Hugging Face's logo

PoetschLab
/
GROVER

Fill-Mask
Transformers
PyTorch
bert
biology
DNA
genomics
Inference Endpoints
Model card Files Community
2
melissasanabria commited on
Commit
134f3a1
•
1 Parent(s): c67f305

adding model files

Browse files
Files changed (4) hide show
  1. config.json +52 -0
  2. pytorch_model.bin +3 -0
  3. tokenizer.json +1358 -0
  4. tokenizer_config.json +17 -0
config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertForMaskedLM"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "bos_token_id": 0,
7
+ "do_sample": false,
8
+ "eos_token_ids": 0,
9
+ "finetuning_task": null,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 768,
13
+ "id2label": {
14
+ "0": "LABEL_0",
15
+ "1": "LABEL_1"
16
+ },
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 3072,
19
+ "is_decoder": false,
20
+ "label2id": {
21
+ "LABEL_0": 0,
22
+ "LABEL_1": 1
23
+ },
24
+ "layer_norm_eps": 1e-12,
25
+ "length_penalty": 1.0,
26
+ "max_length": 20,
27
+ "max_position_embeddings": 512,
28
+ "model_type": "bert",
29
+ "num_attention_heads": 12,
30
+ "num_beams": 1,
31
+ "num_hidden_layers": 12,
32
+ "num_labels": 2,
33
+ "num_return_sequences": 1,
34
+ "num_rnn_layer": 1,
35
+ "output_attentions": false,
36
+ "output_hidden_states": false,
37
+ "output_past": true,
38
+ "pad_token_id": 0,
39
+ "pruned_heads": {},
40
+ "repetition_penalty": 1.0,
41
+ "rnn": "lstm",
42
+ "rnn_dropout": 0.0,
43
+ "rnn_hidden": 768,
44
+ "split": 10,
45
+ "temperature": 1.0,
46
+ "top_k": 50,
47
+ "top_p": 1.0,
48
+ "torchscript": false,
49
+ "type_vocab_size": 2,
50
+ "use_bfloat16": false,
51
+ "vocab_size": 609
52
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81fa4ee056e65ff1c06c32512589f3ba3bcb9d054960f35f634b9cf6da7aadda
3
+ size 348488361
tokenizer.json ADDED
@@ -0,0 +1,1358 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 1,
8
+ "special": true,
9
+ "content": "[UNK]",
10
+ "single_word": false,
11
+ "lstrip": false,
12
+ "rstrip": false,
13
+ "normalized": false
14
+ },
15
+ {
16
+ "id": 2,
17
+ "special": true,
18
+ "content": "[CLS]",
19
+ "single_word": false,
20
+ "lstrip": false,
21
+ "rstrip": false,
22
+ "normalized": false
23
+ },
24
+ {
25
+ "id": 3,
26
+ "special": true,
27
+ "content": "[SEP]",
28
+ "single_word": false,
29
+ "lstrip": false,
30
+ "rstrip": false,
31
+ "normalized": false
32
+ },
33
+ {
34
+ "id": 0,
35
+ "special": true,
36
+ "content": "[PAD]",
37
+ "single_word": false,
38
+ "lstrip": false,
39
+ "rstrip": false,
40
+ "normalized": false
41
+ },
42
+ {
43
+ "id": 4,
44
+ "special": true,
45
+ "content": "[MASK]",
46
+ "single_word": false,
47
+ "lstrip": false,
48
+ "rstrip": false,
49
+ "normalized": false
50
+ }
51
+ ],
52
+ "normalizer": null,
53
+ "pre_tokenizer": {
54
+ "type": "Whitespace"
55
+ },
56
+ "post_processor": {
57
+ "type": "TemplateProcessing",
58
+ "single": [
59
+ {
60
+ "SpecialToken": {
61
+ "id": "[CLS]",
62
+ "type_id": 0
63
+ }
64
+ },
65
+ {
66
+ "Sequence": {
67
+ "id": "A",
68
+ "type_id": 0
69
+ }
70
+ },
71
+ {
72
+ "SpecialToken": {
73
+ "id": "[SEP]",
74
+ "type_id": 0
75
+ }
76
+ }
77
+ ],
78
+ "pair": [
79
+ {
80
+ "SpecialToken": {
81
+ "id": "[CLS]",
82
+ "type_id": 0
83
+ }
84
+ },
85
+ {
86
+ "Sequence": {
