2 -2
- parser/model +2 -2
- transformer/cfg +1 -1
- transformer/model +2 -2
- vocab/strings.json +2 -2
README.md
CHANGED
@@ -14,72 +14,72 @@ model-index:
|
|
14 |
metrics:
|
15 |
- name: NER Precision
|
16 |
type: precision
|
17 |
-
value: 0.
|
18 |
- name: NER Recall
|
19 |
type: recall
|
20 |
-
value: 0.
|
21 |
- name: NER F Score
|
22 |
type: f_score
|
23 |
-
value: 0.
|
24 |
- task:
|
25 |
name: TAG
|
26 |
type: token-classification
|
27 |
metrics:
|
28 |
- name: TAG (XPOS) Accuracy
|
29 |
type: accuracy
|
30 |
-
value: 0.
|
31 |
- task:
|
32 |
name: POS
|
33 |
type: token-classification
|
34 |
metrics:
|
35 |
- name: POS (UPOS) Accuracy
|
36 |
type: accuracy
|
37 |
-
value: 0.
|
38 |
- task:
|
39 |
name: MORPH
|
40 |
type: token-classification
|
41 |
metrics:
|
42 |
- name: Morph (UFeats) Accuracy
|
43 |
type: accuracy
|
44 |
-
value: 0.
|
45 |
- task:
|
46 |
name: LEMMA
|
47 |
type: token-classification
|
48 |
metrics:
|
49 |
- name: Lemma Accuracy
|
50 |
type: accuracy
|
51 |
-
value: 0.
|
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.
|
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.
|
66 |
- task:
|
67 |
name: SENTS
|
68 |
type: token-classification
|
69 |
metrics:
|
70 |
- name: Sentences F-Score
|
71 |
type: f_score
|
72 |
-
value: 0.
|
73 |
---
|
74 |
### Details: https://spacy.io/models/da#da_core_news_trf
|
75 |
|
76 |
-
Danish transformer pipeline (vesteinn/DanskBERT). Components: transformer, morphologizer, parser, lemmatizer (trainable_lemmatizer), ner, attribute_ruler.
|
77 |
|
78 |
| Feature | Description |
|
79 |
| --- | --- |
|
80 |
| **Name** | `da_core_news_trf` |
|
81 |
-
| **Version** | `3.
|
82 |
-
| **spaCy** | `>=3.
|
83 |
| **Default Pipeline** | `transformer`, `morphologizer`, `parser`, `lemmatizer`, `attribute_ruler`, `ner` |
|
84 |
| **Components** | `transformer`, `morphologizer`, `parser`, `lemmatizer`, `attribute_ruler`, `ner` |
|
85 |
| **Vectors** | 0 keys, 0 unique vectors (0 dimensions) |
|
@@ -109,18 +109,18 @@ Danish transformer pipeline (vesteinn/DanskBERT). Components: transformer, morph
|
|
109 |
| `TOKEN_P` | 99.78 |
|
110 |
| `TOKEN_R` | 99.75 |
|
111 |
| `TOKEN_F` | 99.76 |
|
112 |
-
| `POS_ACC` | 98.
|
113 |
-
| `MORPH_ACC` | 98.
|
114 |
-
| `MORPH_MICRO_P` | 99.
|
115 |
-
| `MORPH_MICRO_R` | 98.
|
116 |
-
| `MORPH_MICRO_F` |
|
117 |
-
| `SENTS_P` |
|
118 |
-
| `SENTS_R` |
|
119 |
-
| `SENTS_F` |
|
120 |
-
| `DEP_UAS` |
|
121 |
-
| `DEP_LAS` |
|
122 |
-
| `LEMMA_ACC` | 95.
|
123 |
-
| `TAG_ACC` | 98.
|
124 |
-
| `ENTS_P` |
|
125 |
-
| `ENTS_R` |
|
126 |
-
| `ENTS_F` |
|
|
|
14 |
metrics:
|
15 |
- name: NER Precision
|
16 |
type: precision
|
17 |
+
value: 0.8866396761
|
18 |
- name: NER Recall
|
19 |
type: recall
|
20 |
+
value: 0.9125
|
21 |
- name: NER F Score
|
22 |
type: f_score
|
23 |
+
value: 0.8993839836
|
24 |
- task:
|
25 |
name: TAG
|
26 |
type: token-classification
|
27 |
metrics:
|
28 |
- name: TAG (XPOS) Accuracy
|
29 |
type: accuracy
|
30 |
+
value: 0.9870683392
|
31 |
- task:
|
32 |
name: POS
|
33 |
type: token-classification
|
34 |
metrics:
|
35 |
- name: POS (UPOS) Accuracy
|
36 |
type: accuracy
|
37 |
+
value: 0.9870683392
|
38 |
- task:
|
39 |
name: MORPH
|
40 |
type: token-classification
|
41 |
metrics:
|
42 |
- name: Morph (UFeats) Accuracy
|
43 |
type: accuracy
|
44 |
+
value: 0.9845498135
|
45 |
- task:
|
46 |
name: LEMMA
|
47 |
type: token-classification
|
48 |
metrics:
|
49 |
- name: Lemma Accuracy
|
50 |
type: accuracy
|
51 |
+
value: 0.9579661017
|
52 |
- task:
|
53 |
name: UNLABELED_DEPENDENCIES
|
54 |
type: token-classification
|
55 |
metrics:
|
56 |
- name: Unlabeled Attachment Score (UAS)
|
57 |
type: f_score
|
58 |
+
value: 0.897934115
|
59 |
- task:
|
60 |
name: LABELED_DEPENDENCIES
|
61 |
type: token-classification
|
62 |
metrics:
|
63 |
- name: Labeled Attachment Score (LAS)
|
64 |
type: f_score
|
65 |
+
value: 0.8740509156
|
66 |
- task:
|
67 |
name: SENTS
|
68 |
type: token-classification
|
69 |
metrics:
|
70 |
- name: Sentences F-Score
|
71 |
type: f_score
|
72 |
+
value: 0.9302736099
|
73 |
---
|
74 |
### Details: https://spacy.io/models/da#da_core_news_trf
|
75 |
|
76 |
+
Danish transformer pipeline (Transformer(name='vesteinn/DanskBERT', piece_encoder='xlm-roberta-sentencepiece', stride=120, type='xlm-roberta', width=768, window=152, vocab_size=50005)). Components: transformer, morphologizer, parser, lemmatizer (trainable_lemmatizer), ner, attribute_ruler.
|
77 |
|
78 |
| Feature | Description |
|
79 |
| --- | --- |
|
80 |
| **Name** | `da_core_news_trf` |
|
81 |
+
| **Version** | `3.7.2` |
|
82 |
+
| **spaCy** | `>=3.7.0,<3.8.0` |
|
83 |
| **Default Pipeline** | `transformer`, `morphologizer`, `parser`, `lemmatizer`, `attribute_ruler`, `ner` |
|
84 |
| **Components** | `transformer`, `morphologizer`, `parser`, `lemmatizer`, `attribute_ruler`, `ner` |
|
85 |
| **Vectors** | 0 keys, 0 unique vectors (0 dimensions) |
|
|
|
109 |
| `TOKEN_P` | 99.78 |
|
110 |
| `TOKEN_R` | 99.75 |
|
111 |
| `TOKEN_F` | 99.76 |
|
112 |
+
| `POS_ACC` | 98.71 |
|
113 |
+
| `MORPH_ACC` | 98.45 |
|
114 |
+
| `MORPH_MICRO_P` | 99.30 |
|
115 |
+
| `MORPH_MICRO_R` | 98.86 |
|
116 |
+
| `MORPH_MICRO_F` | 99.08 |
|
117 |
+
| `SENTS_P` | 92.62 |
|
118 |
+
| `SENTS_R` | 93.44 |
|
119 |
+
| `SENTS_F` | 93.03 |
|
120 |
+
| `DEP_UAS` | 89.79 |
|
121 |
+
| `DEP_LAS` | 87.41 |
|
122 |
+
| `LEMMA_ACC` | 95.80 |
|
123 |
+
| `TAG_ACC` | 98.71 |
|
124 |
+
| `ENTS_P` | 88.66 |
|
125 |
+
| `ENTS_R` | 91.25 |
|
126 |
+
| `ENTS_F` | 89.94 |
|
accuracy.json
CHANGED
@@ -3,51 +3,51 @@
|
|
3 |
"token_p": 0.9977732598,
|
4 |
"token_r": 0.9974835463,
|
5 |
"token_f": 0.997628382,
|
6 |
-
"pos_acc": 0.
|
7 |
-
"morph_acc": 0.
|
8 |
-
"morph_micro_p": 0.
|
9 |
-
"morph_micro_r": 0.
|
10 |
-
"morph_micro_f": 0.
|
11 |
"morph_per_feat": {
|
12 |
"Mood": {
|
13 |
-
"p": 0.
|
14 |
"r": 0.9942802669,
|
15 |
-
"f": 0.
|
16 |
},
|
17 |
"Tense": {
|
18 |
-
"p": 0.
|
19 |
-
"r": 0.
|
20 |
-
"f": 0.
|
21 |
},
|
22 |
"VerbForm": {
|
23 |
-
"p": 0.
|
24 |
-
"r": 0.
|
25 |
-
"f": 0.
|
26 |
},
|
27 |
"Voice": {
|
28 |
-
"p": 0.
|
29 |
-
"r": 0.
|
30 |
-
"f": 0.
|
31 |
},
|
32 |
"Definite": {
|
33 |
-
"p": 0.
|
34 |
-
"r": 0.
|
35 |
-
"f": 0.
|
36 |
},
|
37 |
"Gender": {
|
38 |
-
"p": 0.
|
39 |
-
"r": 0.
|
40 |
-
"f": 0.
|
41 |
},
|
42 |
"Number": {
|
43 |
-
"p": 0.
|
44 |
-
"r": 0.
|
45 |
-
"f": 0.
|
46 |
},
|
47 |
"AdpType": {
|
48 |
-
"p": 0.
|
49 |
-
"r": 0.
|
50 |
-
"f": 0.
|
51 |
},
|
52 |
"PartType": {
|
53 |
"p": 1.0,
|
@@ -55,29 +55,29 @@
|
|
55 |
"f": 1.0
|
56 |
},
|
57 |
"Case": {
|
58 |
-
"p": 0.
|
59 |
"r": 0.9889415482,
|
60 |
-
"f": 0.
|
61 |
},
|
62 |
"Person": {
|
63 |
-
"p": 0.
|
64 |
-
"r": 0.
|
65 |
-
"f": 0.
|
66 |
},
|
67 |
"PronType": {
|
68 |
-
"p": 0.
|
69 |
-
"r": 0.
