osanseviero
commited on
Commit
•
796dea5
1
Parent(s):
4d9c899
Update spaCy pipeline
Browse files- README.md +23 -20
- accuracy.json +195 -192
- attribute_ruler/patterns +0 -0
- config.cfg +22 -20
- en_core_web_trf-any-py3-none-any.whl +2 -2
- meta.json +199 -196
- ner/model +0 -0
- parser/model +0 -0
- tagger/cfg +2 -1
- tagger/model +0 -0
- tokenizer +0 -0
- transformer/{model/pytorch_model.bin → model} +2 -2
- transformer/model/config.json +0 -26
- transformer/model/merges.txt +0 -0
- transformer/model/special_tokens_map.json +0 -1
- transformer/model/tokenizer.json +0 -0
- transformer/model/tokenizer_config.json +0 -1
- transformer/model/vocab.json +0 -0
- vocab/strings.json +2 -2
- vocab/vectors.cfg +3 -0
README.md
CHANGED
@@ -14,47 +14,47 @@ 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: POS
|
26 |
type: token-classification
|
27 |
metrics:
|
28 |
- name: POS Accuracy
|
29 |
type: accuracy
|
30 |
-
value: 0.
|
31 |
- task:
|
32 |
name: SENTER
|
33 |
type: token-classification
|
34 |
metrics:
|
35 |
- name: SENTER Precision
|
36 |
type: precision
|
37 |
-
value: 0.
|
38 |
- name: SENTER Recall
|
39 |
type: recall
|
40 |
-
value: 0.
|
41 |
- name: SENTER F Score
|
42 |
type: f_score
|
43 |
-
value: 0.
|
44 |
- task:
|
45 |
name: UNLABELED_DEPENDENCIES
|
46 |
type: token-classification
|
47 |
metrics:
|
48 |
- name: Unlabeled Dependencies Accuracy
|
49 |
type: accuracy
|
50 |
-
value: 0.
|
51 |
- task:
|
52 |
name: LABELED_DEPENDENCIES
|
53 |
type: token-classification
|
54 |
metrics:
|
55 |
- name: Labeled Dependencies Accuracy
|
56 |
type: accuracy
|
57 |
-
value: 0.
|
58 |
---
|
59 |
### Details: https://spacy.io/models/en#en_core_web_trf
|
60 |
|
@@ -63,8 +63,8 @@ English transformer pipeline (roberta-base). Components: transformer, tagger, pa
|
|
63 |
| Feature | Description |
|
64 |
| --- | --- |
|
65 |
| **Name** | `en_core_web_trf` |
|
66 |
-
| **Version** | `3.
|
67 |
-
| **spaCy** | `>=3.
|
68 |
| **Default Pipeline** | `transformer`, `tagger`, `parser`, `attribute_ruler`, `lemmatizer`, `ner` |
|
69 |
| **Components** | `transformer`, `tagger`, `parser`, `attribute_ruler`, `lemmatizer`, `ner` |
|
70 |
| **Vectors** | 0 keys, 0 unique vectors (0 dimensions) |
|
@@ -91,12 +91,15 @@ English transformer pipeline (roberta-base). Components: transformer, tagger, pa
|
|
91 |
| Type | Score |
|
92 |
| --- | --- |
|
93 |
| `TOKEN_ACC` | 99.93 |
|
94 |
-
| `
|
95 |
-
| `
|
96 |
-
| `
|
97 |
-
| `
|
98 |
-
| `
|
99 |
-
| `
|
100 |
-
| `
|
101 |
-
| `
|
102 |
-
| `
|
|
|
|
|
|
|
|
14 |
metrics:
|
15 |
- name: NER Precision
|
16 |
type: precision
|
17 |
+
value: 0.898708838
|
18 |
- name: NER Recall
|
19 |
type: recall
|
20 |
+
value: 0.8993389423
|
21 |
- name: NER F Score
|
22 |
type: f_score
|
23 |
+
value: 0.8990237797
|
24 |
- task:
|
25 |
name: POS
|
26 |
type: token-classification
|
27 |
metrics:
|
28 |
- name: POS Accuracy
|
29 |
type: accuracy
|
30 |
+
value: 0.9780627094
|
31 |
- task:
|
32 |
name: SENTER
|
33 |
type: token-classification
|
34 |
metrics:
|
35 |
- name: SENTER Precision
|
36 |
type: precision
|
37 |
+
value: 0.9568827385
|
38 |
- name: SENTER Recall
|
39 |
type: recall
|
40 |
+
value: 0.8666798601
|
41 |
- name: SENTER F Score
|
42 |
type: f_score
|
43 |
+
value: 0.9095503479
|
44 |
- task:
|
45 |
name: UNLABELED_DEPENDENCIES
|
46 |
type: token-classification
|
47 |
metrics:
|
48 |
- name: Unlabeled Dependencies Accuracy
|
49 |
type: accuracy
|
50 |
+
value: 0.9527056513
|
51 |
- task:
|
52 |
name: LABELED_DEPENDENCIES
|
53 |
type: token-classification
|
54 |
metrics:
|
55 |
- name: Labeled Dependencies Accuracy
|
56 |
type: accuracy
|
57 |
+
value: 0.9527056513
|
58 |
---
|
59 |
### Details: https://spacy.io/models/en#en_core_web_trf
|
60 |
|
|
|
63 |
| Feature | Description |
|
64 |
| --- | --- |
|
65 |
| **Name** | `en_core_web_trf` |
|
66 |
+
| **Version** | `3.2.0` |
|
67 |
+
| **spaCy** | `>=3.2.0,<3.3.0` |
|
68 |
| **Default Pipeline** | `transformer`, `tagger`, `parser`, `attribute_ruler`, `lemmatizer`, `ner` |
|
69 |
| **Components** | `transformer`, `tagger`, `parser`, `attribute_ruler`, `lemmatizer`, `ner` |
|
70 |
| **Vectors** | 0 keys, 0 unique vectors (0 dimensions) |
|
|
|
91 |
| Type | Score |
|
92 |
| --- | --- |
|
93 |
| `TOKEN_ACC` | 99.93 |
|
94 |
+
| `TOKEN_P` | 99.57 |
|
95 |
+
| `TOKEN_R` | 99.58 |
|
96 |
+
| `TOKEN_F` | 99.57 |
|
97 |
+
| `TAG_ACC` | 97.81 |
|
98 |
+
| `SENTS_P` | 95.69 |
|
99 |
+
| `SENTS_R` | 86.67 |
|
100 |
+
| `SENTS_F` | 90.96 |
|
101 |
+
| `DEP_UAS` | 95.27 |
|
102 |
+
| `DEP_LAS` | 93.93 |
|
103 |
+
| `ENTS_P` | 89.87 |
|
104 |
+
| `ENTS_R` | 89.93 |
|
105 |
+
| `ENTS_F` | 89.90 |
|
accuracy.json
CHANGED
@@ -1,327 +1,330 @@
|
|
1 |
{
|
2 |
"token_acc": 0.9993053983,
|
3 |
-
"
|
4 |
-
"
|
5 |
-
"
|
6 |
-
"
|
7 |
-
"
|
8 |
-
"
|
9 |
-
"
|
10 |
-
"
|
11 |
-
"
|
12 |
-
"speed": 5467.6411211076,
|
13 |
"dep_las_per_type": {
|
14 |
"prep": {
|
15 |
-
"p": 0.
|
16 |
-
"r": 0.
|
17 |
-
"f": 0.
|
18 |
},
|
19 |
"det": {
|
20 |
-
"p": 0.
|
21 |
-
"r": 0.
|
22 |
-
"f": 0.
|
23 |
},
|
24 |
"pobj": {
|
25 |
-
"p": 0.
|
26 |
-
"r": 0.
|
27 |
-
"f": 0.
|
28 |
},
|
29 |
"nsubj": {
|
30 |
-
"p": 0.
|
31 |
-
"r": 0.
|
32 |
-
"f": 0.
|
33 |
},
|
34 |
"aux": {
|
35 |
-
"p": 0.
|
36 |
-
"r": 0.
|
37 |
-
"f": 0.
|
38 |
},
|
39 |
"advmod": {
|
40 |
-
"p": 0.
|
41 |
-
"r": 0.
|
42 |
-
"f": 0.
|
43 |
},
|
44 |
"relcl": {
|
45 |
-
"p": 0.
|
46 |
-
"r": 0.
|
47 |
-
"f": 0.
|
48 |
},
|
49 |
"root": {
|
50 |
-
"p": 0.