87
+ "id": "A",
88
+ "type_id": 0
89
+ }
90
+ },
91
+ {
92
+ "SpecialToken": {
93
+ "id": "[SEP]",
94
+ "type_id": 0
95
+ }
96
+ },
97
+ {
98
+ "Sequence": {
99
+ "id": "B",
100
+ "type_id": 1
101
+ }
102
+ },
103
+ {
104
+ "SpecialToken": {
105
+ "id": "[SEP]",
106
+ "type_id": 1
107
+ }
108
+ }
109
+ ],
110
+ "special_tokens": {
111
+ "[CLS]": {
112
+ "id": "[CLS]",
113
+ "ids": [
114
+ 2
115
+ ],
116
+ "tokens": [
117
+ "[CLS]"
118
+ ]
119
+ },
120
+ "[SEP]": {
121
+ "id": "[SEP]",
122
+ "ids": [
123
+ 3
124
+ ],
125
+ "tokens": [
126
+ "[SEP]"
127
+ ]
128
+ }
129
+ }
130
+ },
131
+ "decoder": null,
132
+ "model": {
133
+ "type": "BPE",
134
+ "dropout": null,
135
+ "unk_token": "[UNK]",
136
+ "continuing_subword_prefix": null,
137
+ "end_of_word_suffix": null,
138
+ "fuse_unk": false,
139
+ "vocab": {
140
+ "[PAD]": 0,
141
+ "[UNK]": 1,
142
+ "[CLS]": 2,
143
+ "[SEP]": 3,
144
+ "[MASK]": 4,
145
+ "A": 5,
146
+ "C": 6,
147
+ "G": 7,
148
+ "T": 8,
149
+ "TT": 9,
150
+ "AA": 10,
151
+ "TG": 11,
152
+ "AG": 12,
153
+ "CC": 13,
154
+ "TC": 14,
155
+ "AC": 15,
156
+ "GG": 16,
157
+ "ATT": 17,
158
+ "AT": 18,
159
+ "ATG": 19,
160
+ "GC": 20,
161
+ "TAA": 21,
162
+ "TCC": 22,
163
+ "ACC": 23,
164
+ "AAAA": 24,
165
+ "AGG": 25,
166
+ "ATC": 26,
167
+ "AGC": 27,
168
+ "TTC": 28,
169
+ "AAG": 29,
170
+ "TTTT": 30,
171
+ "TGC": 31,
172
+ "TGG": 32,
173
+ "AAC": 33,
174
+ "TTG": 34,
175
+ "TAG": 35,
176
+ "TAC": 36,
177
+ "CCC": 37,
178
+ "TATT": 38,
179
+ "TGGG": 39,
180
+ "TAT": 40,
181
+ "AGAA": 41,
182
+ "AGGG": 42,
183
+ "TTTC": 43,
184
+ "AGGC": 44,
185
+ "AGCC": 45,
186
+ "ATAA": 46,
187
+ "TGTG": 47,
188
+ "TTGG": 48,
189
+ "ATTC": 49,
190
+ "AAGG": 50,
191
+ "ACAC": 51,
192
+ "TCCC": 52,
193
+ "TCTC": 53,
194
+ "TATG": 54,
195
+ "TTTG": 55,
196
+ "TTCC": 56,
197
+ "AGTG": 57,
198
+ "ATGG": 58,
199
+ "AGAC": 59,
200
+ "AAAC": 60,
201
+ "ACCC": 61,
202
+ "TGCC": 62,
203
+ "ATTG": 63,
204
+ "ATCC": 64,
205
+ "AGAG": 65,
206
+ "ATGC": 66,
207
+ "ATAC": 67,
208
+ "TCTG": 68,
209
+ "TTAA": 69,
210
+ "TCAC": 70,
211
+ "TGAA": 71,
212
+ "TGGC": 72,
213
+ "TTGC": 73,
214
+ "TAAG": 74,
215
+ "TATC": 75,
216
+ "TAAC": 76,
217
+ "AAAG": 77,
218
+ "TTAC": 78,
219
+ "AAGC": 79,
220
+ "GGG": 80,
221
+ "TAGC": 81,
222
+ "GGC": 82,
223
+ "ATAT": 83,
224
+ "TACC": 84,
225
+ "AACC": 85,
226
+ "AATG": 86,
227
+ "TAGG": 87,
228
+ "GCC": 88,
229
+ "ATATT": 89,
230
+ "AGTC": 90,
231
+ "TTTTC": 91,
232
+ "AAAAC": 92,
233
+ "TGAC": 93,
234
+ "TTTAA": 94,
235
+ "AAAAG": 95,
236
+ "AATC": 96,
237
+ "TGTC": 97,
238
+ "TTATT": 98,
239
+ "ATAG": 99,
240
+ "TGAG": 100,
241
+ "TTTTG": 101,
242
+ "AAATT": 102,
243
+ "AATT": 103,
244
+ "AATAA": 104,
245
+ "TTTCC": 105,
246
+ "ACAG": 106,
247
+ "TCAG": 107,
248
+ "AAATG": 108,
249
+ "TGGGC": 109,
250
+ "ACTC": 110,
251
+ "AGGCC": 111,
252
+ "TTAG": 112,
253
+ "ACTG": 113,
254
+ "ACG": 114,
255
+ "ATATG": 115,
256
+ "TGGCC": 116,
257
+ "ATTTC": 117,
258
+ "ACAA": 118,
259
+ "ATCTC": 119,
260
+ "TATTC": 120,
261
+ "TGTAA": 121,
262
+ "ACTT": 122,
263
+ "ATGCC": 123,
264
+ "TAAAA": 124,
265
+ "AAAAAAAA": 125,
266
+ "ATTCC": 126,
267
+ "TTTAG": 127,
268
+ "TCCCC": 128,
269
+ "TTTGC": 129,
270
+ "TTCCC": 130,
271
+ "TGGGG": 131,
272
+ "TTCTC": 132,
273
