Initial commit
Browse files- config.json +610 -0
- model_args.json +1 -0
- special_tokens_map.json +7 -0
- vocab.txt +0 -0
config.json
ADDED
@@ -0,0 +1,610 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "classla/bcms-bertic",
|
3 |
+
"architectures": [
|
4 |
+
"ElectraForTokenClassification"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"classifier_dropout": null,
|
8 |
+
"embedding_size": 768,
|
9 |
+
"hidden_act": "gelu",
|
10 |
+
"hidden_dropout_prob": 0.1,
|
11 |
+
"hidden_size": 768,
|
12 |
+
"id2label": {
|
13 |
+
"0": "[0, '', 0, '']",
|
14 |
+
"1": "[1, 'evar', 0, '']",
|
15 |
+
"2": "[1, 'a', 0, '']",
|
16 |
+
"3": "[1, '', 0, '']",
|
17 |
+
"4": "[1, 'oj', 0, '']",
|
18 |
+
"5": "[2, '', 0, '']",
|
19 |
+
"6": "[6, 'ao', 0, '']",
|
20 |
+
"7": "[3, 'ivati', 0, '']",
|
21 |
+
"8": "[0, 'ti', 0, '']",
|
22 |
+
"9": "[0, 'be', 0, '']",
|
23 |
+
"10": "[2, 'an', 0, '']",
|
24 |
+
"11": "[4, 'ao', 0, '']",
|
25 |
+
"12": "[1, 'i', 0, '']",
|
26 |
+
"13": "[2, 'biti', 0, '']",
|
27 |
+
"14": "[2, 'iti', 0, '']",
|
28 |
+
"15": "[2, 'aj', 0, '']",
|
29 |
+
"16": "[1, 'e', 0, '']",
|
30 |
+
"17": "[3, 'an', 0, '']",
|
31 |
+
"18": "[2, 'ar', 0, '']",
|
32 |
+
"19": "[2, 'ji', 0, '']",
|
33 |
+
"20": "[1, 'eti', 0, '']",
|
34 |
+
"21": "[2, 'ti', 0, '']",
|
35 |
+
"22": "[3, 'lac', 0, '']",
|
36 |
+
"23": "[1, 'iti', 0, '']",
|
37 |
+
"24": "[1, 'o', 0, '']",
|
38 |
+
"25": "[2, 'av', 0, '']",
|
39 |
+
"26": "[3, 'iti', 0, '']",
|
40 |
+
"27": "[4, 'ti', 0, '']",
|
41 |
+
"28": "[4, 'iti', 0, '']",
|
42 |
+
"29": "[1, 'ji', 0, '']",
|
43 |
+
"30": "[3, '', 0, '']",
|
44 |
+
"31": "[2, 'ak', 0, '']",
|
45 |
+
"32": "[6, 'ak', 0, '']",
|
46 |
+
"33": "[5, '', 0, '']",
|
47 |
+
"34": "[3, 'o', 0, '']",
|
48 |
+
"35": "[3, 'aj', 0, '']",
|
49 |
+
"36": "[0, 'ti', 1, 'ht']",
|
50 |
+
"37": "[2, 'ti', 1, 'ht']",
|
51 |
+
"38": "[3, 'ak', 0, '']",
|
52 |
+
"39": "[2, 'i', 0, '']",
|
53 |
+
"40": "[3, 'ovati', 0, '']",
|
54 |
+
"41": "[1, 'sti', 0, '']",
|
55 |
+
"42": "[3, 'ka', 0, '']",
|
56 |
+
"43": "[2, 'a', 0, '']",
|
57 |
+
"44": "[4, 'li', 0, '']",
|
58 |
+
"45": "[2, '\u0107i', 0, '']",
|
59 |
+
"46": "[3, 'av', 0, '']",
|
60 |
+
"47": "[4, 'k', 0, '']",
|
61 |
+
"48": "[0, 'j', 0, '']",
|
62 |
+
"49": "[3, 'e', 0, '']",
|
63 |
+
"50": "[4, 'li', 3, '']",
|
64 |
+
"51": "[3, '', 0, 'o']",
|
65 |
+
"52": "[1, 'aj', 0, '']",
|
66 |
+
"53": "[2, 'eti', 0, '']",
|
67 |
+
"54": "[3, 'lik', 3, '']",
|
68 |
+
"55": "[2, 'zati', 0, '']",
|
69 |
+
"56": "[4, 'biti', 0, '']",
|
70 |
+
"57": "[1, '', 0, 'on']",
|
71 |
+
"58": "[1, 'ti', 0, '']",
|
72 |
+
"59": "[2, 'ti', 1, 'b']",
|
73 |
+
"60": "[5, 'iti', 0, '']",
|
74 |
+
"61": "[1, 'j', 0, '']",
|
75 |
+
"62": "[2, 'ka', 0, '']",
|
76 |
+
"63": "[5, 'ivati', 0, '']",
|
77 |
+
"64": "[4, 'ak', 0, '']",
|
78 |
+
"65": "[2, 'vati', 0, '']",
|
79 |
+
"66": "[3, 'st', 0, '']",
|
80 |
+
"67": "[3, 'ek', 0, '']",
|
81 |
+
"68": "[0, 'a', 0, '']",
|
82 |
+
"69": "[2, 'at', 0, '']",
|
83 |
+
"70": "[2, 'ac', 0, '']",
|
84 |
+
"71": "[1, 'ica', 0, '']",
|
85 |
+
"72": "[1, 'i\u010dan', 0, '']",
|
86 |
+
"73": "[1, 'arta', 0, '']",
|
87 |
+
"74": "[2, 'k', 0, '']",
|
88 |
+
"75": "[2, 'o', 0, '']",
|
89 |
+
"76": "[1, 'astarski', 0, '']",
|
90 |
+
"77": "[1, 'ojan', 0, '']",
|
91 |
+
"78": "[4, 'o', 0, '']",
|
92 |
+
"79": "[4, 'ac', 0, '']",
|
93 |
+
"80": "[3, 'o', 3, '']",
|
94 |
+
"81": "[3, 'vo', 0, '']",
|
95 |
+
"82": "[2, 'cati', 0, '']",
|
96 |
+
"83": "[1, 'ati', 0, '']",
|
97 |
+
"84": "[2, 'tak', 0, '']",
|
98 |
+
"85": "[2, 'on', 0, '']",
|
99 |
+
"86": "[2, 'ao', 0, '']",
|
100 |
+
"87": "[3, '\u0107i', 0, '']",
|
101 |
+
"88": "[3, 'ac', 0, '']",
|
102 |
+
"89": "[3, 'ar', 0, '']",
|
103 |
+
"90": "[4, '', 0, 'o']",
|
104 |
+
"91": "[3, 't', 0, '']",
|
105 |
+
"92": "[2, 'sok', 0, '']",
|
106 |
+
"93": "[2, '.o.o.', 0, '']",
|
107 |
+
"94": "[3, 'lo', 0, '']",
|
108 |
+
"95": "[3, 'ati', 0, '']",
|
109 |
+