|
70 |
-
"f": 0.
|
71 |
},
|
72 |
"NumType": {
|
73 |
-
"p": 0.
|
74 |
-
"r": 0.
|
75 |
-
"f": 0.
|
76 |
},
|
77 |
"Degree": {
|
78 |
-
"p": 0.
|
79 |
-
"r": 0.
|
80 |
-
"f": 0.
|
81 |
},
|
82 |
"Reflex": {
|
83 |
"p": 1.0,
|
@@ -85,19 +85,19 @@
|
|
85 |
"f": 1.0
|
86 |
},
|
87 |
"Number[psor]": {
|
88 |
-
"p": 0
|
89 |
-
"r": 0
|
90 |
-
"f": 0
|
91 |
},
|
92 |
"Poss": {
|
93 |
-
"p": 0
|
94 |
-
"r": 0
|
95 |
-
"f": 0
|
96 |
},
|
97 |
"Foreign": {
|
98 |
-
"p": 0
|
99 |
-
"r": 0.
|
100 |
-
"f": 0.
|
101 |
},
|
102 |
"Abbr": {
|
103 |
"p": 1.0,
|
@@ -115,141 +115,146 @@
|
|
115 |
"f": 1.0
|
116 |
}
|
117 |
},
|
118 |
-
"sents_p": 0.
|
119 |
-
"sents_r": 0.
|
120 |
-
"sents_f": 0.
|
121 |
-
"dep_uas": 0.
|
122 |
-
"dep_las": 0.
|
123 |
"dep_las_per_type": {
|
124 |
"advmod": {
|
125 |
-
"p": 0.
|
126 |
-
"r": 0.
|
127 |
-
"f": 0.
|
128 |
},
|
129 |
"root": {
|
130 |
-
"p": 0.
|
131 |
-
"r": 0.
|
132 |
-
"f": 0.
|
133 |
},
|
134 |
"nsubj": {
|
135 |
-
"p": 0.
|
136 |
-
"r": 0.
|
137 |
-
"f": 0.
|
138 |
},
|
139 |
"case": {
|
140 |
-
"p": 0.
|
141 |
-
"r": 0.
|
142 |
-
"f": 0.
|
143 |
},
|
144 |
"obl": {
|
145 |
-
"p": 0.
|
146 |
-
"r": 0.
|
147 |
-
"f": 0.
|
148 |
},
|
149 |
"cc": {
|
150 |
-
"p": 0.
|
151 |
-
"r": 0.
|
152 |
-
"f": 0.
|
153 |
},
|
154 |
"conj": {
|
155 |
-
"p": 0.
|
156 |
-
"r": 0.
|
157 |
-
"f": 0.
|
158 |
},
|
159 |
"obj": {
|
160 |
-
"p": 0.
|
161 |
-
"r": 0.
|
162 |
-
"f": 0.
|
163 |
},
|
164 |
"aux": {
|
165 |
-
"p": 0.
|
166 |
-
"r": 0.
|
167 |
-
"f": 0.
|
168 |
},
|
169 |
"acl:relcl": {
|
170 |
-
"p": 0.
|
171 |
-
"r": 0.
|
172 |
-
"f": 0.
|
173 |
},
|
174 |
"advmod:lmod": {
|
175 |
-
"p": 0.
|
176 |
-
"r": 0.
|
177 |
-
"f": 0.
|
178 |
},
|
179 |
"det": {
|
180 |
-
"p": 0.
|
181 |
"r": 0.9456342669,
|
182 |
-
"f": 0.
|
183 |
},
|
184 |
"amod": {
|
185 |
-
"p": 0.
|
186 |
-
"r": 0.
|
187 |
-
"f": 0.
|
188 |
},
|
189 |
"nmod:poss": {
|
190 |
-
"p": 0.
|
191 |
-
"r": 0.
|
192 |
-
"f": 0.
|
193 |
},
|
194 |
"ccomp": {
|
195 |
-
"p": 0.
|
196 |
-
"r": 0.
|
197 |
-
"f": 0.
|
198 |
},
|
199 |
"nummod": {
|
200 |
-
"p": 0.
|
201 |
-
"r": 0.
|
202 |
-
"f": 0.
|
203 |
},
|
204 |
"flat": {
|
205 |
-
"p": 0.
|
206 |
-
"r": 0.
|
207 |
-
"f": 0.
|
208 |
},
|
209 |
"compound:prt": {
|
210 |
-
"p": 0.
|
211 |
-
"r": 0.
|
212 |
-
"f": 0.
|
213 |
},
|
214 |
"advcl": {
|
215 |
-
"p": 0.
|
216 |
-
"r": 0.
|
217 |
-
"f": 0.
|
218 |
},
|
219 |
"mark": {
|
220 |
-
"p": 0.
|
221 |
-
"r": 0.
|
222 |
-
"f": 0.
|
223 |
},
|
224 |
"cop": {
|
225 |
-
"p": 0.
|
226 |
-
"r": 0.
|
227 |
-
"f": 0.
|
228 |
},
|
229 |
"dep": {
|
230 |
-
"p": 0.
|
231 |
-
"r": 0.
|
232 |
-
"f": 0.
|
233 |
},
|
234 |
"nmod": {
|
235 |
-
"p": 0.
|
236 |
-
"r": 0.
|
237 |
-
"f": 0.
|
238 |
},
|
239 |
"iobj": {
|
240 |
-
"p": 0.
|
241 |
-
"r": 0.
|
242 |
-
"f": 0.
|
243 |
},
|
244 |
"xcomp": {
|
245 |
-
"p": 0.
|
246 |
-
"r": 0.
|
247 |
-
"f": 0.
|
|
|
|
|
|
|
|
|
|
|
248 |
},
|
249 |
"list": {
|
250 |
-
"p": 0.
|
251 |
-
"r": 0.
|
252 |
-
"f": 0.
|
253 |
},
|
254 |
"vocative": {
|
255 |
"p": 0.0,
|
@@ -257,62 +262,57 @@
|
|
257 |
"f": 0.0
|
258 |
},
|
259 |
"fixed": {
|
260 |
-
"p": 0.
|
261 |
-
"r": 0.
|
262 |
-
"f": 0.
|
263 |
},
|
264 |
"expl": {
|
265 |
-
"p": 0.
|
266 |
"r": 0.9117647059,
|
267 |
-
"f": 0.
|
268 |
},
|
269 |
"appos": {
|
270 |
-
"p": 0.
|
271 |
-
"r": 0.
|
272 |
-
"f": 0.
|
273 |
},
|
274 |
"obl:tmod": {
|
275 |
-
"p": 0
|
276 |
-
"r": 0.
|
277 |
-
"f": 0.
|
278 |
},
|
279 |
"discourse": {
|
280 |
"p": 0.0,
|
281 |
"r": 0.0,
|
282 |
"f": 0.0
|
283 |
-
},
|
284 |
-
"obl:lmod": {
|
285 |
-
"p": 0.0,
|
286 |
-
"r": 0.0,
|
287 |
-
"f": 0.0
|
288 |
}
|
289 |
},
|
290 |
-
"lemma_acc": 0.
|
291 |
-
"tag_acc": 0.
|
292 |
-
"ents_p": 0.
|
293 |
-
"ents_r": 0.
|
294 |
-
"ents_f": 0.
|
295 |
"ents_per_type": {
|
|
|
|
|
|
|
|
|
|
|
296 |
"PER": {
|
297 |
-
"p": 0.
|
298 |
-
"r": 0.
|
299 |
-
"f": 0.
|
300 |
},
|
301 |
"ORG": {
|
302 |
-
"p": 0.
|
303 |
-
"r": 0.
|
304 |
-
"f": 0.
|
305 |
-
},
|
306 |
-
"MISC": {
|
307 |
-
"p": 0.7163120567,
|
308 |
-
"r": 0.8938053097,
|
309 |
-
"f": 0.7952755906
|
310 |
},
|
311 |
"LOC": {
|
312 |
-
"p": 0.
|
313 |
"r": 0.9459459459,
|
314 |
-
"f": 0.