+ "ATAAAA": 133,
274
+ "AGAAG": 134,
275
+ "TTTTTTTT": 135,
276
+ "ACCCC": 136,
277
+ "AGGGC": 137,
278
+ "ACCTC": 138,
279
+ "AGATG": 139,
280
+ "ATTAC": 140,
281
+ "AAGCC": 141,
282
+ "GGCC": 142,
283
+ "AGGAG": 143,
284
+ "TCAA": 144,
285
+ "ATTGC": 145,
286
+ "TATTG": 146,
287
+ "ATAAC": 147,
288
+ "ATATC": 148,
289
+ "TTTAC": 149,
290
+ "ATGGC": 150,
291
+ "AAGGC": 151,
292
+ "ACCAC": 152,
293
+ "GTG": 153,
294
+ "ATCCC": 154,
295
+ "AGAAC": 155,
296
+ "ATTTT": 156,
297
+ "TTGCC": 157,
298
+ "AAATC": 158,
299
+ "ATAAG": 159,
300
+ "TTGGC": 160,
301
+ "TGGAG": 161,
302
+ "ATGGG": 162,
303
+ "AAAGC": 163,
304
+ "AGGGG": 164,
305
+ "ATCAC": 165,
306
+ "ATTTG": 166,
307
+ "AATTC": 167,
308
+ "TGCAC": 168,
309
+ "TTTGG": 169,
310
+ "TCG": 170,
311
+ "AGAGC": 171,
312
+ "AAAGG": 172,
313
+ "GGGC": 173,
314
+ "TTGGG": 174,
315
+ "AGAAAA": 175,
316
+ "TATCC": 176,
317
+ "TCTCC": 177,
318
+ "ATAGC": 178,
319
+ "TGAGG": 179,
320
+ "TTTATT": 180,
321
+ "AGTAA": 181,
322
+ "AGAGG": 182,
323
+ "TCTTC": 183,
324
+ "ACATT": 184,
325
+ "TCCTG": 185,
326
+ "AGCCC": 186,
327
+ "TATGC": 187,
328
+ "TTAAAA": 188,
329
+ "AGATT": 189,
330
+ "TTAAC": 190,
331
+ "GGGG": 191,
332
+ "AAGAC": 192,
333
+ "TCATT": 193,
334
+ "TTCTG": 194,
335
+ "AGACC": 195,
336
+ "AAGGG": 196,
337
+ "ATACC": 197,
338
+ "TTTAT": 198,
339
+ "AAGTG": 199,
340
+ "TTATG": 200,
341
+ "AAGAA": 201,
342
+ "TAGCC": 202,
343
+ "TTCAC": 203,
344
+ "AGGTG": 204,
345
+ "TTGAA": 205,
346
+ "ATCTG": 206,
347
+ "AGCAC": 207,
348
+ "TGCTG": 208,
349
+ "AAACC": 209,
350
+ "ATGTG": 210,
351
+ "TTTTCC": 211,
352
+ "AGTTC": 212,
353
+ "TCCTC": 213,
354
+ "TATGG": 214,
355
+ "AATAC": 215,
356
+ "AGTGG": 216,
357
+ "TAGGC": 217,
358
+ "AGCTC": 218,
359
+ "ATAGG": 219,
360
+ "TTATC": 220,
361
+ "TTAAG": 221,
362
+ "TACCC": 222,
363
+ "TTTTTG": 223,
364
+ "AACAC": 224,
365
+ "TGCTC": 225,
366
+ "AGATC": 226,
367
+ "TCCCAGC": 227,
368
+ "AGCTG": 228,
369
+ "AATAG": 229,
370
+ "TCTTG": 230,
371
+ "AGTGGC": 231,
372
+ "ATTGG": 232,
373
+ "TACTC": 233,
374
+ "TAAAC": 234,
375
+ "AATGG": 235,
376
+ "AGGTC": 236,
377
+ "AGGAC": 237,
378
+ "TTGTG": 238,
379
+ "TATAC": 239,
380
+ "ATTTTC": 240,
381
+ "ATATAA": 241,
382
+ "AGGCTG": 242,
383
+ "ATTTAA": 243,
384
+ "AGTT": 244,
385
+ "AGTAG": 245,
386
+ "ATGAC": 246,
387
+ "AATGC": 247,
388
+ "TCCAC": 248,
389
+ "CCCC": 249,
390
+ "ATGTC": 250,
391
+ "AACTC": 251,
392
+ "TTTTTC": 252,
393
+ "TAAGC": 253,
394
+ "AAGTC": 254,
395
+ "TGGTG": 255,
396
+ "TATAA": 256,
397
+ "AGTGC": 257,
398
+ "TAAGG": 258,
399
+ "ACCTG": 259,
400
+ "TTAGC": 260,
401
+ "AAATAA": 261,
402
+ "TGCCTC": 262,
403
+ "AATCC": 263,
404
+ "TTGGCC": 264,
405
+ "TAGGG": 265,
406
+ "TGGAC": 266,
407
+ "TTGTC": 267,
408
+ "AACCC": 268,
409
+ "TTACC": 269,
410
+ "TAACC": 270,
411
+ "AATTTT": 271,
412
+ "AAAGAA": 272,
413
+ "ATTATT": 273,
414
+ "AGCG": 274,
415
+ "AAAAAC": 275,
416
+ "TAATG": 276,
417
+ "TTGAC": 277,
418
+ "AGTCC": 278,
419
+ "AACTG": 279,
420
+ "AGTTG": 280,
421
+ "AATTG": 281,
422
+ "TCTGC": 282,
423
+ "TTAGG": 283,
424
+ "TACAC": 284,
425
+ "AGAAGG": 285,
426
+ "ATATTC": 286,
427
+ "AAAACC": 287,
428
+ "AAAAGC": 288,
429
+ "TGCCC": 289,
430
+ "ACTGC": 290,
431
+ "AGAAGC": 291,
432
+ "TAATAA": 292,
433
+ "AATATT": 293,
434