|
51 |
-
"r": 0.
|
52 |
-
"f": 0.
|
53 |
},
|
54 |
"xcomp": {
|
55 |
-
"p": 0.
|
56 |
-
"r": 0.
|
57 |
-
"f": 0.
|
58 |
},
|
59 |
"amod": {
|
60 |
-
"p": 0.
|
61 |
-
"r": 0.
|
62 |
-
"f": 0.
|
63 |
},
|
64 |
"compound": {
|
65 |
-
"p": 0.
|
66 |
-
"r": 0.
|
67 |
-
"f": 0.
|
68 |
},
|
69 |
"poss": {
|
70 |
-
"p": 0.
|
71 |
-
"r": 0.
|
72 |
-
"f": 0.
|
73 |
},
|
74 |
"ccomp": {
|
75 |
-
"p": 0.
|
76 |
-
"r": 0.
|
77 |
-
"f": 0.
|
78 |
},
|
79 |
"attr": {
|
80 |
-
"p": 0.
|
81 |
-
"r": 0.
|
82 |
-
"f": 0.
|
83 |
},
|
84 |
"case": {
|
85 |
-
"p": 0.
|
86 |
-
"r": 0.
|
87 |
-
"f": 0.
|
88 |
},
|
89 |
"mark": {
|
90 |
-
"p": 0.
|
91 |
-
"r": 0.
|
92 |
-
"f": 0.
|
93 |
},
|
94 |
"intj": {
|
95 |
-
"p": 0.
|
96 |
-
"r": 0.
|
97 |
-
"f": 0.
|
98 |
},
|
99 |
"advcl": {
|
100 |
-
"p": 0.
|
101 |
-
"r": 0.
|
102 |
-
"f": 0.
|
103 |
},
|
104 |
"cc": {
|
105 |
-
"p": 0.
|
106 |
-
"r": 0.
|
107 |
-
"f": 0.
|
108 |
},
|
109 |
"neg": {
|
110 |
-
"p": 0.
|
111 |
-
"r": 0.
|
112 |
-
"f": 0.
|
113 |
},
|
114 |
"conj": {
|
115 |
-
"p": 0.
|
116 |
-
"r": 0.
|
117 |
-
"f": 0.
|
118 |
},
|
119 |
"nsubjpass": {
|
120 |
-
"p": 0.
|
121 |
-
"r": 0.
|
122 |
-
"f": 0.
|
123 |
},
|
124 |
"auxpass": {
|
125 |
-
"p": 0.
|
126 |
-
"r": 0.
|
127 |
-
"f": 0.
|
128 |
},
|
129 |
"dobj": {
|
130 |
-
"p": 0.
|
131 |
-
"r": 0.
|
132 |
-
"f": 0.
|
133 |
},
|
134 |
"nummod": {
|
135 |
-
"p": 0.
|
136 |
-
"r": 0.
|
137 |
-
"f": 0.
|
138 |
},
|
139 |
"npadvmod": {
|
140 |
-
"p": 0.
|
141 |
-
"r": 0.
|
142 |
-
"f": 0.
|
143 |
},
|
144 |
"prt": {
|
145 |
-
"p": 0.
|
146 |
-
"r": 0.
|
147 |
-
"f": 0.
|
148 |
},
|
149 |
"pcomp": {
|
150 |
-
"p": 0.
|
151 |
-
"r": 0.
|
152 |
-
"f": 0.
|
153 |
},
|
154 |
"expl": {
|
155 |
-
"p": 0.
|
156 |
"r": 0.9957173448,
|
157 |
-
"f": 0.
|
158 |
},
|
159 |
"acl": {
|
160 |
-
"p": 0.
|
161 |
-
"r": 0.
|
162 |
-
"f": 0.
|
163 |
},
|
164 |
"agent": {
|
165 |
-
"p": 0.
|
166 |
-
"r": 0.
|
167 |
-
"f": 0.
|
168 |
},
|
169 |
"dative": {
|
170 |
-
"p": 0.
|
171 |
-
"r": 0.
|
172 |
-
"f": 0.
|
173 |
},
|
174 |
"acomp": {
|
175 |
-
"p": 0.
|
176 |
-
"r": 0.
|
177 |
-
"f": 0.
|
178 |
},
|
179 |
"dep": {
|
180 |
-
"p": 0.
|
181 |
-
"r": 0.
|
182 |
-
"f": 0.
|
183 |
},
|
184 |
"csubj": {
|
185 |
-
"p": 0.
|
186 |
-
"r": 0.
|
187 |
-
"f": 0.
|
188 |
},
|
189 |
"quantmod": {
|
190 |
-
"p": 0.
|
191 |
-
"r": 0.
|
192 |
-
"f": 0.
|
193 |
},
|
194 |
"nmod": {
|
195 |
-
"p": 0.
|
196 |
-
"r": 0.
|
197 |
-
"f": 0.
|
198 |
},
|
199 |
"appos": {
|
200 |
-
"p": 0.
|
201 |
-
"r": 0.
|
202 |
-
"f": 0.
|
203 |
},
|
204 |
"predet": {
|
205 |
-
"p": 0.
|
206 |
-
"r": 0.
|
207 |
-
"f": 0.
|
208 |
},
|
209 |
"preconj": {
|
210 |
-
"p": 0.
|
211 |
-
"r": 0.
|
212 |
-
"f": 0.
|
213 |
},
|
214 |
"oprd": {
|
215 |
-
"p": 0.
|
216 |
-
"r": 0.
|
217 |
-
"f": 0.
|
218 |
},
|
219 |
"parataxis": {
|
220 |
-
"p": 0.
|
221 |
-
"r": 0.
|
222 |
-
"f": 0.
|
223 |
},
|
224 |
"meta": {
|
225 |
-
"p": 0.
|
226 |
"r": 0.6538461538,
|
227 |
-
"f": 0.
|
228 |
},
|
229 |
"csubjpass": {
|
230 |
-
"p":
|
231 |
"r": 0.8333333333,
|
232 |
-
"f": 0.
|
233 |
}
|
234 |
},
|
|
|
|
|
|
|
235 |
"ents_per_type": {
|
236 |
"DATE": {
|
237 |
-
"p": 0.
|
238 |
-
"r": 0.
|
239 |
-
"f": 0.
|
240 |
},
|
241 |
"GPE": {
|
242 |
-
"p": 0.
|
243 |
-
"r": 0.
|
244 |
-
"f": 0.
|
245 |
},
|
246 |
"ORDINAL": {
|
247 |
-
"p": 0.
|
248 |
"r": 0.8229813665,
|
249 |
-
"f": 0.
|
250 |
},
|
251 |
"ORG": {
|
252 |
-
"p": 0.
|
253 |
-
"r": 0.
|
254 |
-
"f": 0.
|
255 |
},
|
256 |
"FAC": {
|
257 |
-
"p": 0.
|
258 |
-
"r": 0.
|
259 |
-
"f": 0.
|
260 |
},
|
261 |
"QUANTITY": {
|
262 |
-
"p": 0.
|
263 |
-
"r": 0.
|
264 |
-
"f": 0.
|
265 |
},
|
266 |
"LOC": {
|
267 |
-
"p": 0.
|
268 |
-
"r": 0.
|
269 |
-
"f": 0.
|
270 |
},
|
271 |
"CARDINAL": {
|
272 |
-
"p": 0.
|
273 |
-
"r": 0.
|
274 |
-
"f": 0.
|
275 |
},
|
276 |
"PERSON": {
|
277 |
-
"p": 0.
|
278 |
-
"r": 0.
|
279 |
-
"f": 0.
|
280 |
},
|
281 |
"NORP": {
|
282 |
-
"p": 0.
|
283 |
-
"r": 0.
|
284 |
-
"f": 0.
|
285 |
},
|
286 |
"LAW": {
|
287 |
-
"p": 0.
|
288 |
-
"r": 0.
|
289 |
-
"f": 0.
|
290 |
},
|
291 |
"PRODUCT": {
|
292 |
-
"p": 0.
|
293 |
"r": 0.5260663507,
|
294 |
-
"f": 0.
|
|
|
|
|
|
|
|
|
|
|
295 |
},
|
296 |
"TIME": {
|
297 |
-
"p": 0.
|
298 |
"r": 0.7777777778,
|
299 |
-
"f": 0.
|
300 |
},
|
301 |
"EVENT": {
|
302 |
-
"p": 0.
|
303 |
-
"r": 0.