|
315 |
}
|
316 |
},
|
317 |
-
"speed":
|
318 |
}
|
|
|
3 |
"token_p": 0.9977732598,
|
4 |
"token_r": 0.9974835463,
|
5 |
"token_f": 0.997628382,
|
6 |
+
"pos_acc": 0.9870683392,
|
7 |
+
"morph_acc": 0.9845498135,
|
8 |
+
"morph_micro_p": 0.9929531052,
|
9 |
+
"morph_micro_r": 0.9886426733,
|
10 |
+
"morph_micro_f": 0.9907932011,
|
11 |
"morph_per_feat": {
|
12 |
"Mood": {
|
13 |
+
"p": 0.9952290076,
|
14 |
"r": 0.9942802669,
|
15 |
+
"f": 0.9947544111
|
16 |
},
|
17 |
"Tense": {
|
18 |
+
"p": 0.9924242424,
|
19 |
+
"r": 0.9864457831,
|
20 |
+
"f": 0.9894259819
|
21 |
},
|
22 |
"VerbForm": {
|
23 |
+
"p": 0.9919901417,
|
24 |
+
"r": 0.9853121175,
|
25 |
+
"f": 0.9886398526
|
26 |
},
|
27 |
"Voice": {
|
28 |
+
"p": 0.9947643979,
|
29 |
+
"r": 0.9940209268,
|
30 |
+
"f": 0.9943925234
|
31 |
},
|
32 |
"Definite": {
|
33 |
+
"p": 0.9948227798,
|
34 |
+
"r": 0.9869616752,
|
35 |
+
"f": 0.9908766363
|
36 |
},
|
37 |
"Gender": {
|
38 |
+
"p": 0.9886439546,
|
39 |
+
"r": 0.9837155201,
|
40 |
+
"f": 0.9861735799
|
41 |
},
|
42 |
"Number": {
|
43 |
+
"p": 0.9918763103,
|
44 |
+
"r": 0.987219614,
|
45 |
+
"f": 0.9895424837
|
46 |
},
|
47 |
"AdpType": {
|
48 |
+
"p": 0.9991134752,
|
49 |
+
"r": 0.9964633068,
|
50 |
+
"f": 0.9977866313
|
51 |
},
|
52 |
"PartType": {
|
53 |
"p": 1.0,
|
|
|
55 |
"f": 1.0
|
56 |
},
|
57 |
"Case": {
|
58 |
+
"p": 0.9968152866,
|
59 |
"r": 0.9889415482,
|
60 |
+
"f": 0.9928628073
|
61 |
},
|
62 |
"Person": {
|
63 |
+
"p": 0.9982238011,
|
64 |
+
"r": 0.9982238011,
|
65 |
+
"f": 0.9982238011
|
66 |
},
|
67 |
"PronType": {
|
68 |
+
"p": 0.9975308642,
|
69 |
+
"r": 0.9967105263,
|
70 |
+
"f": 0.9971205265
|
71 |
},
|
72 |
"NumType": {
|
73 |
+
"p": 0.9867549669,
|
74 |
+
"r": 0.9867549669,
|
75 |
+
"f": 0.9867549669
|
76 |
},
|
77 |
"Degree": {
|
78 |
+
"p": 0.9795918367,
|
79 |
+
"r": 0.9831325301,
|
80 |
+
"f": 0.9813589898
|
81 |
},
|
82 |
"Reflex": {
|
83 |
"p": 1.0,
|
|
|
85 |
"f": 1.0
|
86 |
},
|
87 |
"Number[psor]": {
|
88 |
+
"p": 1.0,
|
89 |
+
"r": 1.0,
|
90 |
+
"f": 1.0
|
91 |
},
|
92 |
"Poss": {
|
93 |
+
"p": 1.0,
|
94 |
+
"r": 1.0,
|
95 |
+
"f": 1.0
|
96 |
},
|
97 |
"Foreign": {
|
98 |
+
"p": 1.0,
|
99 |
+
"r": 0.7,
|
100 |
+
"f": 0.8235294118
|
101 |
},
|
102 |
"Abbr": {
|
103 |
"p": 1.0,
|
|
|
115 |
"f": 1.0
|
116 |
}
|
117 |
},
|
118 |
+
"sents_p": 0.9261862917,
|
119 |
+
"sents_r": 0.9343971631,
|
120 |
+
"sents_f": 0.9302736099,
|
121 |
+
"dep_uas": 0.897934115,
|
122 |
+
"dep_las": 0.8740509156,
|
123 |
"dep_las_per_type": {
|
124 |
"advmod": {
|
125 |
+
"p": 0.8345323741,
|
126 |
+
"r": 0.8192090395,
|
127 |
+
"f": 0.8267997149
|
128 |
},
|
129 |
"root": {
|
130 |
+
"p": 0.9014084507,
|
131 |
+
"r": 0.9078014184,
|
132 |
+
"f": 0.9045936396
|
133 |
},
|
134 |
"nsubj": {
|
135 |
+
"p": 0.9345991561,
|
136 |
+
"r": 0.9345991561,
|
137 |
+
"f": 0.9345991561
|
138 |
},
|
139 |
"case": {
|
140 |
+
"p": 0.9358717435,
|
141 |
+
"r": 0.9211045365,
|
142 |
+
"f": 0.9284294235
|
143 |
},
|
144 |
"obl": {
|
145 |
+
"p": 0.8388625592,
|
146 |
+
"r": 0.8245341615,
|
147 |
+
"f": 0.8316366484
|
148 |
},
|
149 |
"cc": {
|
150 |
+
"p": 0.898255814,
|
151 |
+
"r": 0.898255814,
|
152 |
+
"f": 0.898255814
|
153 |
},
|
154 |
"conj": {
|
155 |
+
"p": 0.816,
|
156 |
+
"r": 0.816,
|
157 |
+
"f": 0.816
|
158 |
},
|
159 |
"obj": {
|
160 |
+
"p": 0.9128787879,
|
161 |
+
"r": 0.9359223301,
|
162 |
+
"f": 0.9242569511
|
163 |
},
|
164 |
"aux": {
|
165 |
+
"p": 0.9176470588,
|
166 |
+
"r": 0.9096209913,
|
167 |
+
"f": 0.9136163982
|
168 |
},
|
169 |
"acl:relcl": {
|
170 |
+
"p": 0.7932960894,
|
171 |
+
"r": 0.7675675676,
|
172 |
+
"f": 0.7802197802
|
173 |
},
|
174 |
"advmod:lmod": {
|
175 |
+
"p": 0.8805970149,
|
176 |
+
"r": 0.8805970149,
|
177 |
+
"f": 0.8805970149
|
178 |
},
|
179 |
"det": {
|
180 |
+
"p": 0.9456342669,
|
181 |
"r": 0.9456342669,
|
182 |
+
"f": 0.9456342669
|
183 |
},
|
184 |
"amod": {
|
185 |
+
"p": 0.8939393939,
|
186 |
+
"r": 0.9061433447,
|
187 |
+
"f": 0.9
|
188 |
},
|
189 |
"nmod:poss": {
|
190 |
+
"p": 0.8,
|
191 |
+
"r": 0.7920792079,
|
192 |
+
"f": 0.7960199005
|
193 |
},
|
194 |
"ccomp": {
|
195 |
+
"p": 0.7105263158,
|
196 |
+
"r": 0.8709677419,
|
197 |
+
"f": 0.7826086957
|
198 |
},
|
199 |
"nummod": {
|
200 |
+
"p": 0.8174603175,
|
201 |
+
"r": 0.8583333333,
|
202 |
+
"f": 0.837398374
|
203 |
},
|
204 |
"flat": {
|
205 |
+
"p": 0.8805031447,
|
206 |
+
"r": 0.9271523179,
|
207 |
+
"f": 0.9032258065
|
208 |
},
|
209 |
"compound:prt": {
|
210 |
+
"p": 0.6666666667,
|
211 |
+
"r": 0.6341463415,
|
212 |
+
"f": 0.65
|
213 |
},
|
214 |
"advcl": {
|
215 |
+
"p": 0.7699115044,
|
216 |
+
"r": 0.75,
|
217 |
+
"f": 0.7598253275
|
218 |
},
|
219 |
"mark": {
|
220 |
+
"p": 0.9371069182,
|
221 |
+
"r": 0.9178644764,
|
222 |
+
"f": 0.9273858921
|
223 |
},
|
224 |
"cop": {
|
225 |
+
"p": 0.9152542373,
|
226 |
+
"r": 0.9257142857,
|
227 |
+
"f": 0.9204545455
|
228 |
},
|
229 |
"dep": {
|
230 |
+
"p": 0.2095238095,
|
231 |
+
"r": 0.4150943396,
|
232 |
+
"f": 0.2784810127
|
233 |
},
|
234 |
"nmod": {
|
235 |
+
"p": 0.784989858,
|
236 |
+
"r": 0.755859375,
|
237 |
+
"f": 0.7701492537
|
238 |
},
|
239 |
"iobj": {
|
240 |
+
"p": 0.9375,
|
241 |
+
"r": 0.6818181818,
|
242 |
+
"f": 0.7894736842
|
243 |
},
|
244 |
"xcomp": {
|
245 |
+
"p": 0.7346938776,
|
246 |
+
"r": 0.6101694915,
|
247 |
+
"f": 0.6666666667
|
248 |
+
},
|
249 |
+
"obl:lmod": {
|
250 |
+
"p": 0.0,
|
251 |
+
"r": 0.0,
|
252 |
+
"f": 0.0
|
253 |
},
|
254 |
"list": {
|
255 |
+
"p": 0.4,
|
256 |
+
"r": 0.3333333333,
|
257 |
+
"f": 0.3636363636
|
258 |
},
|
259 |
"vocative": {
|
260 |
"p": 0.0,
|
|
|
262 |
"f": 0.0
|
263 |
},
|
264 |
"fixed": {
|
265 |
+
"p": 0.9473684211,
|
266 |
+
"r": 0.8780487805,
|
267 |
+
"f": 0.9113924051
|
268 |
},
|
269 |
"expl": {
|
270 |
+
"p": 0.96875,
|
271 |
"r": 0.9117647059,
|
272 |
+
"f": 0.9393939394
|
273 |
},
|
274 |
"appos": {
|
275 |
+
"p": 0.7297297297,
|
276 |
+
"r": 0.8181818182,
|
277 |
+
"f": 0.7714285714
|
278 |
},
|
279 |
"obl:tmod": {
|
280 |
+
"p": 1.0,
|
281 |
+
"r": 0.6111111111,
|
282 |
+
"f": 0.7586206897
|
283 |
},
|
284 |
"discourse": {
|
285 |
"p": 0.0,
|
286 |
"r": 0.0,
|
287 |
"f": 0.0
|
|
|
|
|
|
|
|
|
|
|
288 |
}
|
289 |
},
|
290 |
+
"lemma_acc": 0.9579661017,
|
291 |
+
"tag_acc": 0.9870683392,
|
292 |
+
"ents_p": 0.8866396761,
|
293 |
+
"ents_r": 0.9125,
|
294 |
+
"ents_f": 0.8993839836,
|
295 |
"ents_per_type": {
|
296 |
+
"MISC": {
|
297 |
+
"p": 0.7846153846,
|
298 |
+
"r": 0.9026548673,
|
299 |
+
"f": 0.8395061728
|
300 |
+
},
|
301 |
"PER": {
|
302 |
+
"p": 0.950310559,
|
303 |
+
"r": 0.921686747,
|
304 |
+
"f": 0.9357798165
|
305 |
},
|
306 |
"ORG": {
|
307 |
+
"p": 0.8764044944,
|
308 |
+
"r": 0.8666666667,
|
309 |
+
"f": 0.8715083799
|
|
|
|
|
|
|
|
|
|
|
310 |
},
|
311 |
"LOC": {
|
312 |
+
"p": 0.9210526316,
|
313 |
"r": 0.9459459459,
|
314 |
+
"f": 0.9333333333
|
315 |
}
|
316 |
},
|
317 |
+
"speed": 523.1005752745
|
318 |
}
|
config.cfg
CHANGED
@@ -17,6 +17,7 @@ after_creation = null
|
|
17 |
after_pipeline_creation = null
|
18 |
batch_size = 64
|
19 |
tokenizer = {"@tokenizers":"spacy.Tokenizer.v1"}
|
|
|
20 |
|
21 |
[components]
|
22 |
|
@@ -39,10 +40,11 @@ nO = null
|
|
39 |
normalize = false
|
40 |
|
41 |
[components.lemmatizer.model.tok2vec]
|
42 |
-
@architectures = "spacy-transformers.
|
43 |
-
|
44 |
upstream = "transformer"
|
45 |
pooling = {"@layers":"reduce_mean.v1"}
|
|
|
46 |
|
47 |
[components.morphologizer]
|
48 |
factory = "morphologizer"
|
@@ -57,10 +59,11 @@ nO = null
|
|
57 |
normalize = false
|
58 |
|
59 |
[components.morphologizer.model.tok2vec]
|
60 |
-
@architectures = "spacy-transformers.
|
61 |
-
|
62 |
upstream = "transformer"
|
63 |
pooling = {"@layers":"reduce_mean.v1"}
|
|
|
64 |
|
65 |
[components.ner]
|
66 |
factory = "ner"
|
@@ -79,10 +82,11 @@ use_upper = false
|
|
79 |
nO = null
|
80 |
|
81 |
[components.ner.model.tok2vec]
|
82 |
-
@architectures = "spacy-transformers.
