Upload tokenizer
Browse files- tokenizer.json +126 -44
tokenizer.json
CHANGED
@@ -145,53 +145,135 @@
|
|
145 |
"[CLS]": 5,
|
146 |
"[SEP]": 6,
|
147 |
"C": 7,
|
148 |
-
"
|
149 |
-
"
|
150 |
-
"
|
151 |
-
"
|
152 |
"1": 12,
|
153 |
-
"
|
154 |
"=": 14,
|
155 |
-
"
|
156 |
-
"
|
157 |
-
"
|
158 |
-
"[C
|
159 |
-
"
|
160 |
-
"
|
161 |
-
"
|
162 |
-
"
|
163 |
-
"
|
164 |
-
"
|
165 |
-
"
|
166 |
-
"
|
167 |
-
"[
|
168 |
-
"
|
169 |
-
"
|
170 |
-
"
|
171 |
-
"
|
172 |
-
"
|
173 |
-
"[
|
174 |
-
"
|
175 |
-
"
|
176 |
-
"
|
177 |
-
"[
|
178 |
-
"[
|
179 |
-
"
|
180 |
-
"
|
181 |
-
"
|
182 |
-
"
|
183 |
-
"
|
184 |
-
"
|
185 |
-
"
|
186 |
"[Si]": 46,
|
187 |
-
"
|
188 |
-
"[
|
189 |
-
"[
|
190 |
-
"[
|
191 |
-
"[
|
192 |
-
"[N
|
193 |
-
"
|
194 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
195 |
},
|
196 |
"unk_token": "[UNK]"
|
197 |
}
|
|
|
145 |
"[CLS]": 5,
|
146 |
"[SEP]": 6,
|
147 |
"C": 7,
|
148 |
+
"c": 8,
|
149 |
+
"(": 9,
|
150 |
+
")": 10,
|
151 |
+
"O": 11,
|
152 |
"1": 12,
|
153 |
+
"2": 13,
|
154 |
"=": 14,
|
155 |
+
"N": 15,
|
156 |
+
"3": 16,
|
157 |
+
"n": 17,
|
158 |
+
"[C@H]": 18,
|
159 |
+
"[C@@H]": 19,
|
160 |
+
"4": 20,
|
161 |
+
"F": 21,
|
162 |
+
"[NH+]": 22,
|
163 |
+
"S": 23,
|
164 |
+
"o": 24,
|
165 |
+
"Cl": 25,
|
166 |
+
"s": 26,
|
167 |
+
"[nH]": 27,
|
168 |
+
"5": 28,
|
169 |
+
"[NH2+]": 29,
|
170 |
+
"#": 30,
|
171 |
+
"/": 31,
|
172 |
+
"Br": 32,
|
173 |
+
"[C@@]": 33,
|
174 |
+
"[C@]": 34,
|
175 |
+
"[O-]": 35,
|
176 |
+
"\\": 36,
|
177 |
+
"[nH+]": 37,
|
178 |
+
"[NH3+]": 38,
|
179 |
+
"[n-]": 39,
|
180 |
+
"6": 40,
|
181 |
+
"I": 41,
|
182 |
+
"[N-]": 42,
|
183 |
+
"-": 43,
|
184 |
+
"7": 44,
|
185 |
+
"[H]": 45,
|
186 |
"[Si]": 46,
|
187 |
+
"P": 47,
|
188 |
+
"[n+]": 48,
|
189 |
+
"[S-]": 49,
|
190 |
+
"[S@]": 50,
|
191 |
+
"[S@@]": 51,
|
192 |
+
"[N+]": 52,
|
193 |
+
"8": 53,
|
194 |
+
"B": 54,
|
195 |
+
"[CH]": 55,
|
196 |
+
"[C]": 56,
|
197 |
+
"9": 57,
|
198 |
+
"[P@@]": 58,
|
199 |
+
"[P@]": 59,
|
200 |
+
"[S+]": 60,
|
201 |
+
"[N@@+]": 61,
|
202 |
+
"[N@+]": 62,
|
203 |
+
"[CH2]": 63,
|
204 |
+
"[O]": 64,
|
205 |
+
"[s+]": 65,
|
206 |
+
"[Sn]": 66,
|
207 |
+
"[P+]": 67,
|
208 |
+
"[B-]": 68,
|
209 |
+
"[S@@+]": 69,
|
210 |
+
"[S@+]": 70,
|
211 |
+
"p": 71,
|
212 |
+
"[N]": 72,
|
213 |
+
"%10": 73,
|
214 |
+
"[C+]": 74,
|
215 |
+
"[o+]": 75,
|
216 |
+
"%11": 76,
|
217 |
+
"[N@]": 77,
|
218 |
+
"[P@@H]": 78,
|
219 |
+
"[n@]": 79,
|
220 |
+
"[C-]": 80,
|
221 |
+
"[c+]": 81,
|
222 |
+
"[IH2]": 82,
|
223 |
+
"%13": 83,
|
224 |
+
"[Si@@]": 84,
|
225 |
+
"%12": 85,
|
226 |
+
"[Si@]": 86,
|
227 |
+
"[N@@]": 87,
|
228 |
+
"[BH3-]": 88,
|
229 |
+
"[P@H]": 89,
|
230 |
+
"[CH-]": 90,
|
231 |
+
"[Sn@]": 91,
|
232 |
+
"[s@@]": 92,
|
233 |
+
"[s@]": 93,
|
234 |
+
"[P@+]": 94,
|
235 |
+
"[P@@+]": 95,
|
236 |
+
"[Sn@@]": 96,
|
237 |
+
"[c-]": 97,
|
238 |
+
"[17O]": 98,
|
239 |
+
"[BH-]": 99,
|
240 |
+
"[SnH4+2]": 100,
|
241 |
+
"[B@-]": 101,
|
242 |
+
"[B@@-]": 102,
|
243 |
+
"[cH-]": 103,
|
244 |
+
"[O+]": 104,
|
245 |
+
"[SnH2+]": 105,
|
246 |
+
"[SnH]": 106,
|
247 |
+
"%14": 107,
|
248 |
+
"[Sn+2]": 108,
|
249 |
+
"[I+]": 109,
|
250 |
+
"[P@@H+]": 110,
|
251 |
+
"%15": 111,
|
252 |
+
"%16": 112,
|
253 |
+
"%18": 113,
|
254 |
+
"[Br+]": 114,
|
255 |
+
"[NH]": 115,
|
256 |
+
"[Sn+]": 116,
|
257 |
+
"[n@@]": 117,
|
258 |
+
"%17": 118,
|
259 |
+
"%19": 119,
|
260 |
+
"%20": 120,
|
261 |
+
"%21": 121,
|
262 |
+
"%22": 122,
|
263 |
+
"[18OH]": 123,
|
264 |
+
"[BH2-]": 124,
|
265 |
+
"[S@@-]": 125,
|
266 |
+
"[S@@H]": 126,
|
267 |
+
"[Sn+3]": 127,
|
268 |
+
"[SnH2]": 128,
|
269 |
+
"[SnH6+3]": 129,
|
270 |
+
"[pH]": 130,
|
271 |
+
"[S@H]": 131,
|
272 |
+
"[SH3]": 132,
|
273 |
+
"[SiH2]": 133,
|
274 |
+
"[SiH3]": 134,
|
275 |
+
"[Sn-]": 135,
|
276 |
+
"[p+]": 136
|
277 |
},
|
278 |
"unk_token": "[UNK]"
|
279 |
}
|