+ "ACCATG": 294,
435
+ "TGGTC": 295,
436
+ "TTTTGC": 296,
437
+ "AACG": 297,
438
+ "TACTG": 298,
439
+ "ACACACAC": 299,
440
+ "ATTTTG": 300,
441
+ "TCCG": 301,
442
+ "TGCG": 302,
443
+ "AAAATG": 303,
444
+ "ACATG": 304,
445
+ "TCAGC": 305,
446
+ "ATCG": 306,
447
+ "AGTAC": 307,
448
+ "TTTTGG": 308,
449
+ "AATAT": 309,
450
+ "AGAGAA": 310,
451
+ "TTCG": 311,
452
+ "TCCAGCC": 312,
453
+ "ATATAC": 313,
454
+ "TCACC": 314,
455
+ "AAAAGG": 315,
456
+ "TGTGTGTG": 316,
457
+ "TCATC": 317,
458
+ "TGCTGGG": 318,
459
+ "TGAAG": 319,
460
+ "TGTAG": 320,
461
+ "TGTGG": 321,
462
+ "AAAAATT": 322,
463
+ "ACTTC": 323,
464
+ "TTCCCC": 324,
465
+ "ATAGAA": 325,
466
+ "TTGCCC": 326,
467
+ "AGGAGG": 327,
468
+ "TTTCCC": 328,
469
+ "TATATT": 329,
470
+ "ACCG": 330,
471
+ "ACTAC": 331,
472
+ "TCACTGC": 332,
473
+ "GCG": 333,
474
+ "TTTGTG": 334,
475
+ "ACAGC": 335,
476
+ "TCATG": 336,
477
+ "AGTTTT": 337,
478
+ "AGGAA": 338,
479
+ "TTTATG": 339,
480
+ "ATATTG": 340,
481
+ "TGATG": 341,
482
+ "TCTAA": 342,
483
+ "TGTGC": 343,
484
+ "AGGAAG": 344,
485
+ "TTTGGG": 345,
486
+ "TGTTC": 346,
487
+ "AGCCCC": 347,
488
+ "AGTTTC": 348,
489
+ "AGGCTGG": 349,
490
+ "TTTGCC": 350,
491
+ "ATTTCC": 351,
492
+ "ATACAC": 352,
493
+ "AAAATAA": 353,
494
+ "TAGAC": 354,
495
+ "AGGAGAA": 355,
496
+ "TGAGC": 356,
497
+ "TGGAA": 357,
498
+ "TTTTTAA": 358,
499
+ "AGCCTCCC": 359,
500
+ "ATGAA": 360,
501
+ "TTTAAG": 361,
502
+ "TCTGG": 362,
503
+ "TTTATC": 363,
504
+ "TTATAA": 364,
505
+ "TGATT": 365,
506
+ "AACAA": 366,
507
+ "TAGCTGGG": 367,
508
+ "TCAAG": 368,
509
+ "AAAAAA": 369,
510
+ "ACTTTGGG": 370,
511
+ "TATTCC": 371,
512
+ "TCAGG": 372,
513
+ "AACAG": 373,
514
+ "TTCTTC": 374,
515
+ "TGTGGC": 375,
516
+ "ATATGC": 376,
517
+ "ATTACAGGC": 377,
518
+ "AGGGGC": 378,
519
+ "AGGGCC": 379,
520
+ "TTATTC": 380,
521
+ "ATATCC": 381,
522
+ "TGTAATCCCAGC": 382,
523
+ "TACG": 383,
524
+ "AGAAAC": 384,
525
+ "TGTCC": 385,
526
+ "AGATGG": 386,
527
+ "TGTGCC": 387,
528
+ "TTTCTC": 388,
529
+ "TGAAC": 389,
530
+ "AGTCTC": 390,
531
+ "TGTTG": 391,
532
+ "ATTTTTT": 392,
533
+ "AAGAAG": 393,
534
+ "TGGGGC": 394,
535
+ "AGCAGC": 395,
536
+ "GCCC": 396,
537
+ "TTTGGC": 397,
538
+ "AGGCTGAGGC": 398,
539
+ "TGGGCC": 399,
540
+ "TTCTCC": 400,
541
+ "TAGAA": 401,
542
+ "TGGAGTGC": 402,
543
+ "ATTAA": 403,
544
+ "AGTGCC": 404,
545
+ "TGTCTC": 405,
546
+ "ATATGG": 406,
547
+ "ACATC": 407,
548
+ "TGGGGG": 408,
549
+ "TGACC": 409,
550
+ "ACTCC": 410,
551
+ "TAAAAC": 411,
552
+ "AGATAA": 412,
553
+ "TAATTTT": 413,
554
+ "TCAAC": 414,
555
+ "TCTAC": 415,
556
+ "TCTAG": 416,
557
+ "GAG": 417,
558
+ "TAAATG": 418,
559
+ "AGCAA": 419,
560
+ "TATATG": 420,
561
+ "ATATATAT": 421,
562
+ "ATTTGC": 422,
563
+ "TCCTCC": 423,
564
+ "CCCAC": 424,
565
+ "ATTTATT": 425,
566
+ "TCTGCC": 426,
567
+ "ATGGCC": 427,
568
+ "TCGC": 428,
569
+ "AGTATT": 429,
570
+ "AGAACC": 430,
571
+ "TTAAAC": 431,
572
+ "AAATTC": 432,
573
+ "AGAGAC": 433,
574
+ "ATTTAC": 434,
575
+ "ATTGCC": 435,
576
+ "AACAAC": 436,
577
+ "TTTAAC": 437,
578
+ "ACGG": 438,
579
+ "AAGAAAA": 439,
580
+ "TCTGGC": 440,
581
+ "ATTCTCC": 441,
582
+ "AGGTGG": 442,
583
+ "TGCTGC": 443,
584
+ "TTCAAG": 444,
585
+ "AGAGGG": 445,
586
+ "ACACC": 446,
587
+ "TCTTTT": 447,
588
+ "AGAGGC": 448,
589