"96": "[3, 'biti', 0, '']",
|
110 |
+
"97": "[2, 'g', 0, '']",
|
111 |
+
"98": "[5, 'an', 3, '']",
|
112 |
+
"99": "[1, 'istarski', 0, '']",
|
113 |
+
"100": "[3, 'tan', 0, '']",
|
114 |
+
"101": "[0, '', 3, '\u0161t']",
|
115 |
+
"102": "[4, '\u0161ta', 0, '']",
|
116 |
+
"103": "[2, 'sti', 0, '']",
|
117 |
+
"104": "[5, '\u0107i', 0, '']",
|
118 |
+
"105": "[1, 'an', 0, '']",
|
119 |
+
"106": "[2, 'lik', 0, '']",
|
120 |
+
"107": "[1, 'rokuplje', 0, '']",
|
121 |
+
"108": "[1, 'epublika', 0, '']",
|
122 |
+
"109": "[3, 'sti', 0, '']",
|
123 |
+
"110": "[2, 'eko', 0, '']",
|
124 |
+
"111": "[4, 'diti', 0, '']",
|
125 |
+
"112": "[2, 'ga', 0, '']",
|
126 |
+
"113": "[4, 'lac', 0, '']",
|
127 |
+
"114": "[0, 'i', 0, '']",
|
128 |
+
"115": "[4, 'av', 0, '']",
|
129 |
+
"116": "[2, 'zu', 0, '']",
|
130 |
+
"117": "[3, 'zak', 0, '']",
|
131 |
+
"118": "[4, 'titi', 0, '']",
|
132 |
+
"119": "[3, 'tak', 3, '']",
|
133 |
+
"120": "[4, 'vati', 0, '']",
|
134 |
+
"121": "[3, 'lo', 3, '']",
|
135 |
+
"122": "[3, 'ba', 0, '']",
|
136 |
+
"123": "[1, 'u\u017eben', 0, '']",
|
137 |
+
"124": "[1, 'ugi', 0, '']",
|
138 |
+
"125": "[4, 'at', 0, '']",
|
139 |
+
"126": "[3, 'titi', 0, '']",
|
140 |
+
"127": "[2, 'sati', 0, '']",
|
141 |
+
"128": "[4, 'deti', 0, '']",
|
142 |
+
"129": "[5, 'stiti', 0, '']",
|
143 |
+
"130": "[6, 'ogo', 0, 'm']",
|
144 |
+
"131": "[3, 'deti', 0, '']",
|
145 |
+
"132": "[4, 'an', 0, '']",
|
146 |
+
"133": "[2, 'tan', 0, '']",
|
147 |
+
"134": "[2, 'gati', 0, '']",
|
148 |
+
"135": "[1, 'i', 3, 'bi']",
|
149 |
+
"136": "[4, 'mnogo', 0, '']",
|
150 |
+
"137": "[4, 'g', 0, '']",
|
151 |
+
"138": "[4, '', 0, '']",
|
152 |
+
"139": "[2, 'ek', 0, '']",
|
153 |
+
"140": "[2, 'nuti', 0, '']",
|
154 |
+
"141": "[5, 'lac', 0, '']",
|
155 |
+
"142": "[1, 'uti', 0, '']",
|
156 |
+
"143": "[3, 'li', 0, '']",
|
157 |
+
"144": "[1, 'i\u010dno', 0, '']",
|
158 |
+
"145": "[6, 'r', 1, 'dob']",
|
159 |
+
"146": "[4, '\u0161to', 0, '']",
|
160 |
+
"147": "[2, 'zak', 0, '']",
|
161 |
+
"148": "[1, 'r\u0161ilac', 0, '']",
|
162 |
+
"149": "[1, 'u\u017enost', 0, '']",
|
163 |
+
"150": "[1, 'av', 0, '']",
|
164 |
+
"151": "[4, 'am', 0, '']",
|
165 |
+
"152": "[3, 'diti', 0, '']",
|
166 |
+
"153": "[4, 'tak', 0, '']",
|
167 |
+
"154": "[2, 'ha', 0, '']",
|
168 |
+
"155": "[5, 'slati', 0, '']",
|
169 |
+
"156": "[3, 'lik', 0, '']",
|
170 |
+
"157": "[1, 'ok', 0, '']",
|
171 |
+
"158": "[4, 'i', 0, 'o']",
|
172 |
+
"159": "[2, 'go', 0, '']",
|
173 |
+
"160": "[2, 'te', 0, '']",
|
174 |
+
"161": "[1, 'ak', 0, '']",
|
175 |
+
"162": "[4, 'ziti', 0, '']",
|
176 |
+
"163": "[2, 'tac', 0, '']",
|
177 |
+
"164": "[4, 'an', 3, '']",
|
178 |
+
"165": "[2, 'tko', 0, '']",
|
179 |
+
"166": "[2, 'zak', 3, '']",
|
180 |
+
"167": "[3, 'sok', 0, '']",
|
181 |
+
"168": "[3, 'ao', 0, '']",
|
182 |
+
"169": "[3, 'g', 0, '']",
|
183 |
+
"170": "[2, 'sko', 0, '']",
|
184 |
+
"171": "[2, 'tati', 0, '']",
|
185 |
+
"172": "[4, '\u0107i', 0, '']",
|
186 |
+
"173": "[1, 'a\u010dka', 0, '']",
|
187 |
+
"174": "[2, 'eko', 3, '']",
|
188 |
+
"175": "[3, 'i', 0, 'o']",
|
189 |
+
"176": "[6, 'o', 0, '']",
|
190 |
+
"177": "[2, 'go', 3, '']",
|
191 |
+
"178": "[3, 'avati', 0, '']",
|
192 |
+
"179": "[2, '', 0, 'o']",
|
193 |
+
"180": "[0, '', 3, 'sl']",
|
194 |
+
"181": "[3, 'siti', 0, '']",
|
195 |
+
"182": "[1, 'agistar', 0, '']",
|
196 |
+
"183": "[6, 'iti', 0, '']",
|
197 |
+
"184": "[4, 'citi', 0, '']",
|
198 |
+
"185": "[3, 'vojeru\u010dno', 0, '']",
|
199 |
+
"186": "[4, 'gati', 0, '']",
|
200 |
+
"187": "[4, 'stiti', 0, '']",
|
201 |
+
"188": "[3, 'ti', 0, '']",
|
202 |
+
"189": "[1, 'odina', 0, '']",
|
203 |
+
"190": "[1, '\u0161i', 0, '']",
|
204 |
+
"191": "[1, 'u\u0107an', 0, '']",
|
205 |
+
"192": "[1, 'resuda', 0, '']",
|
206 |
+
"193": "[1, 'p\u0161tinski', 0, '']",
|
207 |
+
"194": "[1, 'ud', 0, '']",
|
208 |
+
"195": "[2, 'an', 5, 'prav']",
|
209 |
+
"196": "[1, 'okat', 0, '']",
|
210 |
+