|
304 |
-
"f": 0.
|
305 |
},
|
306 |
"WORK_OF_ART": {
|
307 |
-
"p": 0.
|
308 |
-
"r": 0.
|
309 |
-
"f": 0.
|
310 |
-
},
|
311 |
-
"MONEY": {
|
312 |
-
"p": 0.9120481928,
|
313 |
-
"r": 0.893742621,
|
314 |
-
"f": 0.9028026237
|
315 |
},
|
316 |
"PERCENT": {
|
317 |
-
"p": 0.
|
318 |
-
"r": 0.
|
319 |
-
"f": 0.
|
320 |
},
|
321 |
"LANGUAGE": {
|
322 |
"p": 1.0,
|
323 |
-
"r": 0.
|
324 |
-
"f": 0.
|
325 |
}
|
326 |
-
}
|
|
|
327 |
}
|
|
|
1 |
{
|
2 |
"token_acc": 0.9993053983,
|
3 |
+
"token_p": 0.9956742163,
|
4 |
+
"token_r": 0.9957505887,
|
5 |
+
"token_f": 0.9957124011,
|
6 |
+
"tag_acc": 0.9780627094,
|
7 |
+
"sents_p": 0.9568827385,
|
8 |
+
"sents_r": 0.8666798601,
|
9 |
+
"sents_f": 0.9095503479,
|
10 |
+
"dep_uas": 0.9527056513,
|
11 |
+
"dep_las": 0.9393298266,
|
|
|
12 |
"dep_las_per_type": {
|
13 |
"prep": {
|
14 |
+
"p": 0.9229604964,
|
15 |
+
"r": 0.9255169797,
|
16 |
+
"f": 0.9242369702
|
17 |
},
|
18 |
"det": {
|
19 |
+
"p": 0.9898046572,
|
20 |
+
"r": 0.9896028704,
|
21 |
+
"f": 0.9897037535
|
22 |
},
|
23 |
"pobj": {
|
24 |
+
"p": 0.9832869081,
|
25 |
+
"r": 0.9842136265,
|
26 |
+
"f": 0.9837500491
|
27 |
},
|
28 |
"nsubj": {
|
29 |
+
"p": 0.9803139249,
|
30 |
+
"r": 0.9795837897,
|
31 |
+
"f": 0.9799487213
|
32 |
},
|
33 |
"aux": {
|
34 |
+
"p": 0.989571263,
|
35 |
+
"r": 0.9883379329,
|
36 |
+
"f": 0.9889542134
|
37 |
},
|
38 |
"advmod": {
|
39 |
+
"p": 0.8962161255,
|
40 |
+
"r": 0.8987043581,
|
41 |
+
"f": 0.8974585171
|
42 |
},
|
43 |
"relcl": {
|
44 |
+
"p": 0.8798724309,
|
45 |
+
"r": 0.9009433962,
|
46 |
+
"f": 0.8902832556
|
47 |
},
|
48 |
"root": {
|
49 |
+
"p": 0.9664043142,
|
50 |
+
"r": 0.8747938518,
|
51 |
+
"f": 0.9183200028
|
52 |
},
|
53 |
"xcomp": {
|
54 |
+
"p": 0.9438444924,
|
55 |
+
"r": 0.9411342426,
|
56 |
+
"f": 0.9424874191
|
57 |
},
|
58 |
"amod": {
|
59 |
+
"p": 0.9413514387,
|
60 |
+
"r": 0.943181082,
|
61 |
+
"f": 0.9422653722
|
62 |
},
|
63 |
"compound": {
|
64 |
+
"p": 0.951964576,
|
65 |
+
"r": 0.9458676765,
|
66 |
+
"f": 0.9489063329
|
67 |
},
|
68 |
"poss": {
|
69 |
+
"p": 0.986728333,
|
70 |
+
"r": 0.9877214171,
|
71 |
+
"f": 0.9872246253
|
72 |
},
|
73 |
"ccomp": {
|
74 |
+
"p": 0.8604651163,
|
75 |
+
"r": 0.9118126273,
|
76 |
+
"f": 0.8853950361
|
77 |
},
|
78 |
"attr": {
|
79 |
+
"p": 0.9478225144,
|
80 |
+
"r": 0.9701429773,
|
81 |
+
"f": 0.9588528678
|
82 |
},
|
83 |
"case": {
|
84 |
+
"p": 0.9890547264,
|
85 |
+
"r": 0.994994995,
|
86 |
+
"f": 0.9920159681
|
87 |
},
|
88 |
"mark": {
|
89 |
+
"p": 0.9492105263,
|
90 |
+
"r": 0.9557498675,
|
91 |
+
"f": 0.9524689728
|
92 |
},
|
93 |
"intj": {
|
94 |
+
"p": 0.6237388724,
|
95 |
+
"r": 0.76996337,
|
96 |
+
"f": 0.6891803279
|
97 |
},
|
98 |
"advcl": {
|
99 |
+
"p": 0.8195836546,
|
100 |
+
"r": 0.803072274,
|
101 |
+
"f": 0.8112439583
|
102 |
},
|
103 |
"cc": {
|
104 |
+
"p": 0.8986197049,
|
105 |
+
"r": 0.9032412391,
|
106 |
+
"f": 0.9009245452
|
107 |
},
|
108 |
"neg": {
|
109 |
+
"p": 0.9574148297,
|
110 |
+
"r": 0.958855996,
|
111 |
+
"f": 0.9581348709
|
112 |
},
|
113 |
"conj": {
|
114 |
+
"p": 0.8712776177,
|
115 |
+
"r": 0.9133937563,
|
116 |
+
"f": 0.8918387414
|
117 |
},
|
118 |
"nsubjpass": {
|
119 |
+
"p": 0.9616182573,
|
120 |
+
"r": 0.9507692308,
|
121 |
+
"f": 0.9561629706
|
122 |
},
|
123 |
"auxpass": {
|
124 |
+
"p": 0.9676979072,
|
125 |
+
"r": 0.9690205011,
|
126 |
+
"f": 0.9683587526
|
127 |
},
|
128 |
"dobj": {
|
129 |
+
"p": 0.9742177522,
|
130 |
+
"r": 0.9725874572,
|
131 |
+
"f": 0.9734019221
|
132 |
},
|
133 |
"nummod": {
|
134 |
+
"p": 0.9582800102,
|
135 |
+
"r": 0.9454545455,
|
136 |
+
"f": 0.9518240753
|
137 |
},
|
138 |
"npadvmod": {
|
139 |
+
"p": 0.8582474227,
|
140 |
+
"r": 0.8280639432,
|
141 |
+
"f": 0.8428855541
|
142 |
},
|
143 |
"prt": {
|
144 |
+
"p": 0.896,
|
145 |
+
"r": 0.9032258065,
|
146 |
+
"f": 0.8995983936
|
147 |
},
|
148 |
"pcomp": {
|
149 |
+
"p": 0.9316901408,
|
150 |
+
"r": 0.9264705882,
|
151 |
+
"f": 0.9290730337
|
152 |
},
|
153 |
"expl": {
|
154 |
+
"p": 0.9893617021,
|
155 |
"r": 0.9957173448,
|
156 |
+
"f": 0.992529349
|
157 |
},
|
158 |
"acl": {
|
159 |
+
"p": 0.8476138829,
|
160 |
+
"r": 0.852700491,
|
161 |
+
"f": 0.8501495785
|
162 |
},
|
163 |
"agent": {
|
164 |
+
"p": 0.9559859155,
|
165 |
+
"r": 0.9731182796,
|
166 |
+
"f": 0.9644760213
|
167 |
},
|
168 |
"dative": {
|
169 |
+
"p": 0.8236607143,
|
170 |
+
"r": 0.8463302752,
|
171 |
+
"f": 0.834841629
|
172 |
},
|
173 |
"acomp": {
|
174 |
+
"p": 0.9352355072,
|
175 |
+
"r": 0.9365079365,
|
176 |
+
"f": 0.9358712894
|
177 |
},
|
178 |
"dep": {
|
179 |
+
"p": 0.3949044586,
|
180 |
+
"r": 0.4025974026,
|
181 |
+
"f": 0.3987138264
|
182 |
},
|
183 |
"csubj": {
|
184 |
+
"p": 0.8928571429,
|
185 |
+
"r": 0.8875739645,
|
186 |
+
"f": 0.8902077151
|
187 |
},
|
188 |
"quantmod": {
|
189 |
+
"p": 0.8867449664,
|