|
83 |
-
|
84 |
upstream = "transformer"
|
85 |
pooling = {"@layers":"reduce_mean.v1"}
|
|
|
86 |
|
87 |
[components.parser]
|
88 |
factory = "parser"
|
@@ -102,32 +106,44 @@ use_upper = false
|
|
102 |
nO = null
|
103 |
|
104 |
[components.parser.model.tok2vec]
|
105 |
-
@architectures = "spacy-transformers.
|
106 |
-
|
107 |
upstream = "transformer"
|
108 |
pooling = {"@layers":"reduce_mean.v1"}
|
|
|
109 |
|
110 |
[components.transformer]
|
111 |
-
factory = "
|
112 |
-
|
113 |
-
|
114 |
|
115 |
[components.transformer.model]
|
116 |
-
|
117 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
118 |
mixed_precision = false
|
119 |
-
|
120 |
-
[components.transformer.model.get_spans]
|
121 |
-
@span_getters = "spacy-transformers.strided_spans.v1"
|
122 |
-
window = 128
|
123 |
-
stride = 96
|
124 |
|
125 |
[components.transformer.model.grad_scaler_config]
|
126 |
|
127 |
-
[components.transformer.model.
|
128 |
-
|
129 |
-
|
130 |
-
|
|
|
131 |
|
132 |
[corpora]
|
133 |
|
@@ -164,11 +180,11 @@ annotating_components = []
|
|
164 |
before_update = null
|
165 |
|
166 |
[training.batcher]
|
167 |
-
@batchers = "spacy.
|
168 |
-
discard_oversize =
|
169 |
-
get_length = null
|
170 |
size = 2000
|
171 |
-
|
|
|
172 |
|
173 |
[training.logger]
|
174 |
@loggers = "spacy.ConsoleLogger.v1"
|
@@ -246,6 +262,18 @@ require = false
|
|
246 |
path = "corpus/labels/parser.json"
|
247 |
require = false
|
248 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
249 |
[initialize.lookups]
|
250 |
@misc = "spacy.LookupsDataLoader.v1"
|
251 |
lang = ${nlp.lang}
|
|
|
17 |
after_pipeline_creation = null
|
18 |
batch_size = 64
|
19 |
tokenizer = {"@tokenizers":"spacy.Tokenizer.v1"}
|
20 |
+
vectors = {"@vectors":"spacy.Vectors.v1"}
|
21 |
|
22 |
[components]
|
23 |
|
|
|
40 |
normalize = false
|
41 |
|
42 |
[components.lemmatizer.model.tok2vec]
|
43 |
+
@architectures = "spacy-curated-transformers.LastTransformerLayerListener.v1"
|
44 |
+
width = ${components.transformer.model.hidden_width}
|
45 |
upstream = "transformer"
|
46 |
pooling = {"@layers":"reduce_mean.v1"}
|
47 |
+
grad_factor = 1.0
|
48 |
|
49 |
[components.morphologizer]
|
50 |
factory = "morphologizer"
|
|
|
59 |
normalize = false
|
60 |
|
61 |
[components.morphologizer.model.tok2vec]
|
62 |
+
@architectures = "spacy-curated-transformers.LastTransformerLayerListener.v1"
|
63 |
+
width = ${components.transformer.model.hidden_width}
|
64 |
upstream = "transformer"
|
65 |
pooling = {"@layers":"reduce_mean.v1"}
|
66 |
+
grad_factor = 1.0
|
67 |
|
68 |
[components.ner]
|
69 |
factory = "ner"
|
|
|
82 |
nO = null
|
83 |
|
84 |
[components.ner.model.tok2vec]
|
85 |
+
@architectures = "spacy-curated-transformers.LastTransformerLayerListener.v1"
|
86 |
+
width = ${components.transformer.model.hidden_width}
|
87 |
upstream = "transformer"
|
88 |
pooling = {"@layers":"reduce_mean.v1"}
|
89 |
+
grad_factor = 1.0
|
90 |
|
91 |
[components.parser]
|
92 |
factory = "parser"
|
|
|
106 |
nO = null
|
107 |
|
108 |
[components.parser.model.tok2vec]
|
109 |
+
@architectures = "spacy-curated-transformers.LastTransformerLayerListener.v1"
|
110 |
+
width = ${components.transformer.model.hidden_width}
|
111 |
upstream = "transformer"
|
112 |
pooling = {"@layers":"reduce_mean.v1"}
|
113 |
+
grad_factor = 1.0
|
114 |
|
115 |
[components.transformer]
|
116 |
+
factory = "curated_transformer"
|
117 |
+
all_layer_outputs = false
|
118 |
+
frozen = false
|
119 |
|
120 |
[components.transformer.model]
|
121 |
+
@architectures = "spacy-curated-transformers.XlmrTransformer.v1"
|
122 |
+
vocab_size = 50005
|
123 |
+
hidden_width = 768
|
124 |
+
piece_encoder = {"@architectures":"spacy-curated-transformers.XlmrSentencepieceEncoder.v1"}
|
125 |
+
attention_probs_dropout_prob = 0.1
|
126 |
+
hidden_act = "gelu"
|
127 |
+
hidden_dropout_prob = 0.1
|
128 |
+
intermediate_width = 3072
|
129 |
+
layer_norm_eps = 0.00001
|
130 |
+
max_position_embeddings = 514
|
131 |
+
model_max_length = 512
|
132 |
+
num_attention_heads = 12
|
133 |
+
num_hidden_layers = 12
|
134 |
+
padding_idx = 1
|
135 |
+
type_vocab_size = 1
|
136 |
+
torchscript = false
|
137 |
mixed_precision = false
|
138 |
+
wrapped_listener = null
|
|
|
|
|
|
|
|
|
139 |
|
140 |
[components.transformer.model.grad_scaler_config]
|
141 |
|
142 |
+
[components.transformer.model.with_spans]
|
143 |
+
@architectures = "spacy-curated-transformers.WithStridedSpans.v1"
|
144 |
+
stride = 120
|
145 |
+
window = 152
|
146 |
+
batch_size = 384
|
147 |
|
148 |
[corpora]
|
149 |
|
|
|
180 |
before_update = null
|
181 |
|
182 |
[training.batcher]
|
183 |
+
@batchers = "spacy.batch_by_words.v1"
|
184 |
+
discard_oversize = false
|
|
|
185 |
size = 2000
|
186 |
+
tolerance = 0.2
|
187 |
+
get_length = null
|
188 |
|
189 |
[training.logger]
|
190 |
@loggers = "spacy.ConsoleLogger.v1"
|
|
|
262 |
path = "corpus/labels/parser.json"
|
263 |
require = false
|
264 |
|
265 |
+
[initialize.components.transformer]
|
266 |
+
|
267 |
+
[initialize.components.transformer.encoder_loader]
|
268 |
+
@model_loaders = "spacy-curated-transformers.HFTransformerEncoderLoader.v1"
|
269 |
+
name = "vesteinn/DanskBERT"
|
270 |
+
revision = "main"
|
271 |
+
|
272 |
+
[initialize.components.transformer.piecer_loader]
|
273 |
+
@model_loaders = "spacy-curated-transformers.HFPieceEncoderLoader.v1"
|
274 |
+
name = "vesteinn/DanskBERT"
|
275 |
+
revision = "main"
|
276 |
+
|
277 |
[initialize.lookups]
|
278 |
@misc = "spacy.LookupsDataLoader.v1"
|
279 |
lang = ${nlp.lang}
|
da_core_news_trf-any-py3-none-any.whl
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:89bfa83b2c7a53376976a12ac112ac91b910990e341b7c46d7faaef5fcc620e2
|
3 |
+
size 440703099
|
lemmatizer/cfg
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
{
|
2 |
"labels":[
|
3 |
-
|
4 |
2,
|
5 |
4,
|
6 |
6,
|
@@ -10,16 +10,15 @@
|
|
10 |
14,
|
11 |
16,
|
12 |
18,
|
13 |
-
|
14 |
-
|
15 |
28,
|
16 |
30,
|
17 |
32,
|
18 |
-
|
19 |
-
|
20 |
39,
|
21 |
41,
|
22 |
-
42,
|
23 |
43,
|
24 |
45,
|
25 |
47,
|
@@ -27,10 +26,10 @@
|
|
27 |
51,
|
28 |
53,
|
29 |
55,
|
30 |
-
|
31 |
-
|
32 |
-
65,
|
33 |
67,
|
|
|
34 |
71,
|
35 |
73,
|
36 |
75,
|
@@ -44,62 +43,60 @@
|
|
44 |
91,
|
45 |
93,
|
46 |
95,
|
47 |
-
|
48 |
101,
|
49 |
-
|
50 |
104,
|
51 |
-
|
|
|
52 |
111,
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
131,
|
61 |
133,
|
62 |
-
|
63 |
136,
|
64 |
138,
|
65 |
140,
|
66 |
142,
|
67 |
144,
|
68 |
-
|
69 |
-
147,
|
70 |
148,
|
71 |
-
|
72 |
-
|
73 |