"197": "[1, 'en', 0, '']",
|
211 |
+
"198": "[1, 'oletan', 0, '']",
|
212 |
+
"199": "[3, 'ta', 0, '']",
|
213 |
+
"200": "[3, 'tac', 0, '']",
|
214 |
+
"201": "[2, 'am', 0, '']",
|
215 |
+
"202": "[1, 'izija', 0, '']",
|
216 |
+
"203": "[4, 'lja', 0, '']",
|
217 |
+
"204": "[1, 'omo\u0107nik', 0, '']",
|
218 |
+
"205": "[2, 'j', 0, '']",
|
219 |
+
"206": "[7, 'r', 1, 'dob']",
|
220 |
+
"207": "[1, 'ka', 0, '']",
|
221 |
+
"208": "[3, 'i', 0, '']",
|
222 |
+
"209": "[5, 'ki', 0, '']",
|
223 |
+
"210": "[1, 'n', 1, '']",
|
224 |
+
"211": "[1, 'ar\u0161al', 0, '']",
|
225 |
+
"212": "[5, 'k', 1, 'uz']",
|
226 |
+
"213": "[1, 'ovodan', 0, '']",
|
227 |
+
"214": "[4, 'sti', 0, '']",
|
228 |
+
"215": "[5, 'ovati', 0, '']",
|
229 |
+
"216": "[3, 'ajati', 0, '']",
|
230 |
+
"217": "[3, 'zac', 0, '']",
|
231 |
+
"218": "[0, 'k', 0, '']",
|
232 |
+
"219": "[2, '', 5, 'isposa']",
|
233 |
+
"220": "[3, 'a', 0, '']",
|
234 |
+
"221": "[4, 'tati', 0, '']",
|
235 |
+
"222": "[1, 'ana', 0, '']",
|
236 |
+
"223": "[2, 'hteti', 0, '']",
|
237 |
+
"224": "[3, 'teti', 0, '']",
|
238 |
+
"225": "[1, 'esor', 0, '']",
|
239 |
+
"226": "[3, 'li', 3, '']",
|
240 |
+
"227": "[0, 'ar', 0, '']",
|
241 |
+
"228": "[2, 'oktor', 0, '']",
|
242 |
+
"229": "[5, '\u010dovek', 0, '']",
|
243 |
+
"230": "[3, 'ti', 1, 'b']",
|
244 |
+
"231": "[3, 'anj', 0, '']",
|
245 |
+
"232": "[4, 'siti', 0, '']",
|
246 |
+
"233": "[2, 'lik', 3, '']",
|
247 |
+
"234": "[1, 'uriranje', 0, '']",
|
248 |
+
"235": "[5, 'sti', 0, '']",
|
249 |
+
"236": "[3, 'dak', 0, '']",
|
250 |
+
"237": "[6, 'stiti', 0, '']",
|
251 |
+
"238": "[0, 'ugi', 0, '']",
|
252 |
+
"239": "[4, 'ar', 0, 'do']",
|
253 |
+
"240": "[4, 'ivati', 0, '']",
|
254 |
+
"241": "[2, 'ja', 0, '']",
|
255 |
+
"242": "[1, 'amostalno', 0, '']",
|
256 |
+
"243": "[1, 'reduze\u0107e', 0, '']",
|
257 |
+
"244": "[2, 'zo', 0, '']",
|
258 |
+
"245": "[2, 'n', 0, '']",
|
259 |
+
"246": "[5, 'an', 0, '']",
|
260 |
+
"247": "[1, 'in', 0, '']",
|
261 |
+
"248": "[3, 'at', 0, '']",
|
262 |
+
"249": "[3, 'slan', 0, '']",
|
263 |
+
"250": "[6, 'ar', 0, '']",
|
264 |
+
"251": "[5, 'ent', 0, '']",
|
265 |
+
"252": "[3, 'ti', 1, 'sl']",
|
266 |
+
"253": "[3, 'tak', 0, '']",
|
267 |
+
"254": "[1, 'osno', 0, '']",
|
268 |
+
"255": "[4, 'cati', 0, '']",
|
269 |
+
"256": "[3, 'mo', 0, '']",
|
270 |
+
"257": "[1, 'efon', 0, '']",
|
271 |
+
"258": "[1, 'ugo', 0, '']",
|
272 |
+
"259": "[1, 'omiran', 0, '']",
|
273 |
+
"260": "[4, 'zak', 0, '']",
|
274 |
+
"261": "[1, 'ina', 0, '']",
|
275 |
+
"262": "[3, 'vati', 0, '']",
|
276 |
+
"263": "[1, 'oktor', 0, '']",
|
277 |
+
"264": "[0, 'an', 0, '']",
|
278 |
+
"265": "[0, 'ica', 0, '']",
|
279 |
+
"266": "[3, '\u0161ta', 0, '']",
|
280 |
+
"267": "[4, 'ti', 2, '']",
|
281 |
+
"268": "[2, 'ti', 4, 'inten']",
|
282 |
+
"269": "[0, '.', 0, '']",
|
283 |
+
"270": "[2, 'sok', 3, '']",
|
284 |
+
"271": "[0, 'botica', 0, '']",
|
285 |
+
"272": "[1, 'ektri\u010dan', 0, '']",
|
286 |
+
"273": "[1, 'i', 2, '']",
|
287 |
+
"274": "[0, 'ti', 1, 'b']",
|
288 |
+
"275": "[1, 'ena', 0, '']",
|
289 |
+
"276": "[2, '', 5, 'vodod']",
|
290 |
+
"277": "[5, 'zniti', 0, '']",
|
291 |
+
"278": "[0, 'n', 0, '']",
|
292 |
+
"279": "[2, 'g', 3, '']",
|
293 |
+
"280": "[4, 'len', 0, '']",
|
294 |
+
"281": "[1, 'omo\u0107je', 0, '']",
|
295 |
+
"282": "[2, 'ati', 0, '']",
|
296 |
+
"283": "[1, 'ospodin', 0, '']",
|
297 |
+
"284": "[2, '', 11, 'prvotu\u017een']",
|
298 |
+
"285": "[4, 'dak', 0, '']",
|
299 |
+
"286": "[3, 'len', 0, '']",
|
300 |
+
"287": "[2, 'ok', 0, '']"
|
301 |
+
},
|
302 |
+
"initializer_range": 0.02,
|
303 |
+
"intermediate_size": 3072,
|
304 |
+
"label2id": {
|
305 |
+
"[0, '', 0, '']": 0,
|
306 |
+
"[0, '', 3, 'sl']": 180,
|
307 |
+
"[0, '', 3, '\u0161t']": 101,
|
308 |
+
"[0, '.', 0, '']": 269,
|
309 |
+
"[0, 'a', 0, '']": 68,
|
310 |
+
"[0, 'an', 0, '']": 264,
|
311 |
+
"[0, 'ar', 0, '']": 227,
|
312 |
+
"[0, 'be', 0, '']": 9,
|
313 |
+