190 |
+
"r": 0.8586515028,
|
191 |
+
"f": 0.872472142
|
192 |
},
|
193 |
"nmod": {
|
194 |
+
"p": 0.82904795,
|
195 |
+
"r": 0.7269957343,
|
196 |
+
"f": 0.7746753247
|
197 |
},
|
198 |
"appos": {
|
199 |
+
"p": 0.8099701747,
|
200 |
+
"r": 0.8247288503,
|
201 |
+
"f": 0.8172828891
|
202 |
},
|
203 |
"predet": {
|
204 |
+
"p": 0.8508064516,
|
205 |
+
"r": 0.9055793991,
|
206 |
+
"f": 0.8773388773
|
207 |
},
|
208 |
"preconj": {
|
209 |
+
"p": 0.7083333333,
|
210 |
+
"r": 0.7906976744,
|
211 |
+
"f": 0.7472527473
|
212 |
},
|
213 |
"oprd": {
|
214 |
+
"p": 0.9022082019,
|
215 |
+
"r": 0.8537313433,
|
216 |
+
"f": 0.8773006135
|
217 |
},
|
218 |
"parataxis": {
|
219 |
+
"p": 0.5602189781,
|
220 |
+
"r": 0.6659436009,
|
221 |
+
"f": 0.6085232904
|
222 |
},
|
223 |
"meta": {
|
224 |
+
"p": 0.1683168317,
|
225 |
"r": 0.6538461538,
|
226 |
+
"f": 0.2677165354
|
227 |
},
|
228 |
"csubjpass": {
|
229 |
+
"p": 0.8333333333,
|
230 |
"r": 0.8333333333,
|
231 |
+
"f": 0.8333333333
|
232 |
}
|
233 |
},
|
234 |
+
"ents_p": 0.898708838,
|
235 |
+
"ents_r": 0.8993389423,
|
236 |
+
"ents_f": 0.8990237797,
|
237 |
"ents_per_type": {
|
238 |
"DATE": {
|
239 |
+
"p": 0.8928683632,
|
240 |
+
"r": 0.9022222222,
|
241 |
+
"f": 0.8975209222
|
242 |
},
|
243 |
"GPE": {
|
244 |
+
"p": 0.9575551783,
|
245 |
+
"r": 0.9439330544,
|
246 |
+
"f": 0.9506953224
|
247 |
},
|
248 |
"ORDINAL": {
|
249 |
+
"p": 0.8079268293,
|
250 |
"r": 0.8229813665,
|
251 |
+
"f": 0.8153846154
|
252 |
},
|
253 |
"ORG": {
|
254 |
+
"p": 0.9032515991,
|
255 |
+
"r": 0.8984623542,
|
256 |
+
"f": 0.9008506114
|
257 |
},
|
258 |
"FAC": {
|
259 |
+
"p": 0.5567010309,
|
260 |
+
"r": 0.8307692308,
|
261 |
+
"f": 0.6666666667
|
262 |
},
|
263 |
"QUANTITY": {
|
264 |
+
"p": 0.8187134503,
|
265 |
+
"r": 0.7692307692,
|
266 |
+
"f": 0.7932011331
|
267 |
},
|
268 |
"LOC": {
|
269 |
+
"p": 0.8610169492,
|
270 |
+
"r": 0.8089171975,
|
271 |
+
"f": 0.8341543514
|
272 |
},
|
273 |
"CARDINAL": {
|
274 |
+
"p": 0.8530092593,
|
275 |
+
"r": 0.8763376932,
|
276 |
+
"f": 0.864516129
|
277 |
},
|
278 |
"PERSON": {
|
279 |
+
"p": 0.9436258808,
|
280 |
+
"r": 0.9614882507,
|
281 |
+
"f": 0.9524733269
|
282 |
},
|
283 |
"NORP": {
|
284 |
+
"p": 0.9211783439,
|
285 |
+
"r": 0.9256,
|
286 |
+
"f": 0.9233838787
|
287 |
},
|
288 |
"LAW": {
|
289 |
+
"p": 0.5324675325,
|
290 |
+
"r": 0.640625,
|
291 |
+
"f": 0.5815602837
|
292 |
},
|
293 |
"PRODUCT": {
|
294 |
+
"p": 0.649122807,
|
295 |
"r": 0.5260663507,
|
296 |
+
"f": 0.5811518325
|
297 |
+
},
|
298 |
+
"MONEY": {
|
299 |
+
"p": 0.9157641396,
|
300 |
+
"r": 0.8984651712,
|
301 |
+
"f": 0.9070321812
|
302 |
},
|
303 |
"TIME": {
|
304 |
+
"p": 0.7514124294,
|
305 |
"r": 0.7777777778,
|
306 |
+
"f": 0.7643678161
|
307 |
},
|
308 |
"EVENT": {
|
309 |
+
"p": 0.776119403,
|
310 |
+
"r": 0.5977011494,
|
311 |
+
"f": 0.6753246753
|
312 |
},
|
313 |
"WORK_OF_ART": {
|
314 |
+
"p": 0.5684210526,
|
315 |
+
"r": 0.5567010309,
|
316 |
+
"f": 0.5625
|
|
|
|
|
|
|
|
|
|
|
317 |
},
|
318 |
"PERCENT": {
|
319 |
+
"p": 0.9154929577,
|
320 |
+
"r": 0.8958652374,
|
321 |
+
"f": 0.9055727554
|
322 |
},
|
323 |
"LANGUAGE": {
|
324 |
"p": 1.0,
|
325 |
+
"r": 0.71875,
|
326 |
+
"f": 0.8363636364
|
327 |
}
|
328 |
+
},
|
329 |
+
"speed": 6163.7320781314
|
330 |
}
|
attribute_ruler/patterns
CHANGED
Binary files a/attribute_ruler/patterns and b/attribute_ruler/patterns differ
|
|
config.cfg
CHANGED
@@ -1,10 +1,8 @@
|
|
1 |
[paths]
|
2 |
-
train =
|
3 |
-
dev =
|
4 |
vectors = null
|
5 |
-
raw = null
|
6 |
init_tok2vec = null
|
7 |
-
vocab_data = null
|
8 |
|
9 |
[system]
|
10 |
gpu_allocator = "pytorch"
|
@@ -24,6 +22,7 @@ tokenizer = {"@tokenizers":"spacy.Tokenizer.v1"}
|
|
24 |
|
25 |
[components.attribute_ruler]
|
26 |
factory = "attribute_ruler"
|
|
|
27 |
validate = false
|
28 |
|
29 |
[components.lemmatizer]
|
@@ -31,11 +30,13 @@ factory = "lemmatizer"
|
|
31 |
mode = "rule"
|
32 |
model = null
|
33 |
overwrite = false
|
|
|
34 |
|
35 |
[components.ner]
|
36 |
factory = "ner"
|
37 |
incorrect_spans_key = null
|
38 |
moves = null
|
|
|
39 |
update_with_oracle_cut_size = 100
|
40 |
|
41 |
[components.ner.model]
|
@@ -58,6 +59,7 @@ factory = "parser"
|
|
58 |
learn_tokens = false
|
59 |
min_action_freq = 30
|
60 |
moves = null
|
|
|
61 |
update_with_oracle_cut_size = 100
|
62 |
|
63 |
[components.parser.model]
|
@@ -77,6 +79,8 @@ pooling = {"@layers":"reduce_mean.v1"}
|
|
77 |
|
78 |
[components.tagger]
|
79 |
factory = "tagger"
|
|
|
|
|
80 |
|
81 |
[components.tagger.model]
|
82 |
@architectures = "spacy.Tagger.v1"
|
@@ -94,42 +98,39 @@ max_batch_items = 4096
|
|
94 |
set_extra_annotations = {"@annotation_setters":"spacy-transformers.null_annotation_setter.v1"}
|
95 |
|
96 |
[components.transformer.model]
|
97 |
-
@architectures = "spacy-transformers.TransformerModel.