-
|
|
|
74 |
158,
|
|
|
75 |
161,
|
76 |
164,
|
77 |
-
|
78 |
-
168,
|
79 |
170,
|
80 |
172,
|
81 |
174,
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
182,
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
190,
|
90 |
192,
|
91 |
194,
|
92 |
196,
|
93 |
-
|
94 |
-
|
95 |
-
|
96 |
204,
|
97 |
207,
|
98 |
-
208,
|
99 |
209,
|
100 |
211,
|
101 |
212,
|
102 |
-
213,
|
103 |
215,
|
104 |
217,
|
105 |
219,
|
@@ -107,173 +104,176 @@
|
|
107 |
222,
|
108 |
224,
|
109 |
226,
|
110 |
-
|
|
|
111 |
231,
|
112 |
-
|
113 |
-
234,
|
114 |
235,
|
115 |
-
237,
|
116 |
238,
|
117 |
-
|
|
|
|
|
|
|
|
|
118 |
248,
|
|
|
119 |
252,
|
120 |
-
254,
|
121 |
256,
|
122 |
-
258,
|
123 |
-
260,
|
124 |
262,
|
125 |
-
|
126 |
-
264,
|
127 |
-
265,
|
128 |
268,
|
129 |
270,
|
130 |
-
|
131 |
-
|
132 |
-
|
133 |
277,
|
|
|
134 |
279,
|
135 |
-
|
|
|
136 |
285,
|
137 |
-
|
138 |
-
|
139 |
-
|
140 |
292,
|
141 |
294,
|
142 |
-
|
143 |
-
|
144 |
-
298,
|
145 |
-
299,
|
146 |
301,
|
147 |
303,
|
148 |
-
|
|
|
149 |
309,
|
150 |
-
|
151 |
-
|
|
|
152 |
315,
|
153 |
-
|
154 |
-
318,
|
155 |
319,
|
156 |
-
|
157 |
-
|
158 |
-
324,
|
159 |
326,
|
160 |
327,
|
161 |
329,
|
162 |
-
|
163 |
333,
|
164 |
335,
|
165 |
-
|
|
|
166 |
339,
|
167 |
340,
|
168 |
342,
|
169 |
343,
|
170 |
-
|
171 |
-
|
172 |
350,
|
173 |
-
|
174 |
-
|
175 |
-
|
|
|
176 |
363,
|
177 |
-
|
178 |
-
|
179 |
368,
|
180 |
-
369,
|
181 |
371,
|
182 |
372,
|
183 |
-
|
|
|
184 |
376,
|
185 |
-
|
186 |
380,
|
187 |
383,
|
188 |
386,
|
|
|
189 |
390,
|
190 |
-
|
191 |
-
394,
|
192 |
395,
|
193 |
396,
|
194 |
397,
|
|
|
195 |
399,
|
|
|
196 |
401,
|
197 |
403,
|
198 |
405,
|
199 |
406,
|
200 |
-
|
201 |
409,
|
202 |
-
|
203 |
-
|
|
|
204 |
416,
|
205 |
418,
|
206 |
419,
|
207 |
420,
|
208 |
-
421,
|
209 |
422,
|
210 |
424,
|
211 |
426,
|
212 |
-
|
213 |
430,
|
214 |
-
|
215 |
434,
|
216 |
435,
|
217 |
437,
|
|
|
218 |
439,
|
219 |
441,
|
220 |
-
|
221 |
-
|
222 |
-
|
223 |
-
|
|
|
224 |
451,
|
225 |
453,
|
226 |
454,
|
227 |
456,
|
228 |
457,
|
229 |
459,
|
230 |
-
|
231 |
-
|
232 |
464,
|
233 |
-
|
234 |
-
469,
|
235 |
470,
|
236 |
471,
|
237 |
-
|
238 |
474,
|
239 |
-
475,
|
240 |
477,
|
241 |
-
|
242 |
481,
|
243 |
-
|
244 |
485,
|
|
|
245 |
488,
|
246 |
489,
|
247 |
-
|
248 |
493,
|
249 |
-
|
250 |
-
497,
|
251 |
498,
|
252 |
501,
|
253 |
503,
|
|
|
|
|
254 |
506,
|
|
|
255 |
508,
|
256 |
-
509,
|
257 |
510,
|
258 |
511,
|
259 |
-
|
260 |
514,
|
261 |
-
|
262 |
517,
|
|
|
263 |
519,
|
264 |
521,
|
265 |
522,
|
266 |
-
523,
|
267 |
524,
|
268 |
-
|
269 |
-
527,
|
270 |
528,
|
271 |
530,
|
272 |
-
531,
|
273 |
532,
|
274 |
533,
|
|
|
275 |
535,
|
276 |
-
536,
|
277 |
537,
|
278 |
538,
|
279 |
540,
|
@@ -284,95 +284,95 @@
|
|
284 |
548,
|
285 |
549,
|
286 |
551,
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
560,
|
291 |
561,
|
292 |
562,
|
293 |
563,
|
294 |
-
564,
|
295 |
565,
|
296 |
567,
|
|
|
297 |
569,
|
298 |
570,
|
299 |
572,
|
300 |
574,
|
301 |
575,
|
302 |
-
576,
|
303 |
577,
|
304 |
579,
|
305 |
580,
|
306 |
-
|
307 |
-
|
|
|
308 |
589,
|
309 |
-
|
310 |
-
|
311 |
596,
|
312 |
598,
|
|
|
313 |
602,
|
314 |
-
603,
|
315 |
-
605,
|
316 |
606,
|
|
|
|
|
317 |
610,
|
318 |
-
|
319 |
-
|
320 |
-
|
321 |
-
|
322 |
-
621,
|
323 |
-
622,
|
324 |
623,
|
325 |
-
624,
|
326 |
625,
|
327 |
626,
|
|
|
328 |
628,
|
|
|
329 |
630,
|
330 |
-
|
331 |
-
633,
|
332 |
634,
|
|
|
333 |
636,
|
334 |
-
|
335 |
-
|
|
|
336 |
642,
|
337 |
-
|
338 |
-
|
|
|
339 |
648,
|
340 |
650,
|
341 |
-
|
342 |
655,
|
343 |
656,
|
344 |
658,
|
345 |
-
|
346 |
-
660,
|
347 |
-
662,
|
348 |
663,
|
349 |
664,
|
350 |
666,
|
351 |
-
|
352 |
669,
|
|
|
353 |
671,
|
354 |
673,
|
355 |
-
|
356 |
-
|
357 |
-
678,
|
358 |
679,
|
359 |
-
|
360 |
-
|
361 |
684,
|
362 |
685,
|
363 |
687,
|
364 |
-
|
365 |
-
690,
|
366 |
691,
|
367 |
-
|
368 |
694,
|
369 |
695,
|
|
|
|
|
370 |
698,
|
371 |
-
699,
|
372 |
-
700,
|
373 |
701,
|
|
|
374 |
703,
|
375 |
-
|
376 |
706,
|
377 |
708,
|
378 |
709,
|
@@ -385,72 +385,66 @@
|
|
385 |
725,
|
386 |
727,
|
387 |
730,
|
|
|
388 |
732,
|
389 |
-
|
390 |
735,
|
391 |
-
|
392 |
738,
|
|
|
393 |
740,
|
394 |
741,
|
395 |
742,
|
396 |
-
743,
|
397 |
-
744,
|
398 |
745,
|
399 |
746,
|
400 |
-
747,
|
401 |
748,
|
402 |
-
751,
|
403 |
752,
|
404 |
-
754,
|
405 |
755,
|
|
|
406 |
759,
|
|
|
407 |
762,
|
408 |
-
|
409 |
765,
|
410 |
-
766,
|
411 |
767,
|
|
|
412 |
769,
|
413 |
770,
|
414 |
-
|
415 |
-
774,
|
416 |
775,
|
417 |
-
776,
|
418 |
777,
|
419 |
778,
|
420 |
-
|
421 |
-
|
422 |
-
|
423 |
786,
|
|
|
424 |
789,
|
|
|
425 |
792,
|
426 |
793,
|
427 |
795,
|
|
|
428 |
797,
|
429 |
798,
|
430 |
-
|
431 |
801,
|
432 |
802,
|
433 |
803,
|
434 |
-
804,
|
435 |
805,
|
436 |
807,
|
437 |
808,
|
438 |
-
809,
|
439 |
810,
|
440 |
811,
|
441 |
813,
|
442 |
815,
|
443 |
-
|
444 |
818,
|
445 |
-
|
446 |
-
|
447 |
823,
|
448 |
825,
|
449 |
827,
|
450 |
-
829
|
451 |
-
831,
|
452 |
-
833,
|
453 |
-
835,
|
454 |
-
837
|
455 |
]
|
456 |
}
|
|
|
1 |
{
|
2 |
"labels":[
|
3 |
+
0,
|
4 |
2,
|
5 |
4,
|
6 |
6,
|
|
|
10 |
14,
|
11 |
16,
|
12 |
18,
|
13 |
+
22,
|
14 |
+
26,
|
15 |
28,
|
16 |
30,
|
17 |
32,
|
18 |
+
35,
|
19 |
+
37,
|
20 |
39,
|
21 |
41,
|
|
|
22 |
43,
|
23 |
45,
|
24 |
47,
|
|
|
26 |
51,
|
27 |
53,
|
28 |
55,
|
29 |
+
59,
|
30 |
+
63,
|
|
|
31 |
67,
|
32 |
+
69,
|
33 |
71,
|
34 |
73,
|
35 |
75,
|
|
|
43 |
91,
|
44 |
93,
|
45 |
95,
|
46 |
+
97,
|
47 |
101,
|
48 |
+
103,
|
49 |
104,
|
50 |
+
106,
|
51 |
+
108,
|
52 |
111,
|
53 |
+
115,
|
54 |
+
117,
|
55 |
+
120,
|
56 |
+
123,
|
57 |
+
126,
|
58 |
+
129,
|
59 |
+
130,
|
|
|
60 |
133,
|
61 |
+
135,
|
62 |
136,
|
63 |
138,
|
64 |
140,
|
65 |
142,
|
66 |
144,
|
67 |
+
146,
|
|
|
68 |
148,
|
69 |
+
150,
|
70 |
+
151,
|
71 |
+
152,
|
72 |
+
154,
|
73 |
158,
|
74 |
+
160,
|
75 |
161,
|
76 |
164,
|
77 |
+
167,
|
|
|
78 |
170,
|
79 |
172,
|
80 |
174,
|
81 |
+
176,
|
82 |
+
178,
|
83 |
+
180,
|
84 |
182,
|
85 |
+
183,
|
86 |
+
185,
|
87 |
+
187,
|
88 |
190,
|
89 |
192,
|
90 |
194,
|
91 |
196,
|
92 |
+
198,
|
93 |
+
200,
|
94 |
+
202,
|
95 |
204,
|
96 |
207,
|
|
|
97 |
209,
|
98 |
211,
|
99 |
212,
|
|
|
100 |
215,
|
101 |
217,
|
102 |
219,
|
|
|
104 |
222,
|
105 |
224,
|
106 |
226,
|
107 |
+
228,
|
108 |
+
230,
|
109 |
231,
|
110 |
+
233,
|
|
|
111 |
235,
|
|
|
112 |
238,
|
113 |
+
240,
|
114 |
+
241,
|
115 |
+
243,
|
116 |
+
245,
|
117 |
+
246,
|
118 |
248,
|
119 |
+
250,
|
120 |
252,
|
|
|
121 |
256,
|
|
|
|
|
122 |
262,
|
123 |
+
266,
|
|
|
|
|
124 |
268,
|
125 |
270,