"[0, 'botica', 0, '']": 271,
|
314 |
+
"[0, 'i', 0, '']": 114,
|
315 |
+
"[0, 'ica', 0, '']": 265,
|
316 |
+
"[0, 'j', 0, '']": 48,
|
317 |
+
"[0, 'k', 0, '']": 218,
|
318 |
+
"[0, 'n', 0, '']": 278,
|
319 |
+
"[0, 'ti', 0, '']": 8,
|
320 |
+
"[0, 'ti', 1, 'b']": 274,
|
321 |
+
"[0, 'ti', 1, 'ht']": 36,
|
322 |
+
"[0, 'ugi', 0, '']": 238,
|
323 |
+
"[1, '', 0, '']": 3,
|
324 |
+
"[1, '', 0, 'on']": 57,
|
325 |
+
"[1, 'a', 0, '']": 2,
|
326 |
+
"[1, 'agistar', 0, '']": 182,
|
327 |
+
"[1, 'aj', 0, '']": 52,
|
328 |
+
"[1, 'ak', 0, '']": 161,
|
329 |
+
"[1, 'amostalno', 0, '']": 242,
|
330 |
+
"[1, 'an', 0, '']": 105,
|
331 |
+
"[1, 'ana', 0, '']": 222,
|
332 |
+
"[1, 'arta', 0, '']": 73,
|
333 |
+
"[1, 'ar\u0161al', 0, '']": 211,
|
334 |
+
"[1, 'astarski', 0, '']": 76,
|
335 |
+
"[1, 'ati', 0, '']": 83,
|
336 |
+
"[1, 'av', 0, '']": 150,
|
337 |
+
"[1, 'a\u010dka', 0, '']": 173,
|
338 |
+
"[1, 'e', 0, '']": 16,
|
339 |
+
"[1, 'efon', 0, '']": 257,
|
340 |
+
"[1, 'ektri\u010dan', 0, '']": 272,
|
341 |
+
"[1, 'en', 0, '']": 197,
|
342 |
+
"[1, 'ena', 0, '']": 275,
|
343 |
+
"[1, 'epublika', 0, '']": 108,
|
344 |
+
"[1, 'esor', 0, '']": 225,
|
345 |
+
"[1, 'eti', 0, '']": 20,
|
346 |
+
"[1, 'evar', 0, '']": 1,
|
347 |
+
"[1, 'i', 0, '']": 12,
|
348 |
+
"[1, 'i', 2, '']": 273,
|
349 |
+
"[1, 'i', 3, 'bi']": 135,
|
350 |
+
"[1, 'ica', 0, '']": 71,
|
351 |
+
"[1, 'in', 0, '']": 247,
|
352 |
+
"[1, 'ina', 0, '']": 261,
|
353 |
+
"[1, 'istarski', 0, '']": 99,
|
354 |
+
"[1, 'iti', 0, '']": 23,
|
355 |
+
"[1, 'izija', 0, '']": 202,
|
356 |
+
"[1, 'i\u010dan', 0, '']": 72,
|
357 |
+
"[1, 'i\u010dno', 0, '']": 144,
|
358 |
+
"[1, 'j', 0, '']": 61,
|
359 |
+
"[1, 'ji', 0, '']": 29,
|
360 |
+
"[1, 'ka', 0, '']": 207,
|
361 |
+
"[1, 'n', 1, '']": 210,
|
362 |
+
"[1, 'o', 0, '']": 24,
|
363 |
+
"[1, 'odina', 0, '']": 189,
|
364 |
+
"[1, 'oj', 0, '']": 4,
|
365 |
+
"[1, 'ojan', 0, '']": 77,
|
366 |
+
"[1, 'ok', 0, '']": 157,
|
367 |
+
"[1, 'okat', 0, '']": 196,
|
368 |
+
"[1, 'oktor', 0, '']": 263,
|
369 |
+
"[1, 'oletan', 0, '']": 198,
|
370 |
+
"[1, 'omiran', 0, '']": 259,
|
371 |
+
"[1, 'omo\u0107je', 0, '']": 281,
|
372 |
+
"[1, 'omo\u0107nik', 0, '']": 204,
|
373 |
+
"[1, 'osno', 0, '']": 254,
|
374 |
+
"[1, 'ospodin', 0, '']": 283,
|
375 |
+
"[1, 'ovodan', 0, '']": 213,
|
376 |
+
"[1, 'p\u0161tinski', 0, '']": 193,
|
377 |
+
"[1, 'reduze\u0107e', 0, '']": 243,
|
378 |
+
"[1, 'resuda', 0, '']": 192,
|
379 |
+
"[1, 'rokuplje', 0, '']": 107,
|
380 |
+
"[1, 'r\u0161ilac', 0, '']": 148,
|
381 |
+
"[1, 'sti', 0, '']": 41,
|
382 |
+
"[1, 'ti', 0, '']": 58,
|
383 |
+
"[1, 'ud', 0, '']": 194,
|
384 |
+
"[1, 'ugi', 0, '']": 124,
|
385 |
+
"[1, 'ugo', 0, '']": 258,
|
386 |
+
"[1, 'uriranje', 0, '']": 234,
|
387 |
+
"[1, 'uti', 0, '']": 142,
|
388 |
+
"[1, 'u\u0107an', 0, '']": 191,
|
389 |
+
"[1, 'u\u017eben', 0, '']": 123,
|
390 |
+
"[1, 'u\u017enost', 0, '']": 149,
|
391 |
+
"[1, '\u0161i', 0, '']": 190,
|
392 |
+
"[2, '', 0, '']": 5,
|
393 |
+
"[2, '', 0, 'o']": 179,
|
394 |
+
"[2, '', 11, 'prvotu\u017een']": 284,
|
395 |
+
"[2, '', 5, 'isposa']": 219,
|
396 |
+
"[2, '', 5, 'vodod']": 276,
|
397 |
+
"[2, '.o.o.', 0, '']": 93,
|
398 |
+
"[2, 'a', 0, '']": 43,
|
399 |
+
"[2, 'ac', 0, '']": 70,
|
400 |
+
"[2, 'aj', 0, '']": 15,
|
401 |
+
"[2, 'ak', 0, '']": 31,
|
402 |
+
"[2, 'am', 0, '']": 201,
|
403 |
+
"[2, 'an', 0, '']": 10,
|
404 |
+
"[2, 'an', 5, 'prav']": 195,
|
405 |
+
"[2, 'ao', 0, '']": 86,
|
406 |
+
"[2, 'ar', 0, '']": 18,
|
407 |
+
"[2, 'at', 0, '']": 69,
|
408 |
+
"[2, 'ati', 0, '']": 282,
|
409 |
+
"[2, 'av', 0, '']": 25,
|
410 |
+
"[2, 'biti', 0, '']": 13,
|
411 |
+
"[2, 'cati', 0, '']": 82,
|
412 |
+
"[2, 'ek', 0, '']": 139,
|
413 |
+
"[2, 'eko', 0, '']": 110,
|
414 |
+
"[2, 'eko', 3, '']": 174,
|
415 |
+
"[2, 'eti', 0, '']": 53,
|
416 |
+
"[2, 'g', 0, '']": 97,
|
417 |
+
"[2, 'g', 3, '']": 279,
|
418 |
+
"[2, 'ga', 0, '']": 112,
|
419 |
+
"[2, 'gati', 0, '']": 134,
|
420 |
+