|
98 |
name = "roberta-base"
|
|
|
99 |
|
100 |
[components.transformer.model.get_spans]
|
101 |
@span_getters = "spacy-transformers.strided_spans.v1"
|
102 |
window = 128
|
103 |
stride = 96
|
104 |
|
|
|
|
|
105 |
[components.transformer.model.tokenizer_config]
|
106 |
use_fast = true
|
107 |
|
|
|
|
|
108 |
[corpora]
|
109 |
|
110 |
[corpora.dev]
|
111 |
@readers = "spacy.Corpus.v1"
|
112 |
-
|
113 |
-
max_length = 0
|
114 |
-
path = ${paths:dev}
|
115 |
gold_preproc = false
|
|
|
|
|
116 |
augmenter = null
|
117 |
|
118 |
[corpora.train]
|
119 |
@readers = "spacy.Corpus.v1"
|
120 |
-
path = ${paths
|
121 |
-
max_length = 500
|
122 |
gold_preproc = false
|
|
|
123 |
limit = 0
|
124 |
-
|
125 |
-
[corpora.train.augmenter]
|
126 |
-
@augmenters = "spacy.orth_variants.v1"
|
127 |
-
level = 0.2
|
128 |
-
lower = 0.5
|
129 |
-
|
130 |
-
[corpora.train.augmenter.orth_variants]
|
131 |
-
@readers = "srsly.read_json.v1"
|
132 |
-
path = "assets/orth_variants.json"
|
133 |
|
134 |
[training]
|
135 |
train_corpus = "corpora.train"
|
@@ -186,11 +187,12 @@ ents_f = 0.16
|
|
186 |
ents_p = 0.0
|
187 |
ents_r = 0.0
|
188 |
ents_per_type = null
|
|
|
189 |
|
190 |
[pretraining]
|
191 |
|
192 |
[initialize]
|
193 |
-
vocab_data =
|
194 |
vectors = ${paths.vectors}
|
195 |
init_tok2vec = ${paths.init_tok2vec}
|
196 |
before_init = null
|
|
|
1 |
[paths]
|
2 |
+
train = null
|
3 |
+
dev = null
|
4 |
vectors = null
|
|
|
5 |
init_tok2vec = null
|
|
|
6 |
|
7 |
[system]
|
8 |
gpu_allocator = "pytorch"
|
|
|
22 |
|
23 |
[components.attribute_ruler]
|
24 |
factory = "attribute_ruler"
|
25 |
+
scorer = {"@scorers":"spacy.attribute_ruler_scorer.v1"}
|
26 |
validate = false
|
27 |
|
28 |
[components.lemmatizer]
|
|
|
30 |
mode = "rule"
|
31 |
model = null
|
32 |
overwrite = false
|
33 |
+
scorer = {"@scorers":"spacy.lemmatizer_scorer.v1"}
|
34 |
|
35 |
[components.ner]
|
36 |
factory = "ner"
|
37 |
incorrect_spans_key = null
|
38 |
moves = null
|
39 |
+
scorer = {"@scorers":"spacy.ner_scorer.v1"}
|
40 |
update_with_oracle_cut_size = 100
|
41 |
|
42 |
[components.ner.model]
|
|
|
59 |
learn_tokens = false
|
60 |
min_action_freq = 30
|
61 |
moves = null
|
62 |
+
scorer = {"@scorers":"spacy.parser_scorer.v1"}
|
63 |
update_with_oracle_cut_size = 100
|
64 |
|
65 |
[components.parser.model]
|
|
|
79 |
|
80 |
[components.tagger]
|
81 |
factory = "tagger"
|
82 |
+
overwrite = false
|
83 |
+
scorer = {"@scorers":"spacy.tagger_scorer.v1"}
|
84 |
|
85 |
[components.tagger.model]
|
86 |
@architectures = "spacy.Tagger.v1"
|
|
|
98 |
set_extra_annotations = {"@annotation_setters":"spacy-transformers.null_annotation_setter.v1"}
|
99 |
|
100 |
[components.transformer.model]
|
101 |
+
@architectures = "spacy-transformers.TransformerModel.v3"
|
102 |
name = "roberta-base"
|
103 |
+
mixed_precision = false
|
104 |
|
105 |
[components.transformer.model.get_spans]
|
106 |
@span_getters = "spacy-transformers.strided_spans.v1"
|
107 |
window = 128
|
108 |
stride = 96
|
109 |
|
110 |
+
[components.transformer.model.grad_scaler_config]
|
111 |
+
|
112 |
[components.transformer.model.tokenizer_config]
|
113 |
use_fast = true
|
114 |
|
115 |
+
[components.transformer.model.transformer_config]
|
116 |
+
|
117 |
[corpora]
|
118 |
|
119 |
[corpora.dev]
|
120 |
@readers = "spacy.Corpus.v1"
|
121 |
+
path = ${paths.dev}
|
|
|
|
|
122 |
gold_preproc = false
|
123 |
+
max_length = 0
|
124 |
+
limit = 0
|
125 |
augmenter = null
|
126 |
|
127 |
[corpora.train]
|
128 |
@readers = "spacy.Corpus.v1"
|
129 |
+
path = ${paths.train}
|
|
|
130 |
gold_preproc = false
|
131 |
+
max_length = 0
|
132 |
limit = 0
|
133 |
+
augmenter = null
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
134 |
|
135 |
[training]
|
136 |
train_corpus = "corpora.train"
|
|
|
187 |
ents_p = 0.0
|
188 |
ents_r = 0.0
|
189 |
ents_per_type = null
|
190 |
+
speed = 0.0
|
191 |
|
192 |
[pretraining]
|
193 |
|
194 |
[initialize]
|
195 |
+
vocab_data = null
|
196 |
vectors = ${paths.vectors}
|
197 |
init_tok2vec = ${paths.init_tok2vec}
|
198 |
before_init = null
|
en_core_web_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:c6be2ccfc1c30edb5690f5e2baecf2a0cabd23529c0abb84c67aedcf0dac86a2
|
3 |
+
size 460235340
|
meta.json
CHANGED
@@ -1,14 +1,14 @@
|
|
1 |
{
|
2 |
"lang":"en",
|
3 |
"name":"core_web_trf",
|
4 |
-
"version":"3.
|
5 |
"description":"English transformer pipeline (roberta-base). Components: transformer, tagger, parser, ner, attribute_ruler, lemmatizer.",
|
6 |
"author":"Explosion",
|
7 |
"email":"contact@explosion.ai",
|
8 |
"url":"https://explosion.ai",
|
9 |
"license":"MIT",
|
10 |
-
"spacy_version":">=3.
|
11 |
-
"spacy_git_version":"
|
12 |
"vectors":{
|
13 |
"width":0,
|
14 |
"vectors":0,
|
@@ -165,330 +165,333 @@
|
|
165 |
],
|
166 |
"performance":{
|
167 |
"token_acc":0.9993053983,
|
168 |
-
"
|
169 |
-
"
|
170 |
-
"
|
171 |
-
"
|
172 |
-
"
|
173 |
-
"
|
174 |
-
"
|
175 |
-
"
|
176 |
-
"
|
177 |
-
"speed":5467.6411211076,
|
178 |
"dep_las_per_type":{
|
179 |
"prep":{
|
180 |
-
"p":0.
|
181 |
-
"r":0.
|
182 |
-
"f":0.
|
183 |
},
|
184 |
"det":{
|
185 |
-
"p":0.
|
186 |
-
"r":0.
|
187 |
-
"f":0.
|
188 |
},
|
189 |
"pobj":{
|
190 |
-
"p":0.
|
191 |
-
"r":0.
|
192 |
-
"f":0.
|
193 |
},
|
194 |
"nsubj":{
|
195 |
-
"p":0.
|
196 |
-
"r":0.
|
197 |
-
"f":0.
|
198 |
},
|
199 |
"aux":{
|
200 |
-
"p":0.
|
201 |
-
"r":0.
|
202 |
-
"f":0.
|
203 |
},
|
204 |
"advmod":{
|
205 |
-
"p":0.
|
206 |
-
"r":0.
|
207 |
-
"f":0.
|
208 |
},
|
209 |
"relcl":{
|
210 |
-
"p":0.
|
211 |
-
"r":0.
|
212 |
-
"f":0.
|
213 |
},
|
214 |
"root":{
|
215 |
-
"p":0.
|
216 |
-
"r":0.
|
217 |
-
"f":0.
|
218 |
},
|
219 |
"xcomp":{
|
220 |
-
"p":0.
|
221 |
-
"r":0.
|
222 |
-
"f":0.
|
223 |
},
|
224 |
"amod":{
|
225 |
-
"p":0.
|
226 |
-
"r":0.
|
227 |
-
"f":0.
|
228 |
},
|
229 |
"compound":{
|
230 |
-
"p":0.
|
231 |
-
"r":0.
|
232 |
-
"f":0.
|
233 |
},
|
234 |
"poss":{
|
235 |
-
"p":0.
|
236 |
-
"r":0.
|
237 |
-
"f":0.