|
126 |
+
272,
|
127 |
+
274,
|
128 |
+
276,
|
129 |
277,
|
130 |
+
278,
|
131 |
279,
|
132 |
+
282,
|
133 |
+
284,
|
134 |
285,
|
135 |
+
287,
|
136 |
+
289,
|
137 |
+
291,
|
138 |
292,
|
139 |
294,
|
140 |
+
296,
|
141 |
+
300,
|
|
|
|
|
142 |
301,
|
143 |
303,
|
144 |
+
305,
|
145 |
+
307,
|
146 |
309,
|
147 |
+
311,
|
148 |
+
313,
|
149 |
+
314,
|
150 |
315,
|
151 |
+
317,
|
|
|
152 |
319,
|
153 |
+
320,
|
154 |
+
323,
|
|
|
155 |
326,
|
156 |
327,
|
157 |
329,
|
158 |
+
332,
|
159 |
333,
|
160 |
335,
|
161 |
+
336,
|
162 |
+
338,
|
163 |
339,
|
164 |
340,
|
165 |
342,
|
166 |
343,
|
167 |
+
345,
|
168 |
+
346,
|
169 |
350,
|
170 |
+
352,
|
171 |
+
353,
|
172 |
+
357,
|
173 |
+
359,
|
174 |
363,
|
175 |
+
366,
|
176 |
+
367,
|
177 |
368,
|
|
|
178 |
371,
|
179 |
372,
|
180 |
+
374,
|
181 |
+
375,
|
182 |
376,
|
183 |
+
379,
|
184 |
380,
|
185 |
383,
|
186 |
386,
|
187 |
+
387,
|
188 |
390,
|
189 |
+
391,
|
|
|
190 |
395,
|
191 |
396,
|
192 |
397,
|
193 |
+
398,
|
194 |
399,
|
195 |
+
400,
|
196 |
401,
|
197 |
403,
|
198 |
405,
|
199 |
406,
|
200 |
+
408,
|
201 |
409,
|
202 |
+
412,
|
203 |
+
414,
|
204 |
+
415,
|
205 |
416,
|
206 |
418,
|
207 |
419,
|
208 |
420,
|
|
|
209 |
422,
|
210 |
424,
|
211 |
426,
|
212 |
+
428,
|
213 |
430,
|
214 |
+
431,
|
215 |
434,
|
216 |
435,
|
217 |
437,
|
218 |
+
438,
|
219 |
439,
|
220 |
441,
|
221 |
+
442,
|
222 |
+
444,
|
223 |
+
446,
|
224 |
+
448,
|
225 |
+
450,
|
226 |
451,
|
227 |
453,
|
228 |
454,
|
229 |
456,
|
230 |
457,
|
231 |
459,
|
232 |
+
461,
|
233 |
+
463,
|
234 |
464,
|
235 |
+
468,
|
|
|
236 |
470,
|
237 |
471,
|
238 |
+
472,
|
239 |
474,
|
|
|
240 |
477,
|
241 |
+
478,
|
242 |
481,
|
243 |
+
482,
|
244 |
485,
|
245 |
+
486,
|
246 |
488,
|
247 |
489,
|
248 |
+
492,
|
249 |
493,
|
250 |
+
496,
|
|
|
251 |
498,
|
252 |
501,
|
253 |
503,
|
254 |
+
504,
|
255 |
+
505,
|
256 |
506,
|
257 |
+
507,
|
258 |
508,
|
|
|
259 |
510,
|
260 |
511,
|
261 |
+
512,
|
262 |
514,
|
263 |
+
516,
|
264 |
517,
|
265 |
+
518,
|
266 |
519,
|
267 |
521,
|
268 |
522,
|
|
|
269 |
524,
|
270 |
+
526,
|
|
|
271 |
528,
|
272 |
530,
|
|
|
273 |
532,
|
274 |
533,
|
275 |
+
534,
|
276 |
535,
|
|
|
277 |
537,
|
278 |
538,
|
279 |
540,
|
|
|
284 |
548,
|
285 |
549,
|
286 |
551,
|
287 |
+
553,
|
288 |
+
557,
|
289 |
+
559,
|
290 |
560,
|
291 |
561,
|
292 |
562,
|
293 |
563,
|
|
|
294 |
565,
|
295 |
567,
|
296 |
+
568,
|
297 |
569,
|
298 |
570,
|
299 |
572,
|
300 |
574,
|
301 |
575,
|
|
|
302 |
577,
|
303 |
579,
|
304 |
580,
|
305 |
+
582,
|
306 |
+
583,
|
307 |
+
587,
|
308 |
589,
|
309 |
+
590,
|
310 |
+
593,
|
311 |
596,
|
312 |
598,
|
313 |
+
600,
|
314 |
602,
|
|
|
|
|
315 |
606,
|
316 |
+
607,
|
317 |
+
608,
|
318 |
610,
|
319 |
+
611,
|
320 |
+
615,
|
321 |
+
617,
|
322 |
+
619,
|
|
|
|
|
323 |
623,
|
|
|
324 |
625,
|
325 |
626,
|
326 |
+
627,
|
327 |
628,
|
328 |
+
629,
|
329 |
630,
|
330 |
+
632,
|
|
|
331 |
634,
|
332 |
+
635,
|
333 |
636,
|
334 |
+
638,
|
335 |
+
639,
|
336 |
+
641,
|
337 |
642,
|
338 |
+
643,
|
339 |
+
644,
|
340 |
+
645,
|
341 |
648,
|
342 |
650,
|
343 |
+
654,
|
344 |
655,
|
345 |
656,
|
346 |
658,
|
347 |
+
661,
|
|
|
|
|
348 |
663,
|
349 |
664,
|
350 |
666,
|
351 |
+
667,
|
352 |
669,
|
353 |
+
670,
|
354 |
671,
|
355 |
673,
|
356 |
+
675,
|
357 |
+
677,
|
|
|
358 |
679,
|
359 |
+
680,
|
360 |
+
682,
|
361 |
684,
|
362 |
685,
|
363 |
687,
|
364 |
+
689,
|
|
|
365 |
691,
|
366 |
+
692,
|
367 |
694,
|
368 |
695,
|
369 |
+
696,
|
370 |
+
697,
|
371 |
698,
|
|
|
|
|
372 |
701,
|
373 |
+
702,
|
374 |
703,
|
375 |
+
704,
|
376 |
706,
|
377 |
708,
|
378 |
709,
|
|
|
385 |
725,
|
386 |
727,
|
387 |
730,
|
388 |
+
731,
|
389 |
732,
|
390 |
+
733,
|
391 |
735,
|
392 |
+
737,
|
393 |
738,
|
394 |
+
739,
|
395 |
740,
|
396 |
741,
|
397 |
742,
|
|
|
|
|
398 |
745,
|
399 |
746,
|
|
|
400 |
748,
|
|
|
401 |
752,
|
|
|
402 |
755,
|
403 |
+
757,
|
404 |
759,
|
405 |
+
760,
|
406 |
762,
|
407 |
+
763,
|
408 |
765,
|
|
|
409 |
767,
|
410 |
+
768,
|
411 |
769,
|
412 |
770,
|
413 |
+
771,
|
|
|
414 |
775,
|
|
|
415 |
777,
|
416 |
778,
|
417 |
+
779,
|
418 |
+
780,
|
419 |
+
783,
|
420 |
786,
|
421 |
+
787,
|
422 |
789,
|
423 |
+
791,
|
424 |
792,
|
425 |
793,
|
426 |
795,
|
427 |
+
796,
|
428 |
797,
|
429 |
798,
|
430 |
+
800,
|
431 |
801,
|
432 |
802,
|
433 |
803,
|
|
|
434 |
805,
|
435 |
807,
|
436 |
808,
|
|
|
437 |
810,
|
438 |
811,
|
439 |
813,
|
440 |
815,
|
441 |
+
816,
|
442 |
818,
|
443 |
+
819,
|
444 |
+
821,
|
445 |
823,
|
446 |
825,
|
447 |
827,
|
448 |
+
829
|
|
|
|
|
|
|
|
|
449 |
]
|
450 |
}
|
lemmatizer/model
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0804a6d8b32f684ed40c9822a1b0883abe89ac3ed7e479934b3dd1b26d56c4d2
|
3 |
+
size 1372633
|
lemmatizer/trees
CHANGED
Binary files a/lemmatizer/trees and b/lemmatizer/trees differ
|
|
meta.json
CHANGED
@@ -1,14 +1,14 @@
|
|
1 |
{
|
2 |
"lang":"da",
|
3 |
"name":"core_news_trf",
|
4 |
-
"version":"3.
|
5 |
-
"description":"Danish transformer pipeline (vesteinn/DanskBERT). Components: transformer, morphologizer, parser, lemmatizer (trainable_lemmatizer), 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.
|
11 |
-
"spacy_git_version":"
|
12 |
"vectors":{
|
13 |
"width":0,
|
14 |
"vectors":0,
|
@@ -246,51 +246,51 @@
|
|
246 |
"token_p":0.9977732598,
|
247 |
"token_r":0.9974835463,
|
248 |
"token_f":0.997628382,
|
249 |
-
"pos_acc":0.
|
250 |
-
"morph_acc":0.
|
251 |
-
"morph_micro_p":0.
|
252 |
-
"morph_micro_r":0.
|
253 |
-
"morph_micro_f":0.
|
254 |
"morph_per_feat":{
|
255 |
"Mood":{
|
256 |
-
"p":0.
|
257 |
"r":0.9942802669,
|
258 |
-
"f":0.
|
259 |
},
|
260 |
"Tense":{
|
261 |
-
"p":0.
|
262 |
-
"r":0.
|
263 |
-
"f":0.
|
264 |
},
|
265 |
"VerbForm":{
|
266 |
-
"p":0.
|
267 |
-
"r":0.
|
268 |
-
"f":0.
|
269 |
},
|
270 |
"Voice":{
|
271 |
-
"p":0.
|
272 |
-
"r":0.
|
273 |
-
"f":0.
|
274 |
},
|
275 |
"Definite":{
|
276 |
-
"p":0.
|
277 |
-
"r":0.
|
278 |
-
"f":0.
|
279 |
},
|
280 |
"Gender":{
|
281 |
-
"p":0.
|
282 |
-
"r":0.
|
283 |
-
"f":0.
|
284 |
},
|
285 |
"Number":{
|
286 |
-
"p":0.
|
287 |
-
"r":0.
|
288 |
-
"f":0.
|
289 |
},
|
290 |
"AdpType":{
|
291 |
-
"p":0.
|
292 |
-
"r":0.
|
293 |
-
"f":0.
|
294 |
},
|
295 |
"PartType":{
|
296 |
"p":1.0,
|
@@ -298,29 +298,29 @@
|
|
298 |
"f":1.0
|
299 |
},
|
300 |
"Case":{
|
301 |
-
"p":0.
|
302 |
"r":0.9889415482,
|
303 |
-
"f":0.
|
304 |
},
|
305 |
"Person":{
|
306 |
-
"p":0.
|
307 |
-
"r":0.
|
308 |
-
"f":0.
|
309 |
},
|
310 |
"PronType":{
|
311 |
-
"p":0.
|
312 |
-
"r":0.
|
313 |
-
"f":0.