+ "ATCACC": 449,
590
+ "TAAATT": 450,
591
+ "AAGGCC": 451,
592
+ "TTGCAGTG": 452,
593
+ "TGTAC": 453,
594
+ "AATTTC": 454,
595
+ "ATCCCC": 455,
596
+ "ACAAG": 456,
597
+ "ACAGG": 457,
598
+ "ACAAC": 458,
599
+ "TGCCCC": 459,
600
+ "AGATTC": 460,
601
+ "TTAGAA": 461,
602
+ "TTGGGG": 462,
603
+ "AGACAC": 463,
604
+ "TGGAAG": 464,
605
+ "ACCTCC": 465,
606
+ "ATGGGG": 466,
607
+ "AGCCTCC": 467,
608
+ "TTATTG": 468,
609
+ "TAAAAG": 469,
610
+ "ATCTTC": 470,
611
+ "ATCTCC": 471,
612
+ "TGAAGC": 472,
613
+ "TAATC": 473,
614
+ "AAATGC": 474,
615
+ "TTGTTG": 475,
616
+ "ATTCCC": 476,
617
+ "TACTAAAA": 477,
618
+ "ATAGTG": 478,
619
+ "AAATAC": 479,
620
+ "TTGGGC": 480,
621
+ "TAGAGAC": 481,
622
+ "TGTTTT": 482,
623
+ "TTCTGC": 483,
624
+ "TGGCCC": 484,
625
+ "TCTGTC": 485,
626
+ "AGCTCC": 486,
627
+ "AACTCC": 487,
628
+ "TTAGCC": 488,
629
+ "AAAGTGCTGGG": 489,
630
+ "ATAGAC": 490,
631
+ "TATTTTTAG": 491,
632
+ "ACTTG": 492,
633
+ "ACCACC": 493,
634
+ "AAACAC": 494,
635
+ "GTGG": 495,
636
+ "ATTTAG": 496,
637
+ "AGGAGC": 497,
638
+ "AGGCTGGAGTGC": 498,
639
+ "ATACCC": 499,
640
+ "ATGTAA": 500,
641
+ "ACGC": 501,
642
+ "AGTAT": 502,
643
+ "TTTACC": 503,
644
+ "ACTAA": 504,
645
+ "AGGCCC": 505,
646
+ "AAGGGG": 506,
647
+ "TCTCG": 507,
648
+ "ATGAAG": 508,
649
+ "AAAGAC": 509,
650
+ "TGAAAA": 510,
651
+ "AAGGGC": 511,
652
+ "ATAGGC": 512,
653
+ "AGAGTG": 513,
654
+ "AGCTGC": 514,
655
+ "ATGTTC": 515,
656
+ "TATTTC": 516,
657
+ "TGATC": 517,
658
+ "AGTTTG": 518,
659
+ "AGCTAA": 519,
660
+ "AGAGCC": 520,
661
+ "TGCTTC": 521,
662
+ "ATCATC": 522,
663
+ "AACATGG": 523,
664
+ "AGCTTC": 524,
665
+ "AAGAAC": 525,
666
+ "TTTTTTG": 526,
667
+ "AGGGGG": 527,
668
+ "ATAAGC": 528,
669
+ "TAAGCC": 529,
670
+ "ACTGG": 530,
671
+ "ACAAAA": 531,
672
+ "ATCATT": 532,
673
+ "TCTTTC": 533,
674
+ "ATGATG": 534,
675
+ "TGCAA": 535,
676
+ "AGGTTC": 536,
677
+ "AACATT": 537,
678
+ "ATGGGC": 538,
679
+ "ATAGAG": 539,
680
+ "AAATGG": 540,
681
+ "AGTTCC": 541,
682
+ "TTTAGC": 542,
683
+ "AACTTC": 543,
684
+ "AGCAAG": 544,
685
+ "ATAAAAC": 545,
686
+ "AAAATC": 546,
687
+ "AGCCAC": 547,
688
+ "AGGAAC": 548,
689
+ "TTAACC": 549,
690
+ "TATTTATT": 550,
691
+ "TTTCTG": 551,
692
+ "ATAAGG": 552,
693
+ "AGCCACC": 553,
694
+ "AGATGC": 554,
695
+ "TTAAGC": 555,
696
+ "TTGTAA": 556,
697
+ "AGTGTG": 557,
698
+ "AACCCC": 558,
699
+ "TTCATT": 559,
700
+ "ATCATG": 560,
701
+ "AATGAA": 561,
702
+ "AGGTGC": 562,
703
+ "AAAAAAAAAAAAAAAA": 563,
704
+ "AGGATG": 564,
705
+ "AGCCG": 565,
706
+ "TGGTGG": 566,
707
+ "AGTGGG": 567,
708
+ "TGCACTCCAGCC": 568,
709
+ "TATTGC": 569,
710
+ "TAGTC": 570,
711
+ "CCCG": 571,
712
+ "AAGTAA": 572,
713
+ "TAGTG": 573,
714
+ "TTTTTTTTTTTTTTTT": 574,
715
+ "AGCATT": 575,
716
+ "ATCTGC": 576,
717
+ "TCTCAC": 577,
718
+ "AAATTG": 578,
719
+ "TTTAGG": 579,
720
+ "AGACCC": 580,
721
+ "GGGCC": 581,
722
+ "TCCTTC": 582,
723
+ "ATAGGG": 583,
724
+ "AATATG": 584,
725
+ "TTATAC": 585,
726
+ "TAGAAG": 586,
727
+ "AAAGTG": 587,
728
+ "AAATCC": 588,
729
+ "TTCCTC": 589,
730
+ "TTTCAC": 590,
731
+ "AGTATG": 591,
732
+ "TACTAAAAATAC": 592,
733
+ "ATGTGC": 593,
734
+ "AGGAGGC": 594,
735
+ "TATATC": 595,
736
+ "TTCTAA": 596,
737
+ "TGAGGC": 597,
738
+ "ACACAC": 598,
739
+ "TCCCCC": 599,
740
+ "AACATC": 600,
741
+ "AAGCG": 601,