|
238 |
},
|
239 |
"ccomp":{
|
240 |
-
"p":0.
|
241 |
-
"r":0.
|
242 |
-
"f":0.
|
243 |
},
|
244 |
"attr":{
|
245 |
-
"p":0.
|
246 |
-
"r":0.
|
247 |
-
"f":0.
|
248 |
},
|
249 |
"case":{
|
250 |
-
"p":0.
|
251 |
-
"r":0.
|
252 |
-
"f":0.
|
253 |
},
|
254 |
"mark":{
|
255 |
-
"p":0.
|
256 |
-
"r":0.
|
257 |
-
"f":0.
|
258 |
},
|
259 |
"intj":{
|
260 |
-
"p":0.
|
261 |
-
"r":0.
|
262 |
-
"f":0.
|
263 |
},
|
264 |
"advcl":{
|
265 |
-
"p":0.
|
266 |
-
"r":0.
|
267 |
-
"f":0.
|
268 |
},
|
269 |
"cc":{
|
270 |
-
"p":0.
|
271 |
-
"r":0.
|
272 |
-
"f":0.
|
273 |
},
|
274 |
"neg":{
|
275 |
-
"p":0.
|
276 |
-
"r":0.
|
277 |
-
"f":0.
|
278 |
},
|
279 |
"conj":{
|
280 |
-
"p":0.
|
281 |
-
"r":0.
|
282 |
-
"f":0.
|
283 |
},
|
284 |
"nsubjpass":{
|
285 |
-
"p":0.
|
286 |
-
"r":0.
|
287 |
-
"f":0.
|
288 |
},
|
289 |
"auxpass":{
|
290 |
-
"p":0.
|
291 |
-
"r":0.
|
292 |
-
"f":0.
|
293 |
},
|
294 |
"dobj":{
|
295 |
-
"p":0.
|
296 |
-
"r":0.
|
297 |
-
"f":0.
|
298 |
},
|
299 |
"nummod":{
|
300 |
-
"p":0.
|
301 |
-
"r":0.
|
302 |
-
"f":0.
|
303 |
},
|
304 |
"npadvmod":{
|
305 |
-
"p":0.
|
306 |
-
"r":0.
|
307 |
-
"f":0.
|
308 |
},
|
309 |
"prt":{
|
310 |
-
"p":0.
|
311 |
-
"r":0.
|
312 |
-
"f":0.
|
313 |
},
|
314 |
"pcomp":{
|
315 |
-
"p":0.
|
316 |
-
"r":0.
|
317 |
-
"f":0.
|
318 |
},
|
319 |
"expl":{
|
320 |
-
"p":0.
|
321 |
"r":0.9957173448,
|
322 |
-
"f":0.
|
323 |
},
|
324 |
"acl":{
|
325 |
-
"p":0.
|
326 |
-
"r":0.
|
327 |
-
"f":0.
|
328 |
},
|
329 |
"agent":{
|
330 |
-
"p":0.
|
331 |
-
"r":0.
|
332 |
-
"f":0.
|
333 |
},
|
334 |
"dative":{
|
335 |
-
"p":0.
|
336 |
-
"r":0.
|
337 |
-
"f":0.
|
338 |
},
|
339 |
"acomp":{
|
340 |
-
"p":0.
|
341 |
-
"r":0.
|
342 |
-
"f":0.
|
343 |
},
|
344 |
"dep":{
|
345 |
-
"p":0.
|
346 |
-
"r":0.
|
347 |
-
"f":0.
|
348 |
},
|
349 |
"csubj":{
|
350 |
-
"p":0.
|
351 |
-
"r":0.
|
352 |
-
"f":0.
|
353 |
},
|
354 |
"quantmod":{
|
355 |
-
"p":0.
|
356 |
-
"r":0.
|
357 |
-
"f":0.
|
358 |
},
|
359 |
"nmod":{
|
360 |
-
"p":0.
|
361 |
-
"r":0.
|
362 |
-
"f":0.
|
363 |
},
|
364 |
"appos":{
|
365 |
-
"p":0.
|
366 |
-
"r":0.
|
367 |
-
"f":0.
|
368 |
},
|
369 |
"predet":{
|
370 |
-
"p":0.
|
371 |
-
"r":0.
|
372 |
-
"f":0.
|
373 |
},
|
374 |
"preconj":{
|
375 |
-
"p":0.
|
376 |
-
"r":0.
|
377 |
-
"f":0.
|
378 |
},
|
379 |
"oprd":{
|
380 |
-
"p":0.
|
381 |
-
"r":0.
|
382 |
-
"f":0.
|
383 |
},
|
384 |
"parataxis":{
|
385 |
-
"p":0.
|
386 |
-
"r":0.
|
387 |
-
"f":0.
|
388 |
},
|
389 |
"meta":{
|
390 |
-
"p":0.
|
391 |
"r":0.6538461538,
|
392 |
-
"f":0.
|
393 |
},
|
394 |
"csubjpass":{
|
395 |
-
"p":
|
396 |
"r":0.8333333333,
|
397 |
-
"f":0.
|
398 |
}
|
399 |
},
|
|
|
|
|
|
|
400 |
"ents_per_type":{
|
401 |
"DATE":{
|
402 |
-
"p":0.
|
403 |
-
"r":0.
|
404 |
-
"f":0.
|
405 |
},
|
406 |
"GPE":{
|
407 |
-
"p":0.
|
408 |
-
"r":0.
|
409 |
-
"f":0.
|
410 |
},
|
411 |
"ORDINAL":{
|
412 |
-
"p":0.
|
413 |
"r":0.8229813665,
|
414 |
-
"f":0.
|
415 |
},
|
416 |
"ORG":{
|
417 |
-
"p":0.
|
418 |
-
"r":0.
|
419 |
-
"f":0.
|
420 |
},
|
421 |
"FAC":{
|
422 |
-
"p":0.
|
423 |
-
"r":0.
|
424 |
-
"f":0.
|
425 |
},
|
426 |
"QUANTITY":{
|
427 |
-
"p":0.
|
428 |
-
"r":0.
|
429 |
-
"f":0.
|
430 |
},
|
431 |
"LOC":{
|
432 |
-
"p":0.
|
433 |
-
"r":0.
|
434 |
-
"f":0.
|
435 |
},
|
436 |
"CARDINAL":{
|
437 |
-
"p":0.
|
438 |
-
"r":0.
|
439 |
-
"f":0.
|
440 |
},
|
441 |
"PERSON":{
|
442 |
-
"p":0.
|
443 |
-
"r":0.
|
444 |
-
"f":0.
|
445 |
},
|
446 |
"NORP":{
|
447 |
-
"p":0.
|
448 |
-
"r":0.
|
449 |
-
"f":0.
|
450 |
},
|
451 |
"LAW":{
|
452 |
-
"p":0.
|
453 |
-
"r":0.
|
454 |
-
"f":0.
|
455 |
},
|
456 |
"PRODUCT":{
|
457 |
-
"p":0.
|
458 |
"r":0.5260663507,
|
459 |
-
"f":0.
|
|
|
|
|
|
|
|
|
|
|
460 |
},
|
461 |
"TIME":{
|
462 |
-
"p":0.
|
463 |
"r":0.7777777778,
|
464 |
-
"f":0.
|
465 |
},
|
466 |
"EVENT":{
|
467 |
-
"p":0.
|
468 |
-
"r":0.
|
469 |
-
"f":0.
|
470 |
},
|
471 |
"WORK_OF_ART":{
|
472 |
-
"p":0.
|
473 |
-
"r":0.
|
474 |
-
"f":0.
|
475 |
-
},
|
476 |
-
"MONEY":{
|
477 |
-
"p":0.9120481928,
|
478 |
-
"r":0.893742621,
|
479 |
-
"f":0.9028026237
|
480 |
},
|
481 |
"PERCENT":{
|
482 |
-
"p":0.
|
483 |
-
"r":0.
|
484 |
-
"f":0.
|
485 |
},
|
486 |
"LANGUAGE":{
|
487 |
"p":1.0,
|
488 |
-
"r":0.
|
489 |
-
"f":0.
|
490 |
}
|
491 |
-
}
|
|
|
492 |
},
|
493 |
"sources":[
|
494 |
{
|
@@ -517,6 +520,6 @@
|
|
517 |
}
|
518 |
],
|
519 |
"requirements":[
|
520 |
-
"spacy-transformers>=1.