|
314 |
},
|
315 |
"NumType":{
|
316 |
-
"p":0.
|
317 |
-
"r":0.
|
318 |
-
"f":0.
|
319 |
},
|
320 |
"Degree":{
|
321 |
-
"p":0.
|
322 |
-
"r":0.
|
323 |
-
"f":0.
|
324 |
},
|
325 |
"Reflex":{
|
326 |
"p":1.0,
|
@@ -328,19 +328,19 @@
|
|
328 |
"f":1.0
|
329 |
},
|
330 |
"Number[psor]":{
|
331 |
-
"p":0
|
332 |
-
"r":0
|
333 |
-
"f":0
|
334 |
},
|
335 |
"Poss":{
|
336 |
-
"p":0
|
337 |
-
"r":0
|
338 |
-
"f":0
|
339 |
},
|
340 |
"Foreign":{
|
341 |
-
"p":0
|
342 |
-
"r":0.
|
343 |
-
"f":0.
|
344 |
},
|
345 |
"Abbr":{
|
346 |
"p":1.0,
|
@@ -358,141 +358,146 @@
|
|
358 |
"f":1.0
|
359 |
}
|
360 |
},
|
361 |
-
"sents_p":0.
|
362 |
-
"sents_r":0.
|
363 |
-
"sents_f":0.
|
364 |
-
"dep_uas":0.
|
365 |
-
"dep_las":0.
|
366 |
"dep_las_per_type":{
|
367 |
"advmod":{
|
368 |
-
"p":0.
|
369 |
-
"r":0.
|
370 |
-
"f":0.
|
371 |
},
|
372 |
"root":{
|
373 |
-
"p":0.
|
374 |
-
"r":0.
|
375 |
-
"f":0.
|
376 |
},
|
377 |
"nsubj":{
|
378 |
-
"p":0.
|
379 |
-
"r":0.
|
380 |
-
"f":0.
|
381 |
},
|
382 |
"case":{
|
383 |
-
"p":0.
|
384 |
-
"r":0.
|
385 |
-
"f":0.
|
386 |
},
|
387 |
"obl":{
|
388 |
-
"p":0.
|
389 |
-
"r":0.
|
390 |
-
"f":0.
|
391 |
},
|
392 |
"cc":{
|
393 |
-
"p":0.
|
394 |
-
"r":0.
|
395 |
-
"f":0.
|
396 |
},
|
397 |
"conj":{
|
398 |
-
"p":0.
|
399 |
-
"r":0.
|
400 |
-
"f":0.
|
401 |
},
|
402 |
"obj":{
|
403 |
-
"p":0.
|
404 |
-
"r":0.
|
405 |
-
"f":0.
|
406 |
},
|
407 |
"aux":{
|
408 |
-
"p":0.
|
409 |
-
"r":0.
|
410 |
-
"f":0.
|
411 |
},
|
412 |
"acl:relcl":{
|
413 |
-
"p":0.
|
414 |
-
"r":0.
|
415 |
-
"f":0.
|
416 |
},
|
417 |
"advmod:lmod":{
|
418 |
-
"p":0.
|
419 |
-
"r":0.
|
420 |
-
"f":0.
|
421 |
},
|
422 |
"det":{
|
423 |
-
"p":0.
|
424 |
"r":0.9456342669,
|
425 |
-
"f":0.
|
426 |
},
|
427 |
"amod":{
|
428 |
-
"p":0.
|
429 |
-
"r":0.
|
430 |
-
"f":0.
|
431 |
},
|
432 |
"nmod:poss":{
|
433 |
-
"p":0.
|
434 |
-
"r":0.
|
435 |
-
"f":0.
|
436 |
},
|
437 |
"ccomp":{
|
438 |
-
"p":0.
|
439 |
-
"r":0.
|
440 |
-
"f":0.
|
441 |
},
|
442 |
"nummod":{
|
443 |
-
"p":0.
|
444 |
-
"r":0.
|
445 |
-
"f":0.
|
446 |
},
|
447 |
"flat":{
|
448 |
-
"p":0.
|
449 |
-
"r":0.
|
450 |
-
"f":0.
|
451 |
},
|
452 |
"compound:prt":{
|
453 |
-
"p":0.
|
454 |
-
"r":0.
|
455 |
-
"f":0.
|
456 |
},
|
457 |
"advcl":{
|
458 |
-
"p":0.
|
459 |
-
"r":0.
|
460 |
-
"f":0.
|
461 |
},
|
462 |
"mark":{
|
463 |
-
"p":0.
|
464 |
-
"r":0.
|
465 |
-
"f":0.
|
466 |
},
|
467 |
"cop":{
|
468 |
-
"p":0.
|
469 |
-
"r":0.
|
470 |
-
"f":0.
|
471 |
},
|
472 |
"dep":{
|
473 |
-
"p":0.
|
474 |
-
"r":0.
|
475 |
-
"f":0.
|
476 |
},
|
477 |
"nmod":{
|
478 |
-
"p":0.
|
479 |
-
"r":0.
|
480 |
-
"f":0.
|
481 |
},
|
482 |
"iobj":{
|
483 |
-
"p":0.
|
484 |
-
"r":0.
|
485 |
-
"f":0.
|
486 |
},
|
487 |
"xcomp":{
|
488 |
-
"p":0.
|
489 |
-
"r":0.
|
490 |
-
"f":0.
|
|
|
|
|
|
|
|
|
|
|
491 |
},
|
492 |
"list":{
|
493 |
-
"p":0.
|
494 |
-
"r":0.
|
495 |
-
"f":0.
|
496 |
},
|
497 |
"vocative":{
|
498 |
"p":0.0,
|
@@ -500,64 +505,59 @@
|
|
500 |
"f":0.0
|
501 |
},
|
502 |
"fixed":{
|
503 |
-
"p":0.
|
504 |
-
"r":0.
|
505 |
-
"f":0.
|
506 |
},
|
507 |
"expl":{
|
508 |
-
"p":0.
|
509 |
"r":0.9117647059,
|
510 |
-
"f":0.
|
511 |
},
|
512 |
"appos":{
|
513 |
-
"p":0.
|
514 |
-
"r":0.
|
515 |
-
"f":0.
|
516 |
},
|
517 |
"obl:tmod":{
|
518 |
-
"p":0
|
519 |
-
"r":0.
|
520 |
-
"f":0.
|
521 |
},
|
522 |
"discourse":{
|
523 |
"p":0.0,
|
524 |
"r":0.0,
|
525 |
"f":0.0
|
526 |
-
},
|
527 |
-
"obl:lmod":{
|
528 |
-
"p":0.0,
|
529 |
-
"r":0.0,
|
530 |
-
"f":0.0
|
531 |
}
|
532 |
},
|
533 |
-
"lemma_acc":0.
|
534 |
-
"tag_acc":0.
|
535 |
-
"ents_p":0.
|
536 |
-
"ents_r":0.
|
537 |
-
"ents_f":0.
|
538 |
"ents_per_type":{
|
|
|
|
|
|
|
|
|
|
|
539 |
"PER":{
|
540 |
-
"p":0.
|
541 |
-
"r":0.
|
542 |
-
"f":0.
|
543 |
},
|
544 |
"ORG":{
|
545 |
-
"p":0.
|
546 |
-
"r":0.
|
547 |
-
"f":0.
|
548 |
-
},
|
549 |
-
"MISC":{
|
550 |
-
"p":0.7163120567,
|
551 |
-
"r":0.8938053097,
|
552 |
-
"f":0.7952755906
|
553 |
},
|
554 |
"LOC":{
|
555 |
-
"p":0.
|
556 |
"r":0.9459459459,
|
557 |
-
"f":0.