"[2, 'go', 0, '']": 159,
|
421 |
+
"[2, 'go', 3, '']": 177,
|
422 |
+
"[2, 'ha', 0, '']": 154,
|
423 |
+
"[2, 'hteti', 0, '']": 223,
|
424 |
+
"[2, 'i', 0, '']": 39,
|
425 |
+
"[2, 'iti', 0, '']": 14,
|
426 |
+
"[2, 'j', 0, '']": 205,
|
427 |
+
"[2, 'ja', 0, '']": 241,
|
428 |
+
"[2, 'ji', 0, '']": 19,
|
429 |
+
"[2, 'k', 0, '']": 74,
|
430 |
+
"[2, 'ka', 0, '']": 62,
|
431 |
+
"[2, 'lik', 0, '']": 106,
|
432 |
+
"[2, 'lik', 3, '']": 233,
|
433 |
+
"[2, 'n', 0, '']": 245,
|
434 |
+
"[2, 'nuti', 0, '']": 140,
|
435 |
+
"[2, 'o', 0, '']": 75,
|
436 |
+
"[2, 'ok', 0, '']": 287,
|
437 |
+
"[2, 'oktor', 0, '']": 228,
|
438 |
+
"[2, 'on', 0, '']": 85,
|
439 |
+
"[2, 'sati', 0, '']": 127,
|
440 |
+
"[2, 'sko', 0, '']": 170,
|
441 |
+
"[2, 'sok', 0, '']": 92,
|
442 |
+
"[2, 'sok', 3, '']": 270,
|
443 |
+
"[2, 'sti', 0, '']": 103,
|
444 |
+
"[2, 'tac', 0, '']": 163,
|
445 |
+
"[2, 'tak', 0, '']": 84,
|
446 |
+
"[2, 'tan', 0, '']": 133,
|
447 |
+
"[2, 'tati', 0, '']": 171,
|
448 |
+
"[2, 'te', 0, '']": 160,
|
449 |
+
"[2, 'ti', 0, '']": 21,
|
450 |
+
"[2, 'ti', 1, 'b']": 59,
|
451 |
+
"[2, 'ti', 1, 'ht']": 37,
|
452 |
+
"[2, 'ti', 4, 'inten']": 268,
|
453 |
+
"[2, 'tko', 0, '']": 165,
|
454 |
+
"[2, 'vati', 0, '']": 65,
|
455 |
+
"[2, 'zak', 0, '']": 147,
|
456 |
+
"[2, 'zak', 3, '']": 166,
|
457 |
+
"[2, 'zati', 0, '']": 55,
|
458 |
+
"[2, 'zo', 0, '']": 244,
|
459 |
+
"[2, 'zu', 0, '']": 116,
|
460 |
+
"[2, '\u0107i', 0, '']": 45,
|
461 |
+
"[3, '', 0, '']": 30,
|
462 |
+
"[3, '', 0, 'o']": 51,
|
463 |
+
"[3, 'a', 0, '']": 220,
|
464 |
+
"[3, 'ac', 0, '']": 88,
|
465 |
+
"[3, 'aj', 0, '']": 35,
|
466 |
+
"[3, 'ajati', 0, '']": 216,
|
467 |
+
"[3, 'ak', 0, '']": 38,
|
468 |
+
"[3, 'an', 0, '']": 17,
|
469 |
+
"[3, 'anj', 0, '']": 231,
|
470 |
+
"[3, 'ao', 0, '']": 168,
|
471 |
+
"[3, 'ar', 0, '']": 89,
|
472 |
+
"[3, 'at', 0, '']": 248,
|
473 |
+
"[3, 'ati', 0, '']": 95,
|
474 |
+
"[3, 'av', 0, '']": 46,
|
475 |
+
"[3, 'avati', 0, '']": 178,
|
476 |
+
"[3, 'ba', 0, '']": 122,
|
477 |
+
"[3, 'biti', 0, '']": 96,
|
478 |
+
"[3, 'dak', 0, '']": 236,
|
479 |
+
"[3, 'deti', 0, '']": 131,
|
480 |
+
"[3, 'diti', 0, '']": 152,
|
481 |
+
"[3, 'e', 0, '']": 49,
|
482 |
+
"[3, 'ek', 0, '']": 67,
|
483 |
+
"[3, 'g', 0, '']": 169,
|
484 |
+
"[3, 'i', 0, '']": 208,
|
485 |
+
"[3, 'i', 0, 'o']": 175,
|
486 |
+
"[3, 'iti', 0, '']": 26,
|
487 |
+
"[3, 'ivati', 0, '']": 7,
|
488 |
+
"[3, 'ka', 0, '']": 42,
|
489 |
+
"[3, 'lac', 0, '']": 22,
|
490 |
+
"[3, 'len', 0, '']": 286,
|
491 |
+
"[3, 'li', 0, '']": 143,
|
492 |
+
"[3, 'li', 3, '']": 226,
|
493 |
+
"[3, 'lik', 0, '']": 156,
|
494 |
+
"[3, 'lik', 3, '']": 54,
|
495 |
+
"[3, 'lo', 0, '']": 94,
|
496 |
+
"[3, 'lo', 3, '']": 121,
|
497 |
+
"[3, 'mo', 0, '']": 256,
|
498 |
+
"[3, 'o', 0, '']": 34,
|
499 |
+
"[3, 'o', 3, '']": 80,
|
500 |
+
"[3, 'ovati', 0, '']": 40,
|
501 |
+
"[3, 'siti', 0, '']": 181,
|
502 |
+
"[3, 'slan', 0, '']": 249,
|
503 |
+
"[3, 'sok', 0, '']": 167,
|
504 |
+
"[3, 'st', 0, '']": 66,
|
505 |
+
"[3, 'sti', 0, '']": 109,
|
506 |
+
"[3, 't', 0, '']": 91,
|
507 |
+
"[3, 'ta', 0, '']": 199,
|
508 |
+
"[3, 'tac', 0, '']": 200,
|
509 |
+
"[3, 'tak', 0, '']": 253,
|
510 |
+
"[3, 'tak', 3, '']": 119,
|
511 |
+
"[3, 'tan', 0, '']": 100,
|
512 |
+
"[3, 'teti', 0, '']": 224,
|
513 |
+
"[3, 'ti', 0, '']": 188,
|
514 |
+
"[3, 'ti', 1, 'b']": 230,
|
515 |
+
"[3, 'ti', 1, 'sl']": 252,
|
516 |
+
"[3, 'titi', 0, '']": 126,
|
517 |
+
"[3, 'vati', 0, '']": 262,
|
518 |
+
"[3, 'vo', 0, '']": 81,
|
519 |
+
"[3, 'vojeru\u010dno', 0, '']": 185,
|
520 |
+
"[3, 'zac', 0, '']": 217,
|
521 |
+
"[3, 'zak', 0, '']": 117,
|
522 |
+
"[3, '\u0107i', 0, '']": 87,
|
523 |
+
"[3, '\u0161ta', 0, '']": 266,
|
524 |
+
"[4, '', 0, '']": 138,
|
525 |
+
"[4, '', 0, 'o']": 90,
|
526 |
+
"[4, 'ac', 0, '']": 79,
|
527 |
+
"[4, 'ak', 0, '']": 64,
|
528 |
+
"[4, 'am', 0, '']": 151,
|
529 |
+
"[4, 'an', 0, '']": 132,
|
530 |
+
"[4, 'an', 3, '']": 164,
|
531 |
+