|
521 |
]
|
522 |
}
|
|
|
1 |
{
|
2 |
"lang":"en",
|
3 |
"name":"core_web_trf",
|
4 |
+
"version":"3.2.0",
|
5 |
"description":"English transformer pipeline (roberta-base). Components: transformer, tagger, parser, ner, attribute_ruler, lemmatizer.",
|
6 |
"author":"Explosion",
|
7 |
"email":"contact@explosion.ai",
|
8 |
"url":"https://explosion.ai",
|
9 |
"license":"MIT",
|
10 |
+
"spacy_version":">=3.2.0,<3.3.0",
|
11 |
+
"spacy_git_version":"bb26550e2",
|
12 |
"vectors":{
|
13 |
"width":0,
|
14 |
"vectors":0,
|
|
|
165 |
],
|
166 |
"performance":{
|
167 |
"token_acc":0.9993053983,
|
168 |
+
"token_p":0.9956742163,
|
169 |
+
"token_r":0.9957505887,
|
170 |
+
"token_f":0.9957124011,
|
171 |
+
"tag_acc":0.9780627094,
|
172 |
+
"sents_p":0.9568827385,
|
173 |
+
"sents_r":0.8666798601,
|
174 |
+
"sents_f":0.9095503479,
|
175 |
+
"dep_uas":0.9527056513,
|
176 |
+
"dep_las":0.9393298266,
|
|
|
177 |
"dep_las_per_type":{
|
178 |
"prep":{
|
179 |
+
"p":0.9229604964,
|
180 |
+
"r":0.9255169797,
|
181 |
+
"f":0.9242369702
|
182 |
},
|
183 |
"det":{
|
184 |
+
"p":0.9898046572,
|
185 |
+
"r":0.9896028704,
|
186 |
+
"f":0.9897037535
|
187 |
},
|
188 |
"pobj":{
|
189 |
+
"p":0.9832869081,
|
190 |
+
"r":0.9842136265,
|
191 |
+
"f":0.9837500491
|
192 |
},
|
193 |
"nsubj":{
|
194 |
+
"p":0.9803139249,
|
195 |
+
"r":0.9795837897,
|
196 |
+
"f":0.9799487213
|
197 |
},
|
198 |
"aux":{
|
199 |
+
"p":0.989571263,
|
200 |
+
"r":0.9883379329,
|
201 |
+
"f":0.9889542134
|
202 |
},
|
203 |
"advmod":{
|
204 |
+
"p":0.8962161255,
|
205 |
+
"r":0.8987043581,
|
206 |
+
"f":0.8974585171
|
207 |
},
|
208 |
"relcl":{
|
209 |
+
"p":0.8798724309,
|
210 |
+
"r":0.9009433962,
|
211 |
+
"f":0.8902832556
|
212 |
},
|
213 |
"root":{
|
214 |
+
"p":0.9664043142,
|
215 |
+
"r":0.8747938518,
|
216 |
+
"f":0.9183200028
|
217 |
},
|
218 |
"xcomp":{
|
219 |
+
"p":0.9438444924,
|
220 |
+
"r":0.9411342426,
|
221 |
+
"f":0.9424874191
|
222 |
},
|
223 |
"amod":{
|
224 |
+
"p":0.9413514387,
|
225 |
+
"r":0.943181082,
|
226 |
+
"f":0.9422653722
|
227 |
},
|
228 |
"compound":{
|
229 |
+
"p":0.951964576,
|
230 |
+
"r":0.9458676765,
|
231 |
+
"f":0.9489063329
|
232 |
},
|
233 |
"poss":{
|
234 |
+
"p":0.986728333,
|
235 |
+
"r":0.9877214171,
|
236 |
+
"f":0.9872246253
|
237 |
},
|
238 |
"ccomp":{
|
239 |
+
"p":0.8604651163,
|
240 |
+
"r":0.9118126273,
|
241 |
+
"f":0.8853950361
|
242 |
},
|
243 |
"attr":{
|
244 |
+
"p":0.9478225144,
|
245 |
+
"r":0.9701429773,
|
246 |
+
"f":0.9588528678
|
247 |
},
|
248 |
"case":{
|
249 |
+
"p":0.9890547264,
|
250 |
+
"r":0.994994995,
|
251 |
+
"f":0.9920159681
|
252 |
},
|
253 |
"mark":{
|
254 |
+
"p":0.9492105263,
|
255 |
+
"r":0.9557498675,
|
256 |
+
"f":0.9524689728
|
257 |
},
|
258 |
"intj":{
|
259 |
+
"p":0.6237388724,
|
260 |
+
"r":0.76996337,
|
261 |
+
"f":0.6891803279
|
262 |
},
|
263 |
"advcl":{
|
264 |
+
"p":0.8195836546,
|
265 |
+
"r":0.803072274,
|
266 |
+
"f":0.8112439583
|
267 |
},
|
268 |
"cc":{
|
269 |
+
"p":0.8986197049,
|
270 |
+
"r":0.9032412391,
|
271 |
+
"f":0.9009245452
|
272 |
},
|
273 |
"neg":{
|
274 |
+
"p":0.9574148297,
|
275 |
+
"r":0.958855996,
|
276 |
+
"f":0.9581348709
|
277 |
},
|
278 |
"conj":{
|
279 |
+
"p":0.8712776177,
|
280 |
+
"r":0.9133937563,
|
281 |
+
"f":0.8918387414
|
282 |
},
|
283 |
"nsubjpass":{
|
284 |
+
"p":0.9616182573,
|
285 |
+
"r":0.9507692308,
|
286 |
+
"f":0.9561629706
|
287 |
},
|
288 |
"auxpass":{
|
289 |
+
"p":0.9676979072,
|
290 |
+
"r":0.9690205011,
|
291 |
+
"f":0.9683587526
|
292 |
},
|
293 |
"dobj":{
|
294 |
+
"p":0.9742177522,
|
295 |
+
"r":0.9725874572,
|
296 |
+
"f":0.9734019221
|
297 |
},
|
298 |
"nummod":{
|
299 |
+
"p":0.9582800102,
|
300 |
+
"r":0.9454545455,
|
301 |
+
"f":0.9518240753
|
302 |
},
|
303 |
"npadvmod":{
|
304 |
+
"p":0.8582474227,
|
305 |
+
"r":0.8280639432,
|
306 |
+
"f":0.8428855541
|
307 |
},
|
308 |
"prt":{
|
309 |
+
"p":0.896,
|
310 |
+
"r":0.9032258065,
|
311 |
+
"f":0.8995983936
|
312 |
},
|
313 |
"pcomp":{
|
314 |
+
"p":0.9316901408,
|
315 |
+
"r":0.9264705882,
|
316 |
+
"f":0.9290730337
|
317 |
},
|
318 |
"expl":{
|
319 |
+
"p":0.9893617021,
|
320 |
"r":0.9957173448,
|
321 |
+
"f":0.992529349
|
322 |
},
|
323 |
"acl":{
|
324 |
+
"p":0.8476138829,
|
325 |
+
"r":0.852700491,
|
326 |
+
"f":0.8501495785
|
327 |
},
|
328 |
"agent":{
|
329 |
+
"p":0.9559859155,
|
330 |
+
"r":0.9731182796,
|
331 |
+
"f":0.9644760213
|
332 |
},
|
333 |
"dative":{
|
334 |
+
"p":0.8236607143,
|
335 |
+
"r":0.8463302752,
|
336 |
+
"f":0.834841629
|
337 |
},
|
338 |
"acomp":{
|
339 |
+
"p":0.9352355072,
|
340 |
+
"r":0.9365079365,
|
341 |
+
"f":0.9358712894
|
342 |
},
|
343 |
"dep":{
|
344 |
+
"p":0.3949044586,
|
345 |
+
"r":0.4025974026,
|
346 |
+
"f":0.3987138264
|
347 |
},
|
348 |
"csubj":{
|
349 |
+
"p":0.8928571429,
|
350 |
+
"r":0.8875739645,
|
351 |
+
"f":0.8902077151
|
352 |
},
|
353 |
"quantmod":{
|
354 |
+
"p":0.8867449664,
|
355 |
+
"r":0.8586515028,
|
356 |
+
"f":0.872472142
|
357 |
},
|
358 |
"nmod":{
|
359 |
+
"p":0.82904795,
|
360 |
+
"r":0.7269957343,
|
361 |
+
"f":0.7746753247
|
362 |
},
|
363 |
"appos":{
|
364 |
+
"p":0.8099701747,
|
365 |
+
"r":0.8247288503,
|
366 |
+
"f":0.8172828891
|
367 |
},
|
368 |
"predet":{
|
369 |
+
"p":0.8508064516,
|
370 |
+
"r":0.9055793991,
|
371 |
+
"f":0.8773388773
|
372 |
},
|
373 |
"preconj":{
|
374 |
+
"p":0.7083333333,
|
375 |
+
"r":0.7906976744,
|
376 |
+
"f":0.7472527473
|
377 |
},
|
378 |
"oprd":{
|
379 |
+
"p":0.9022082019,
|
380 |
+
"r":0.8537313433,
|
381 |
+
"f":0.8773006135