742
+ "AATGGC": 602,
743
+ "ACCCCC": 603,
744
+ "AGATAC": 604,
745
+ "ATAAAAG": 605,
746
+ "ATGATT": 606,
747
+ "TGGAGG": 607,
748
+ "AGTTAA": 608,
749
+ "": 609
750
+ },
751
+ "merges": [
752
+ "A ",
753
+ "C ",
754
+ "G ",
755
+ "T ",
756
+ "T T",
757
+ "A A",
758
+ "T G",
759
+ "A G",
760
+ "C C",
761
+ "T C",
762
+ "A C",
763
+ "G G",
764
+ "A TT",
765
+ "A T",
766
+ "A TG",
767
+ "G C",
768
+ "T AA",
769
+ "T CC",
770
+ "A CC",
771
+ "AA AA",
772
+ "AG G",
773
+ "A TC",
774
+ "AG C",
775
+ "TT C",
776
+ "AA G",
777
+ "TT TT",
778
+ "TG C",
779
+ "TG G",
780
+ "AA C",
781
+ "TT G",
782
+ "T AG",
783
+ "T AC",
784
+ "CC C",
785
+ "T ATT",
786
+ "TG GG",
787
+ "T AT",
788
+ "AG AA",
789
+ "AG GG",
790
+ "TT TC",
791
+ "AG GC",
792
+ "AG CC",
793
+ "AT AA",
794
+ "TG TG",
795
+ "TT GG",
796
+ "ATT C",
797
+ "AA GG",
798
+ "AC AC",
799
+ "TCC C",
800
+ "TC TC",
801
+ "T ATG",
802
+ "TT TG",
803
+ "TT CC",
804
+ "AG TG",
805
+ "ATG G",
806
+ "AG AC",
807
+ "AA AC",
808
+ "ACC C",
809
+ "TG CC",
810
+ "ATT G",
811
+ "AT CC",
812
+ "AG AG",
813
+ "ATG C",
814
+ "AT AC",
815
+ "TC TG",
816
+ "TT AA",
817
+ "TC AC",
818
+ "TG AA",
819
+ "TG GC",
820
+ "TT GC",
821
+ "TAA G",
822
+ "T ATC",
823
+ "TAA C",
824
+ "AA AG",
825
+ "TT AC",
826
+ "AA GC",
827
+ "GG G",
828
+ "T AGC",
829
+ "GG C",
830
+ "AT AT",
831
+ "T ACC",
832
+ "AA CC",
833
+ "AA TG",
834
+ "T AGG",
835
+ "G CC",
836
+ "AT ATT",
837
+ "AG TC",
838
+ "TT TTC",
839
+ "AAAA C",
840
+ "TG AC",
841
+ "TT TAA",
842
+ "AAAA G",
843
+ "AA TC",
844
+ "TG TC",
845
+ "TT ATT",
846
+ "AT AG",
847
+ "TG AG",
848
+ "TTTT G",
849
+ "AA ATT",
850
+ "AA TT",
851
+ "AA TAA",
852
+ "TT TCC",
853
+ "AC AG",
854
+ "TC AG",
855
+ "AA ATG",
856
+ "TGGG C",
857
+ "AC TC",
858
+ "AGG CC",
859
+ "TT AG",
860
+ "AC TG",
861
+ "AC G",
862
+ "AT ATG",
863
+ "TGG CC",
864
+ "ATT TC",
865
+ "AC AA",
866
+ "ATC TC",
867
+ "TATT C",
868
+ "TG TAA",
869
+ "AC TT",
870
+ "ATG CC",
871
+ "TAA AA",
872
+ "AAAA AAAA",
873
+ "ATT CC",
874
+ "TT TAG",
875
+ "TCC CC",
876
+ "TT TGC",
877
+ "TT CCC",
878
+ "TGGG G",
879
+ "TTC TC",
880
+ "AT AAAA",
881
+ "AG AAG",
882
+ "TTTT TTTT",
883
+ "ACC CC",
884
+ "AGGG C",
885
+ "ACC TC",
886
+ "AG ATG",
887
+ "ATT AC",
888
+ "AAG CC",
889
+ "GG CC",
890
+ "AGG AG",
891
+ "TC AA",
892
+ "ATT GC",
893
+ "TATT G",
894
+ "AT AAC",
895
+ "AT ATC",
896
+ "TT TAC",
897
+ "ATG GC",
898
+ "AAGG C",
899
+ "ACC AC",
900
+ "G TG",
901
+ "AT CCC",
902
+ "AG AAC",
903
+ "ATT TT",
904
+ "TTG CC",
905
+ "AA ATC",
906
+ "AT AAG",
907
+ "TTGG C",
908
+ "TGG AG",
909
+ "ATG GG",
910
+ "AA AGC",
911
+ "AGGG G",
912
+ "ATC AC",
913
+ "ATT TG",
914
+ "AA TTC",
915
+ "TGC AC",
916
+ "TT TGG",
917
+ "TC G",
918
+ "AG AGC",
919
+ "AA AGG",
920
+ "GG GC",
921
+ "TTGG G",
922
+ "AG AAAA",
923
+ "TAT CC",
924
+ "TC TCC",
925
+ "AT AGC",
926
+ "TG AGG",
927
+ "TT TATT",
928
+ "AG TAA",
929
+ "AG AGG",
930
+ "TC TTC",
931
+ "AC ATT",
932
+ "TCC TG",
933
+ "AG CCC",
934
+ "TATG C",
935
+ "TT AAAA",
936
+ "AG ATT",
937
+ "TT AAC",
938
+ "GG GG",
939
+ "AAG AC",
940
+ "TC ATT",
941
+ "TTC TG",
942
+ "AG ACC",
943
+ "AAGG G",
944
+ "AT ACC",
945
+ "TT TAT",
946
+ "AAG TG",
947
+ "TT ATG",
948
+ "AAG AA",
949
+ "TAG CC",
950
+ "TTC AC",
951