"[4, 'ao', 0, '']": 11,
|
532 |
+
"[4, 'ar', 0, 'do']": 239,
|
533 |
+
"[4, 'at', 0, '']": 125,
|
534 |
+
"[4, 'av', 0, '']": 115,
|
535 |
+
"[4, 'biti', 0, '']": 56,
|
536 |
+
"[4, 'cati', 0, '']": 255,
|
537 |
+
"[4, 'citi', 0, '']": 184,
|
538 |
+
"[4, 'dak', 0, '']": 285,
|
539 |
+
"[4, 'deti', 0, '']": 128,
|
540 |
+
"[4, 'diti', 0, '']": 111,
|
541 |
+
"[4, 'g', 0, '']": 137,
|
542 |
+
"[4, 'gati', 0, '']": 186,
|
543 |
+
"[4, 'i', 0, 'o']": 158,
|
544 |
+
"[4, 'iti', 0, '']": 28,
|
545 |
+
"[4, 'ivati', 0, '']": 240,
|
546 |
+
"[4, 'k', 0, '']": 47,
|
547 |
+
"[4, 'lac', 0, '']": 113,
|
548 |
+
"[4, 'len', 0, '']": 280,
|
549 |
+
"[4, 'li', 0, '']": 44,
|
550 |
+
"[4, 'li', 3, '']": 50,
|
551 |
+
"[4, 'lja', 0, '']": 203,
|
552 |
+
"[4, 'mnogo', 0, '']": 136,
|
553 |
+
"[4, 'o', 0, '']": 78,
|
554 |
+
"[4, 'siti', 0, '']": 232,
|
555 |
+
"[4, 'sti', 0, '']": 214,
|
556 |
+
"[4, 'stiti', 0, '']": 187,
|
557 |
+
"[4, 'tak', 0, '']": 153,
|
558 |
+
"[4, 'tati', 0, '']": 221,
|
559 |
+
"[4, 'ti', 0, '']": 27,
|
560 |
+
"[4, 'ti', 2, '']": 267,
|
561 |
+
"[4, 'titi', 0, '']": 118,
|
562 |
+
"[4, 'vati', 0, '']": 120,
|
563 |
+
"[4, 'zak', 0, '']": 260,
|
564 |
+
"[4, 'ziti', 0, '']": 162,
|
565 |
+
"[4, '\u0107i', 0, '']": 172,
|
566 |
+
"[4, '\u0161ta', 0, '']": 102,
|
567 |
+
"[4, '\u0161to', 0, '']": 146,
|
568 |
+
"[5, '', 0, '']": 33,
|
569 |
+
"[5, 'an', 0, '']": 246,
|
570 |
+
"[5, 'an', 3, '']": 98,
|
571 |
+
"[5, 'ent', 0, '']": 251,
|
572 |
+
"[5, 'iti', 0, '']": 60,
|
573 |
+
"[5, 'ivati', 0, '']": 63,
|
574 |
+
"[5, 'k', 1, 'uz']": 212,
|
575 |
+
"[5, 'ki', 0, '']": 209,
|
576 |
+
"[5, 'lac', 0, '']": 141,
|
577 |
+
"[5, 'ovati', 0, '']": 215,
|
578 |
+
"[5, 'slati', 0, '']": 155,
|
579 |
+
"[5, 'sti', 0, '']": 235,
|
580 |
+
"[5, 'stiti', 0, '']": 129,
|
581 |
+
"[5, 'zniti', 0, '']": 277,
|
582 |
+
"[5, '\u0107i', 0, '']": 104,
|
583 |
+
"[5, '\u010dovek', 0, '']": 229,
|
584 |
+
"[6, 'ak', 0, '']": 32,
|
585 |
+
"[6, 'ao', 0, '']": 6,
|
586 |
+
"[6, 'ar', 0, '']": 250,
|
587 |
+
"[6, 'iti', 0, '']": 183,
|
588 |
+
"[6, 'o', 0, '']": 176,
|
589 |
+
"[6, 'ogo', 0, 'm']": 130,
|
590 |
+
"[6, 'r', 1, 'dob']": 145,
|
591 |
+
"[6, 'stiti', 0, '']": 237,
|
592 |
+
"[7, 'r', 1, 'dob']": 206
|
593 |
+
},
|
594 |
+
"layer_norm_eps": 1e-12,
|
595 |
+
"max_position_embeddings": 512,
|
596 |
+
"model_type": "electra",
|
597 |
+
"num_attention_heads": 12,
|
598 |
+
"num_hidden_layers": 12,
|
599 |
+
"pad_token_id": 0,
|
600 |
+
"position_embedding_type": "absolute",
|
601 |
+
"summary_activation": "gelu",
|
602 |
+
"summary_last_dropout": 0.1,
|
603 |
+
"summary_type": "first",
|
604 |
+
"summary_use_proj": true,
|
605 |
+
"torch_dtype": "float32",
|
606 |
+
"transformers_version": "4.40.2",
|
607 |
+
"type_vocab_size": 2,
|
608 |
+
"use_cache": true,
|
609 |
+
"vocab_size": 32000
|
610 |
+
}
|
model_args.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"adafactor_beta1": null, "adafactor_clip_threshold": 1.0, "adafactor_decay_rate": -0.8, "adafactor_eps": [1e-30, 0.001], "adafactor_relative_step": true, "adafactor_scale_parameter": true, "adafactor_warmup_init": true, "adam_betas": [0.9, 0.999], "adam_epsilon": 1e-08, "best_model_dir": "outputs/best_model", "cache_dir": "cache_dir/", "config": {}, "cosine_schedule_num_cycles": 0.5, "custom_layer_parameters": [], "custom_parameter_groups": [], "dataloader_num_workers": 0, "do_lower_case": false, "dynamic_quantize": false, "early_stopping_consider_epochs": false, "early_stopping_delta": 0, "early_stopping_metric": "eval_loss", "early_stopping_metric_minimize": true, "early_stopping_patience": 3, "encoding": null, "eval_batch_size": 100, "evaluate_during_training": false, "evaluate_during_training_silent": true, "evaluate_during_training_steps": 2000, "evaluate_during_training_verbose": false, "evaluate_each_epoch": true, "fp16": true, "gradient_accumulation_steps": 1, "learning_rate": 4e-05, "local_rank": -1, "logging_steps": 