|
382 |
},
|
383 |
"parataxis":{
|
384 |
+
"p":0.5602189781,
|
385 |
+
"r":0.6659436009,
|
386 |
+
"f":0.6085232904
|
387 |
},
|
388 |
"meta":{
|
389 |
+
"p":0.1683168317,
|
390 |
"r":0.6538461538,
|
391 |
+
"f":0.2677165354
|
392 |
},
|
393 |
"csubjpass":{
|
394 |
+
"p":0.8333333333,
|
395 |
"r":0.8333333333,
|
396 |
+
"f":0.8333333333
|
397 |
}
|
398 |
},
|
399 |
+
"ents_p":0.898708838,
|
400 |
+
"ents_r":0.8993389423,
|
401 |
+
"ents_f":0.8990237797,
|
402 |
"ents_per_type":{
|
403 |
"DATE":{
|
404 |
+
"p":0.8928683632,
|
405 |
+
"r":0.9022222222,
|
406 |
+
"f":0.8975209222
|
407 |
},
|
408 |
"GPE":{
|
409 |
+
"p":0.9575551783,
|
410 |
+
"r":0.9439330544,
|
411 |
+
"f":0.9506953224
|
412 |
},
|
413 |
"ORDINAL":{
|
414 |
+
"p":0.8079268293,
|
415 |
"r":0.8229813665,
|
416 |
+
"f":0.8153846154
|
417 |
},
|
418 |
"ORG":{
|
419 |
+
"p":0.9032515991,
|
420 |
+
"r":0.8984623542,
|
421 |
+
"f":0.9008506114
|
422 |
},
|
423 |
"FAC":{
|
424 |
+
"p":0.5567010309,
|
425 |
+
"r":0.8307692308,
|
426 |
+
"f":0.6666666667
|
427 |
},
|
428 |
"QUANTITY":{
|
429 |
+
"p":0.8187134503,
|
430 |
+
"r":0.7692307692,
|
431 |
+
"f":0.7932011331
|
432 |
},
|
433 |
"LOC":{
|
434 |
+
"p":0.8610169492,
|
435 |
+
"r":0.8089171975,
|
436 |
+
"f":0.8341543514
|
437 |
},
|
438 |
"CARDINAL":{
|
439 |
+
"p":0.8530092593,
|
440 |
+
"r":0.8763376932,
|
441 |
+
"f":0.864516129
|
442 |
},
|
443 |
"PERSON":{
|
444 |
+
"p":0.9436258808,
|
445 |
+
"r":0.9614882507,
|
446 |
+
"f":0.9524733269
|
447 |
},
|
448 |
"NORP":{
|
449 |
+
"p":0.9211783439,
|
450 |
+
"r":0.9256,
|
451 |
+
"f":0.9233838787
|
452 |
},
|
453 |
"LAW":{
|
454 |
+
"p":0.5324675325,
|
455 |
+
"r":0.640625,
|
456 |
+
"f":0.5815602837
|
457 |
},
|
458 |
"PRODUCT":{
|
459 |
+
"p":0.649122807,
|
460 |
"r":0.5260663507,
|
461 |
+
"f":0.5811518325
|
462 |
+
},
|
463 |
+
"MONEY":{
|
464 |
+
"p":0.9157641396,
|
465 |
+
"r":0.8984651712,
|
466 |
+
"f":0.9070321812
|
467 |
},
|
468 |
"TIME":{
|
469 |
+
"p":0.7514124294,
|
470 |
"r":0.7777777778,
|
471 |
+
"f":0.7643678161
|
472 |
},
|
473 |
"EVENT":{
|
474 |
+
"p":0.776119403,
|
475 |
+
"r":0.5977011494,
|
476 |
+
"f":0.6753246753
|
477 |
},
|
478 |
"WORK_OF_ART":{
|
479 |
+
"p":0.5684210526,
|
480 |
+
"r":0.5567010309,
|
481 |
+
"f":0.5625
|
|
|
|
|
|
|
|
|
|
|
482 |
},
|
483 |
"PERCENT":{
|
484 |
+
"p":0.9154929577,
|
485 |
+
"r":0.8958652374,
|
486 |
+
"f":0.9055727554
|
487 |
},
|
488 |
"LANGUAGE":{
|
489 |
"p":1.0,
|
490 |
+
"r":0.71875,
|
491 |
+
"f":0.8363636364
|
492 |
}
|
493 |
+
},
|
494 |
+
"speed":6163.7320781314
|
495 |
},
|
496 |
"sources":[
|
497 |
{
|
|
|
520 |
}
|
521 |
],
|
522 |
"requirements":[
|
523 |
+
"spacy-transformers>=1.1.2,<1.2.0"
|
524 |
]
|
525 |
}
|
ner/model
CHANGED
Binary files a/ner/model and b/ner/model differ
|
|
parser/model
CHANGED
Binary files a/parser/model and b/parser/model differ
|
|
tagger/cfg
CHANGED
@@ -49,5 +49,6 @@
|
|
49 |
"WRB",
|
50 |
"XX",
|
51 |
"``"
|
52 |
-
]
|
|
|
53 |
}
|
|
|
49 |
"WRB",
|
50 |
"XX",
|
51 |
"``"
|
52 |
+
],
|
53 |
+
"overwrite":false
|
54 |
}
|
tagger/model
CHANGED
Binary files a/tagger/model and b/tagger/model differ
|
|
tokenizer
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
transformer/{model/pytorch_model.bin → model}
RENAMED
@@ -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 sha256:523fefd9ecd5f2a76cb7f686222af0615942bc51d131339f5883a6190e7fdc72
|
3 |
+
size 501274094
|
transformer/model/config.json
DELETED
@@ -1,26 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"_name_or_path": "/mnt/scratch/tmp/en_core_web_trf/dc04c7b2-9045-42b5-95f8-fa28249ae3b8/training/core/model-best/transformer/model",
|
3 |
-
"architectures": [
|
4 |
-
"RobertaForMaskedLM"
|
5 |
-
],
|
6 |
-
"attention_probs_dropout_prob": 0.1,
|
7 |
-
"bos_token_id": 0,
|
8 |
-
"eos_token_id": 2,
|
9 |
-
"gradient_checkpointing": false,
|
10 |
-
"hidden_act": "gelu",
|
11 |
-
"hidden_dropout_prob": 0.1,
|
12 |
-
"hidden_size": 768,
|
13 |
-
"initializer_range": 0.02,
|
14 |
-
"intermediate_size": 3072,
|
15 |
-
"layer_norm_eps": 1e-05,
|
16 |
-
"max_position_embeddings": 514,
|
17 |
-
"model_type": "roberta",
|
18 |
-
"num_attention_heads": 12,
|
19 |
-
"num_hidden_layers": 12,
|
20 |
-
"pad_token_id": 1,
|
21 |
-
"position_embedding_type": "absolute",
|
22 |
-
"transformers_version": "4.6.1",
|
23 |
-
"type_vocab_size": 1,
|
24 |
-
"use_cache": true,
|
25 |
-
"vocab_size": 50265
|
26 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
transformer/model/merges.txt
DELETED
The diff for this file is too large to render.
See raw diff
|
|
transformer/model/special_tokens_map.json
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}
|
|
|
|
transformer/model/tokenizer.json
DELETED
The diff for this file is too large to render.
See raw diff
|
|
transformer/model/tokenizer_config.json
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "add_prefix_space": false, "errors": "replace", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": "<mask>", "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "/mnt/scratch/tmp/en_core_web_trf/dc04c7b2-9045-42b5-95f8-fa28249ae3b8/training/core/model-best/transformer/model"}
|
|
|
|
transformer/model/vocab.json
DELETED
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:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f6fda43452971a53ac720650d08c74297656aec52627651521ce4fbceaabce87
|
3 |
+
size 1091378
|
vocab/vectors.cfg
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"mode":"default"
|
3 |
+
}
|