+ "AGG TG",
952
+ "TTG AA",
953
+ "ATC TG",
954
+ "AGC AC",
955
+ "TGC TG",
956
+ "AA ACC",
957
+ "ATG TG",
958
+ "TTTT CC",
959
+ "AG TTC",
960
+ "TCC TC",
961
+ "TATG G",
962
+ "AA TAC",
963
+ "AG TGG",
964
+ "TAG GC",
965
+ "AGC TC",
966
+ "AT AGG",
967
+ "TT ATC",
968
+ "TT AAG",
969
+ "T ACCC",
970
+ "TTTT TG",
971
+ "AAC AC",
972
+ "TGC TC",
973
+ "AG ATC",
974
+ "TCCC AGC",
975
+ "AGC TG",
976
+ "AA TAG",
977
+ "TC TTG",
978
+ "AGTG GC",
979
+ "ATT GG",
980
+ "TAC TC",
981
+ "TAA AC",
982
+ "AA TGG",
983
+ "AGG TC",
984
+ "AGG AC",
985
+ "TTG TG",
986
+ "TAT AC",
987
+ "ATT TTC",
988
+ "AT ATAA",
989
+ "AGGC TG",
990
+ "ATT TAA",
991
+ "AG TT",
992
+ "AG TAG",
993
+ "ATG AC",
994
+ "AA TGC",
995
+ "TCC AC",
996
+ "CC CC",
997
+ "ATG TC",
998
+ "AAC TC",
999
+ "TTTT TC",
1000
+ "TAA GC",
1001
+ "AAG TC",
1002
+ "TGG TG",
1003
+ "TAT AA",
1004
+ "AG TGC",
1005
+ "TAA GG",
1006
+ "ACC TG",
1007
+ "TT AGC",
1008
+ "AA ATAA",
1009
+ "TGCC TC",
1010
+ "AA TCC",
1011
+ "TTGG CC",
1012
+ "TAG GG",
1013
+ "TGG AC",
1014
+ "TTG TC",
1015
+ "AA CCC",
1016
+ "TT ACC",
1017
+ "TAA CC",
1018
+ "AA TTTT",
1019
+ "AA AGAA",
1020
+ "ATT ATT",
1021
+ "AGC G",
1022
+ "AAAA AC",
1023
+ "TAA TG",
1024
+ "TTG AC",
1025
+ "AG TCC",
1026
+ "AAC TG",
1027
+ "AG TTG",
1028
+ "AA TTG",
1029
+ "TC TGC",
1030
+ "TT AGG",
1031
+ "TAC AC",
1032
+ "AGAA GG",
1033
+ "AT ATTC",
1034
+ "AAAA CC",
1035
+ "AAAA GC",
1036
+ "TG CCC",
1037
+ "AC TGC",
1038
+ "AGAA GC",
1039
+ "TAA TAA",
1040
+ "AA TATT",
1041
+ "ACC ATG",
1042
+ "TGG TC",
1043
+ "TTTT GC",
1044
+ "AAC G",
1045
+ "TAC TG",
1046
+ "ACAC ACAC",
1047
+ "ATT TTG",
1048
+ "TCC G",
1049
+ "TGC G",
1050
+ "AAAA TG",
1051
+ "AC ATG",
1052
+ "TC AGC",
1053
+ "ATC G",
1054
+ "AG TAC",
1055
+ "TTTT GG",
1056
+ "AA TAT",
1057
+ "AG AGAA",
1058
+ "TTC G",
1059
+ "TCC AGCC",
1060
+ "AT ATAC",
1061
+ "TC ACC",
1062
+ "AAAA GG",
1063
+ "TGTG TGTG",
1064
+ "TC ATC",
1065
+ "TGC TGGG",
1066
+ "TG AAG",
1067
+ "TG TAG",
1068
+ "TG TGG",
1069
+ "AAAA ATT",
1070
+ "AC TTC",
1071
+ "TTCC CC",
1072
+ "AT AGAA",
1073
+ "TTG CCC",
1074
+ "AGG AGG",
1075
+ "TT TCCC",
1076
+ "TAT ATT",
1077
+ "ACC G",
1078
+ "AC TAC",
1079
+ "TCAC TGC",
1080
+ "GC G",
1081
+ "TT TGTG",
1082
+ "AC AGC",
1083
+ "TC ATG",
1084
+ "AG TTTT",
1085
+ "AGG AA",
1086
+ "TT TATG",
1087
+ "AT ATTG",
1088
+ "TG ATG",
1089
+ "TC TAA",
1090
+ "TG TGC",
1091
+ "AGG AAG",
1092
+ "TT TGGG",
1093
+ "TG TTC",
1094
+ "AGCC CC",
1095
+ "AG TTTC",
1096
+ "AGGC TGG",
1097
+ "TTTG CC",
1098
+ "ATT TCC",
1099
+ "AT ACAC",
1100
+ "AAAA TAA",
1101
+ "TAG AC",
1102
+ "AGG AGAA",
1103
+ "TG AGC",
1104
+ "TGG AA",
1105
+ "TTTT TAA",
1106
+ "AGCC TCCC",
1107
+ "ATG AA",
1108
+ "TT TAAG",
1109
+ "TC TGG",
1110
+ "TT TATC",
1111
+ "TT ATAA",
1112
+ "TG ATT",
1113
+ "AAC AA",
1114
+ "TAGC TGGG",
1115
+ "TC AAG",
1116
+ "AAAA AA",
1117
+ "ACTT TGGG",
1118
+ "TATT CC",
1119
+ "TC AGG",
1120
+ "AAC AG",
1121
+ "TTC TTC",
1122
+ "TGTG GC",
1123
+ "AT ATGC",
1124
+ "ATTAC AGGC",
1125
+ "AGGG GC",
1126
+ "AGGG CC",
1127
+ "TT ATTC",
1128
+ "AT ATCC",
1129
+ "TGTAA TCCCAGC",
1130
+ "TAC G",
1131
+ "AGAA AC",
1132
+ "TG TCC",
1133
+ "AG ATGG",
1134
+ "TGTG CC",
1135
+ "TTTC TC",
1136
+ "TG AAC",
1137
+ "AG TCTC",
1138
+ "TG TTG",