50, "loss_type": null, "loss_args": {}, "manual_seed": 64, "max_grad_norm": 1.0, "max_seq_length": 512, "model_name": "classla/bcms-bertic", "model_type": "electra", "multiprocessing_chunksize": -1, "n_gpu": 1, "no_cache": true, "no_save": false, "not_saved_args": [], "num_train_epochs": 20, "optimizer": "AdamW", "output_dir": "outputs/", "overwrite_output_dir": true, "polynomial_decay_schedule_lr_end": 1e-07, "polynomial_decay_schedule_power": 1.0, "process_count": 14, "quantized_model": false, "reprocess_input_data": true, "save_best_model": true, "save_eval_checkpoints": false, "save_model_every_epoch": false, "save_optimizer_and_scheduler": true, "save_steps": 2000, "scheduler": "linear_schedule_with_warmup", "silent": true, "skip_special_tokens": true, "tensorboard_dir": null, "thread_count": null, "tokenizer_name": null, "tokenizer_type": null, "train_batch_size": 8, "train_custom_parameters_only": false, "trust_remote_code": false, "use_cached_eval_features": false, "use_early_stopping": false, "use_hf_datasets": false, "use_multiprocessing": true, "use_multiprocessing_for_evaluation": true, "wandb_kwargs": {}, "wandb_project": null, "warmup_ratio": 0.06, "warmup_steps": 716, "weight_decay": 0.0, "model_class": "NERModel", "classification_report": false, "labels_list": ["[0, '', 0, '']", "[1, 'evar', 0, '']", "[1, 'a', 0, '']", "[1, '', 0, '']", "[1, 'oj', 0, '']", "[2, '', 0, '']", "[6, 'ao', 0, '']", "[3, 'ivati', 0, '']", "[0, 'ti', 0, '']", "[0, 'be', 0, '']", "[2, 'an', 0, '']", "[4, 'ao', 0, '']", "[1, 'i', 0, '']", "[2, 'biti', 0, '']", "[2, 'iti', 0, '']", "[2, 'aj', 0, '']", "[1, 'e', 0, '']", "[3, 'an', 0, '']", "[2, 'ar', 0, '']", "[2, 'ji', 0, '']", "[1, 'eti', 0, '']", "[2, 'ti', 0, '']", "[3, 'lac', 0, '']", "[1, 'iti', 0, '']", "[1, 'o', 0, '']", "[2, 'av', 0, '']", "[3, 'iti', 0, '']", "[4, 'ti', 0, '']", "[4, 'iti', 0, '']", "[1, 'ji', 0, '']", "[3, '', 0, '']", "[2, 'ak', 0, '']", "[6, 'ak', 0, '']", "[5, '', 0, '']", "[3, 'o', 0, '']", "[3, 'aj', 0, '']", "[0, 'ti', 1, 'ht']", "[2, 'ti', 1, 'ht']", "[3, 'ak', 0, '']", "[2, 'i', 0, '']", "[3, 'ovati', 0, '']", "[1, 'sti', 0, '']", "[3, 'ka', 0, '']", "[2, 'a', 0, '']", "[4, 'li', 0, '']", "[2, '\u0107i', 0, '']", "[3, 'av', 0, '']", "[4, 'k', 0, '']", "[0, 'j', 0, '']", "[3, 'e', 0, '']", "[4, 'li', 3, '']", "[3, '', 0, 'o']", "[1, 'aj', 0, '']", "[2, 'eti', 0, '']", "[3, 'lik', 3, '']", "[2, 'zati', 0, '']", "[4, 'biti', 0, '']", "[1, '', 0, 'on']", "[1, 'ti', 0, '']", "[2, 'ti', 1, 'b']", "[5, 'iti', 0, '']", "[1, 'j', 0, '']", "[2, 'ka', 0, '']", "[5, 'ivati', 0, '']", "[4, 'ak', 0, '']", "[2, 'vati', 0, '']", "[3, 'st', 0, '']", "[3, 'ek', 0, '']", "[0, 'a', 0, '']", "[2, 'at', 0, '']", "[2, 'ac', 0, '']", "[1, 'ica', 0, '']", "[1, 'i\u010dan', 0, '']", "[1, 'arta', 0, '']", "[2, 'k', 0, '']", "[2, 'o', 0, '']", "[1, 'astarski', 0, '']", "[1, 'ojan', 0, '']", "[4, 'o', 0, '']", "[4, 'ac', 0, '']", "[3, 'o', 3, '']", "[3, 'vo', 0, '']", "[2, 'cati', 0, '']", "[1, 'ati', 0, '']", "[2, 'tak', 0, '']", "[2, 'on', 0, '']", "[2, 'ao', 0, '']", "[3, '\u0107i', 0, '']", "[3, 'ac', 0, '']", "[3, 'ar', 0, '']", "[4, '', 0, 'o']", "[3, 't', 0, '']", "[2, 'sok', 0, '']", "[2, '.o.o.', 0, '']", "[3, 'lo', 0, '']", "[3, 'ati', 0, '']", "[3, 'biti', 0, '']", "[2, 'g', 0, '']", "[5, 'an', 3, '']", "[1, 'istarski', 0, '']", "[3, 'tan', 0, '']", "[0, '', 3, '\u0161t']", "[4, '\u0161ta', 0, '']", "[2, 'sti', 0, '']", "[5, '\u0107i', 0, '']", "[1, 'an', 0, '']", "[2, 'lik', 0, '']", "[1, 'rokuplje', 0, '']", "[1, 'epublika', 0, '']", "[3, 'sti', 0, '']", "[2, 'eko', 0, '']", "[4, 'diti', 0, '']", "[2, 'ga', 0, '']", "[4, 'lac', 0, '']", "[0, 'i', 0, '']", "[4, 'av', 0, '']", "[2, 'zu', 0, '']", "[3, 'zak', 0, '']", "[4, 'titi', 0, '']", "[3, 'tak', 3, '']", "[4, 'vati', 0, '']", "[3, 'lo', 3, '']", "[3, 'ba', 0, '']", "[1, 'u\u017eben', 0, '']", "[1, 'ugi', 0, '']", "[4, 'at', 0, '']", "[3, 'titi', 