|
558 |
}
|
559 |
},
|
560 |
-
"speed":
|
561 |
},
|
562 |
"sources":[
|
563 |
{
|
@@ -580,6 +580,6 @@
|
|
580 |
}
|
581 |
],
|
582 |
"requirements":[
|
583 |
-
"spacy-transformers>=
|
584 |
]
|
585 |
}
|
|
|
1 |
{
|
2 |
"lang":"da",
|
3 |
"name":"core_news_trf",
|
4 |
+
"version":"3.7.2",
|
5 |
+
"description":"Danish transformer pipeline (Transformer(name='vesteinn/DanskBERT', piece_encoder='xlm-roberta-sentencepiece', stride=120, type='xlm-roberta', width=768, window=152, vocab_size=50005)). Components: transformer, morphologizer, parser, lemmatizer (trainable_lemmatizer), 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":">=3.7.0,<3.8.0",
|
11 |
+
"spacy_git_version":"6b4f77441",
|
12 |
"vectors":{
|
13 |
"width":0,
|
14 |
"vectors":0,
|
|
|
246 |
"token_p":0.9977732598,
|
247 |
"token_r":0.9974835463,
|
248 |
"token_f":0.997628382,
|
249 |
+
"pos_acc":0.9870683392,
|
250 |
+
"morph_acc":0.9845498135,
|
251 |
+
"morph_micro_p":0.9929531052,
|
252 |
+
"morph_micro_r":0.9886426733,
|
253 |
+
"morph_micro_f":0.9907932011,
|
254 |
"morph_per_feat":{
|
255 |
"Mood":{
|
256 |
+
"p":0.9952290076,
|
257 |
"r":0.9942802669,
|
258 |
+
"f":0.9947544111
|
259 |
},
|
260 |
"Tense":{
|
261 |
+
"p":0.9924242424,
|
262 |
+
"r":0.9864457831,
|
263 |
+
"f":0.9894259819
|
264 |
},
|
265 |
"VerbForm":{
|
266 |
+
"p":0.9919901417,
|
267 |
+
"r":0.9853121175,
|
268 |
+
"f":0.9886398526
|
269 |
},
|
270 |
"Voice":{
|
271 |
+
"p":0.9947643979,
|
272 |
+
"r":0.9940209268,
|
273 |
+
"f":0.9943925234
|
274 |
},
|
275 |
"Definite":{
|
276 |
+
"p":0.9948227798,
|
277 |
+
"r":0.9869616752,
|
278 |
+
"f":0.9908766363
|
279 |
},
|
280 |
"Gender":{
|
281 |
+
"p":0.9886439546,
|
282 |
+
"r":0.9837155201,
|
283 |
+
"f":0.9861735799
|
284 |
},
|
285 |
"Number":{
|
286 |
+
"p":0.9918763103,
|
287 |
+
"r":0.987219614,
|
288 |
+
"f":0.9895424837
|
289 |
},
|
290 |
"AdpType":{
|
291 |
+
"p":0.9991134752,
|
292 |
+
"r":0.9964633068,
|
293 |
+
"f":0.9977866313
|
294 |
},
|
295 |
"PartType":{
|
296 |
"p":1.0,
|
|
|
298 |
"f":1.0
|
299 |
},
|
300 |
"Case":{
|
301 |
+
"p":0.9968152866,
|
302 |
"r":0.9889415482,
|
303 |
+
"f":0.9928628073
|
304 |
},
|
305 |
"Person":{
|
306 |
+
"p":0.9982238011,
|
307 |
+
"r":0.9982238011,
|
308 |
+
"f":0.9982238011
|
309 |
},
|
310 |
"PronType":{
|
311 |
+
"p":0.9975308642,
|
312 |
+
"r":0.9967105263,
|
313 |
+
"f":0.9971205265
|
314 |
},
|
315 |
"NumType":{
|
316 |
+
"p":0.9867549669,
|
317 |
+
"r":0.9867549669,
|
318 |
+
"f":0.9867549669
|
319 |
},
|
320 |
"Degree":{
|
321 |
+
"p":0.9795918367,
|
322 |
+
"r":0.9831325301,
|
323 |
+
"f":0.9813589898
|
324 |
},
|
325 |
"Reflex":{
|
326 |
"p":1.0,
|
|
|
328 |
"f":1.0
|
329 |
},
|
330 |
"Number[psor]":{
|
331 |
+
"p":1.0,
|
332 |
+
"r":1.0,
|
333 |
+
"f":1.0
|
334 |
},
|
335 |
"Poss":{
|
336 |
+
"p":1.0,
|
337 |
+
"r":1.0,
|
338 |
+
"f":1.0
|
339 |
},
|
340 |
"Foreign":{
|
341 |
+
"p":1.0,
|
342 |
+
"r":0.7,
|
343 |
+
"f":0.8235294118
|
344 |
},
|
345 |
"Abbr":{
|
346 |
"p":1.0,
|
|
|
358 |
"f":1.0
|
359 |
}
|
360 |
},
|
361 |
+
"sents_p":0.9261862917,
|
362 |
+
"sents_r":0.9343971631,
|
363 |
+
"sents_f":0.9302736099,
|
364 |
+
"dep_uas":0.897934115,
|
365 |
+
"dep_las":0.8740509156,
|
366 |
"dep_las_per_type":{
|
367 |
"advmod":{
|
368 |
+
"p":0.8345323741,
|
369 |
+
"r":0.8192090395,
|
370 |
+
"f":0.8267997149
|
371 |
},
|
372 |
"root":{
|
373 |
+
"p":0.9014084507,
|
374 |
+
"r":0.9078014184,
|
375 |
+
"f":0.9045936396
|
376 |
},
|
377 |
"nsubj":{
|
378 |
+
"p":0.9345991561,
|
379 |
+
"r":0.9345991561,
|
380 |
+
"f":0.9345991561
|
381 |
},
|
382 |
"case":{
|
383 |
+
"p":0.9358717435,
|
384 |
+
"r":0.9211045365,
|
385 |
+
"f":0.9284294235
|
386 |
},
|
387 |
"obl":{
|
388 |
+
"p":0.8388625592,
|
389 |
+
"r":0.8245341615,
|
390 |
+
"f":0.8316366484
|
391 |
},
|
392 |
"cc":{
|
393 |
+
"p":0.898255814,
|
394 |
+
"r":0.898255814,
|
395 |
+
"f":0.898255814
|
396 |
},
|
397 |
"conj":{
|
398 |
+
"p":0.816,
|
399 |
+
"r":0.816,
|
400 |
+
"f":0.816
|
401 |
},
|
402 |
"obj":{
|
403 |
+
"p":0.9128787879,
|
404 |
+
"r":0.9359223301,
|
405 |
+
"f":0.9242569511
|
406 |
},
|
407 |
"aux":{
|
408 |
+
"p":0.9176470588,
|
409 |
+
"r":0.9096209913,
|
410 |
+
"f":0.9136163982
|
411 |
},
|
412 |
"acl:relcl":{
|
413 |
+
"p":0.7932960894,
|
414 |
+
"r":0.7675675676,
|
415 |
+
"f":0.7802197802
|
416 |
},
|
417 |
"advmod:lmod":{
|
418 |
+
"p":0.8805970149,
|
419 |
+
"r":0.8805970149,
|
420 |
+
"f":0.8805970149
|
421 |
},
|
422 |
"det":{
|
423 |
+
"p":0.9456342669,
|
424 |
"r":0.9456342669,
|
425 |
+
"f":0.9456342669
|
426 |
},
|
427 |
"amod":{
|
428 |
+
"p":0.8939393939,
|
429 |
+
"r":0.9061433447,
|
430 |
+
"f":0.9
|
431 |
},
|
432 |
"nmod:poss":{
|
433 |
+
"p":0.8,
|
434 |
+
"r":0.7920792079,
|
435 |
+
"f":0.7960199005
|
436 |
},
|
437 |
"ccomp":{
|
438 |
+
"p":0.7105263158,
|
439 |
+
"r":0.8709677419,
|
440 |
+
"f":0.7826086957
|
441 |
},
|
442 |
"nummod":{
|
443 |
+
"p":0.8174603175,
|
444 |
+
"r":0.8583333333,
|
445 |
+
"f":0.837398374
|
446 |
},
|
447 |
"flat":{
|
448 |
+
"p":0.8805031447,
|
449 |
+
"r":0.9271523179,
|
450 |
+
"f":0.9032258065
|
451 |
},
|
452 |
"compound:prt":{
|
453 |
+
"p":0.6666666667,
|
454 |
+
"r":0.6341463415,
|
455 |
+
"f":0.65
|
456 |
},
|
457 |
"advcl":{
|
458 |
+
"p":0.7699115044,
|
459 |
+
"r":0.75,
|
460 |
+
"f":0.7598253275
|
461 |
},
|
462 |
"mark":{
|
463 |
+
"p":0.9371069182,
|
464 |
+
"r":0.9178644764,
|
465 |
+
"f":0.9273858921
|
466 |
},
|
467 |
"cop":{
|
468 |
+
"p":0.9152542373,
|
469 |
+
"r":0.9257142857,
|
470 |
+
"f":0.9204545455
|
471 |
},
|
472 |
"dep":{
|
473 |
+
"p":0.2095238095,
|
474 |
+
"r":0.4150943396,
|
475 |
+
"f":0.2784810127
|
476 |
},
|
477 |
"nmod":{
|
478 |
+
"p":0.784989858,
|
479 |
+
"r":0.755859375,
|
480 |
+
"f":0.7701492537
|
481 |
},
|
482 |
"iobj":{
|
483 |
+
"p":0.9375,
|
484 |
+
"r":0.6818181818,
|
485 |
+
"f":0.7894736842
|
486 |
},
|
487 |
"xcomp":{
|
488 |
+
"p":0.7346938776,
|
489 |
+
"r":0.6101694915,
|
490 |
+
"f":0.6666666667
|
491 |
+
},
|
492 |
+
"obl:lmod":{
|
493 |
+
"p":0.0,
|
494 |
+
"r":0.0,
|
495 |
+
"f":0.0
|
496 |
},
|
497 |
"list":{
|
498 |
+
"p":0.4,
|
499 |
+
"r":0.3333333333,
|
500 |
+
"f":0.3636363636
|
501 |
},
|
502 |
"vocative":{
|
503 |
"p":0.0,
|
|
|
505 |
"f":0.0
|
506 |
},
|
507 |
"fixed":{
|
508 |
+
"p":0.9473684211,
|
509 |
+
"r":0.8780487805,
|
510 |
+
"f":0.9113924051
|
511 |
},
|
512 |
"expl":{
|
513 |
+
"p":0.96875,
|
514 |
"r":0.9117647059,
|
515 |
+
"f":0.9393939394
|
516 |
},
|
517 |
"appos":{
|
518 |
+
"p":0.7297297297,
|
519 |
+
"r":0.8181818182,
|
520 |
+
"f":0.7714285714
|
521 |
},
|
522 |
"obl:tmod":{
|
523 |
+
"p":1.0,
|
524 |
+
"r":0.6111111111,
|
525 |
+
"f":0.7586206897
|
526 |
},
|
527 |
"discourse":{
|
528 |
"p":0.0,
|
529 |
"r":0.0,
|
530 |
"f":0.0
|
|
|
|
|
|
|
|
|
|
|
531 |
}
|
532 |
},
|
533 |
+
"lemma_acc":0.9579661017,
|
534 |
+
"tag_acc":0.9870683392,
|
535 |
+
"ents_p":0.8866396761,
|
536 |
+
"ents_r":0.9125,
|
537 |
+
"ents_f":0.8993839836,
|
538 |
"ents_per_type":{
|
539 |
+
"MISC":{
|
540 |
+
"p":0.7846153846,
|
541 |
+
"r":0.9026548673,
|
542 |
+
"f":0.8395061728
|
543 |
+
},
|
544 |
"PER":{
|
545 |
+
"p":0.950310559,
|
546 |
+
"r":0.921686747,
|
547 |
+
"f":0.9357798165
|
548 |
},
|
549 |
"ORG":{
|
550 |
+
"p":0.8764044944,
|
551 |
+
"r":0.8666666667,
|
552 |
+
"f":0.8715083799
|
|
|
|
|
|
|
|
|
|
|
553 |
},
|
554 |
"LOC":{
|
555 |
+
"p":0.9210526316,
|
556 |
"r":0.9459459459,
|
557 |
+
"f":0.9333333333
|
558 |
}
|
559 |
},
|
560 |
+
"speed":523.1005752745
|
561 |
},
|
562 |
"sources":[
|
563 |
{
|
|
|
580 |
}
|
581 |
],
|
582 |
"requirements":[
|
583 |
+
"spacy-curated-transformers>=0.2.0,<0.3.0"
|
584 |
]
|
585 |
}
|
morphologizer/model
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:861742d9415e0a80d100979596a4ad498b775e5a8641a16f53e8b7625833248f
|
3 |
+
size 483664
|
ner/model
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6c0844a418557b03f99de94bde947b4955db2fa867945f90f5b4d4a1a9b05272
|
3 |
+
size 226046
|
parser/model
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3f22af1d9791b6a0f960a1e5ce922b195b012cbb336db44fc40551ad4fd55568
|
3 |
+
size 460409
|
transformer/cfg
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
{
|
2 |
-
|
3 |
}
|
|
|
1 |
{
|
2 |
+
|
3 |
}
|
transformer/model
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:96cb3e3b97b02ef81d8042f34b0c160be89ecb7c236a7659c1faa35646facec0
|
3 |
+
size 496570427
|
vocab/strings.json
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:16ca086c283d7cbff4e96efdb151a1f82d10e82a4df47678979f4be6b20e44e1
|
3 |
+
size 469344
|