0, '']", "[2, 'sati', 0, '']", "[4, 'deti', 0, '']", "[5, 'stiti', 0, '']", "[6, 'ogo', 0, 'm']", "[3, 'deti', 0, '']", "[4, 'an', 0, '']", "[2, 'tan', 0, '']", "[2, 'gati', 0, '']", "[1, 'i', 3, 'bi']", "[4, 'mnogo', 0, '']", "[4, 'g', 0, '']", "[4, '', 0, '']", "[2, 'ek', 0, '']", "[2, 'nuti', 0, '']", "[5, 'lac', 0, '']", "[1, 'uti', 0, '']", "[3, 'li', 0, '']", "[1, 'i\u010dno', 0, '']", "[6, 'r', 1, 'dob']", "[4, '\u0161to', 0, '']", "[2, 'zak', 0, '']", "[1, 'r\u0161ilac', 0, '']", "[1, 'u\u017enost', 0, '']", "[1, 'av', 0, '']", "[4, 'am', 0, '']", "[3, 'diti', 0, '']", "[4, 'tak', 0, '']", "[2, 'ha', 0, '']", "[5, 'slati', 0, '']", "[3, 'lik', 0, '']", "[1, 'ok', 0, '']", "[4, 'i', 0, 'o']", "[2, 'go', 0, '']", "[2, 'te', 0, '']", "[1, 'ak', 0, '']", "[4, 'ziti', 0, '']", "[2, 'tac', 0, '']", "[4, 'an', 3, '']", "[2, 'tko', 0, '']", "[2, 'zak', 3, '']", "[3, 'sok', 0, '']", "[3, 'ao', 0, '']", "[3, 'g', 0, '']", "[2, 'sko', 0, '']", "[2, 'tati', 0, '']", "[4, '\u0107i', 0, '']", "[1, 'a\u010dka', 0, '']", "[2, 'eko', 3, '']", "[3, 'i', 0, 'o']", "[6, 'o', 0, '']", "[2, 'go', 3, '']", "[3, 'avati', 0, '']", "[2, '', 0, 'o']", "[0, '', 3, 'sl']", "[3, 'siti', 0, '']", "[1, 'agistar', 0, '']", "[6, 'iti', 0, '']", "[4, 'citi', 0, '']", "[3, 'vojeru\u010dno', 0, '']", "[4, 'gati', 0, '']", "[4, 'stiti', 0, '']", "[3, 'ti', 0, '']", "[1, 'odina', 0, '']", "[1, '\u0161i', 0, '']", "[1, 'u\u0107an', 0, '']", "[1, 'resuda', 0, '']", "[1, 'p\u0161tinski', 0, '']", "[1, 'ud', 0, '']", "[2, 'an', 5, 'prav']", "[1, 'okat', 0, '']", "[1, 'en', 0, '']", "[1, 'oletan', 0, '']", "[3, 'ta', 0, '']", "[3, 'tac', 0, '']", "[2, 'am', 0, '']", "[1, 'izija', 0, '']", "[4, 'lja', 0, '']", "[1, 'omo\u0107nik', 0, '']", "[2, 'j', 0, '']", "[7, 'r', 1, 'dob']", "[1, 'ka', 0, '']", "[3, 'i', 0, '']", "[5, 'ki', 0, '']", "[1, 'n', 1, '']", "[1, 'ar\u0161al', 0, '']", "[5, 'k', 1, 'uz']", "[1, 'ovodan', 0, '']", "[4, 'sti', 0, '']", "[5, 'ovati', 0, '']", "[3, 'ajati', 0, '']", "[3, 'zac', 0, '']", "[0, 'k', 0, '']", "[2, '', 5, 'isposa']", "[3, 'a', 0, '']", "[4, 'tati', 0, '']", "[1, 'ana', 0, '']", "[2, 'hteti', 0, '']", "[3, 'teti', 0, '']", "[1, 'esor', 0, '']", "[3, 'li', 3, '']", "[0, 'ar', 0, '']", "[2, 'oktor', 0, '']", "[5, '\u010dovek', 0, '']", "[3, 'ti', 1, 'b']", "[3, 'anj', 0, '']", "[4, 'siti', 0, '']", "[2, 'lik', 3, '']", "[1, 'uriranje', 0, '']", "[5, 'sti', 0, '']", "[3, 'dak', 0, '']", "[6, 'stiti', 0, '']", "[0, 'ugi', 0, '']", "[4, 'ar', 0, 'do']", "[4, 'ivati', 0, '']", "[2, 'ja', 0, '']", "[1, 'amostalno', 0, '']", "[1, 'reduze\u0107e', 0, '']", "[2, 'zo', 0, '']", "[2, 'n', 0, '']", "[5, 'an', 0, '']", "[1, 'in', 0, '']", "[3, 'at', 0, '']", "[3, 'slan', 0, '']", "[6, 'ar', 0, '']", "[5, 'ent', 0, '']", "[3, 'ti', 1, 'sl']", "[3, 'tak', 0, '']", "[1, 'osno', 0, '']", "[4, 'cati', 0, '']", "[3, 'mo', 0, '']", "[1, 'efon', 0, '']", "[1, 'ugo', 0, '']", "[1, 'omiran', 0, '']", "[4, 'zak', 0, '']", "[1, 'ina', 0, '']", "[3, 'vati', 0, '']", "[1, 'oktor', 0, '']", "[0, 'an', 0, '']", "[0, 'ica', 0, '']", "[3, '\u0161ta', 0, '']", "[4, 'ti', 2, '']", "[2, 'ti', 4, 'inten']", "[0, '.', 0, '']", "[2, 'sok', 3, '']", "[0, 'botica', 0, '']", "[1, 'ektri\u010dan', 0, '']", "[1, 'i', 2, '']", "[0, 'ti', 1, 'b']", "[1, 'ena', 0, '']", "[2, '', 5, 'vodod']", "[5, 'zniti', 0, '']", "[0, 'n', 0, '']", "[2, 'g', 3, '']", "[4, 'len', 0, '']", "[1, 'omo\u0107je', 0, '']", "[2, 'ati', 0, '']", "[1, 'ospodin', 0, '']", "[2, '', 11, 'prvotu\u017een']", "[4, 'dak', 0, '']", "[3, 'len', 0, '']", "[2, 'ok', 0, '']"], "lazy_loading": false, "lazy_loading_start_line": 0, "onnx": false, "special_tokens_list": []}
|
special_tokens_map.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cls_token": "[CLS]",
|
3 |
+
"mask_token": "[MASK]",
|
4 |
+
"pad_token": "[PAD]",
|
5 |
+
"sep_token": "[SEP]",
|
6 |
+
"unk_token": "[UNK]"
|
7 |
+
}
|
vocab.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|