1139
+ "ATT TTTT",
1140
+ "AAG AAG",
1141
+ "TGGG GC",
1142
+ "AGC AGC",
1143
+ "G CCC",
1144
+ "TTTG GC",
1145
+ "AGGCTG AGGC",
1146
+ "TGGG CC",
1147
+ "TTC TCC",
1148
+ "TAG AA",
1149
+ "TGGAG TGC",
1150
+ "ATT AA",
1151
+ "AGTG CC",
1152
+ "TG TCTC",
1153
+ "AT ATGG",
1154
+ "AC ATC",
1155
+ "TGGG GG",
1156
+ "TG ACC",
1157
+ "AC TCC",
1158
+ "TAA AAC",
1159
+ "AG ATAA",
1160
+ "TAA TTTT",
1161
+ "TC AAC",
1162
+ "TC TAC",
1163
+ "TC TAG",
1164
+ "G AG",
1165
+ "TAA ATG",
1166
+ "AGC AA",
1167
+ "TAT ATG",
1168
+ "ATAT ATAT",
1169
+ "ATT TGC",
1170
+ "TCC TCC",
1171
+ "CCC AC",
1172
+ "ATT TATT",
1173
+ "TC TGCC",
1174
+ "ATGG CC",
1175
+ "TC GC",
1176
+ "AG TATT",
1177
+ "AGAA CC",
1178
+ "TT AAAC",
1179
+ "AA ATTC",
1180
+ "AG AGAC",
1181
+ "ATT TAC",
1182
+ "ATTG CC",
1183
+ "AAC AAC",
1184
+ "TT TAAC",
1185
+ "AC GG",
1186
+ "AAG AAAA",
1187
+ "TCTG GC",
1188
+ "ATTC TCC",
1189
+ "AGG TGG",
1190
+ "TGC TGC",
1191
+ "TTC AAG",
1192
+ "AG AGGG",
1193
+ "AC ACC",
1194
+ "TC TTTT",
1195
+ "AG AGGC",
1196
+ "ATC ACC",
1197
+ "TAA ATT",
1198
+ "AAGG CC",
1199
+ "TTGC AGTG",
1200
+ "TG TAC",
1201
+ "AA TTTC",
1202
+ "ATCC CC",
1203
+ "AC AAG",
1204
+ "AC AGG",
1205
+ "AC AAC",
1206
+ "TGCC CC",
1207
+ "AG ATTC",
1208
+ "TT AGAA",
1209
+ "TTGG GG",
1210
+ "AG ACAC",
1211
+ "TGG AAG",
1212
+ "ACC TCC",
1213
+ "ATG GGG",
1214
+ "AGCC TCC",
1215
+ "TT ATTG",
1216
+ "TAA AAG",
1217
+ "ATC TTC",
1218
+ "ATC TCC",
1219
+ "TGAA GC",
1220
+ "TAA TC",
1221
+ "AA ATGC",
1222
+ "TTG TTG",
1223
+ "ATT CCC",
1224
+ "TAC TAAAA",
1225
+ "AT AGTG",
1226
+ "AA ATAC",
1227
+ "TTGG GC",
1228
+ "TAG AGAC",
1229
+ "TG TTTT",
1230
+ "TTC TGC",
1231
+ "TGG CCC",
1232
+ "TCTG TC",
1233
+ "AGC TCC",
1234
+ "AAC TCC",
1235
+ "TT AGCC",
1236
+ "AAAG TGCTGGG",
1237
+ "AT AGAC",
1238
+ "TATT TTTAG",
1239
+ "AC TTG",
1240
+ "ACC ACC",
1241
+ "AA ACAC",
1242
+ "G TGG",
1243
+ "ATT TAG",
1244
+ "AGG AGC",
1245
+ "AGGC TGGAGTGC",
1246
+ "AT ACCC",
1247
+ "ATG TAA",
1248
+ "AC GC",
1249
+ "AG TAT",
1250
+ "TT TACC",
1251
+ "AC TAA",
1252
+ "AGG CCC",
1253
+ "AAGG GG",
1254
+ "TCTC G",
1255
+ "ATG AAG",
1256
+ "AA AGAC",
1257
+ "TG AAAA",
1258
+ "AAGG GC",
1259
+ "AT AGGC",
1260
+ "AG AGTG",
1261
+ "AGC TGC",
1262
+ "ATG TTC",
1263
+ "TATT TC",
1264
+ "TG ATC",
1265
+ "AG TTTG",
1266
+ "AGC TAA",
1267
+ "AG AGCC",
1268
+ "TGC TTC",
1269
+ "ATC ATC",
1270
+ "AAC ATGG",
1271
+ "AGC TTC",
1272
+ "AAG AAC",
1273
+ "TTTT TTG",
1274
+ "AGGG GG",
1275
+ "ATAA GC",
1276
+ "TAAG CC",
1277
+ "AC TGG",
1278
+ "AC AAAA",
1279
+ "ATC ATT",
1280
+ "TC TTTC",
1281
+ "ATG ATG",
1282
+ "TGC AA",
1283
+ "AGG TTC",
1284
+ "AAC ATT",
1285
+ "ATG GGC",
1286
+ "AT AGAG",
1287
+ "AA ATGG",
1288
+ "AG TTCC",
1289
+ "TT TAGC",
1290
+ "AAC TTC",
1291
+ "AGC AAG",
1292
+ "AT AAAAC",
1293
+ "AAAA TC",
1294
+ "AGCC AC",
1295
+ "AGG AAC",
1296
+ "TTAA CC",
1297
+ "TATT TATT",
1298
+ "TTTC TG",
1299
+ "ATAA GG",
1300
+ "AGCC ACC",
1301
+ "AG ATGC",
1302
+ "TTAA GC",
1303
+ "TTG TAA",
1304
+ "AG TGTG",
1305
+ "AACC CC",
1306
+ "TTC ATT",
1307
+ "ATC ATG",
1308
+ "AA TGAA",
1309
+ "AGG TGC",
1310
+ "AAAAAAAA AAAAAAAA",
1311
+ "AGG ATG",
1312
+ "AGCC G",
1313
+ "TGG TGG",
1314
+ "AG TGGG",
1315
+ "TGCAC TCCAGCC",
1316
+ "TATT GC",
1317
+ "TAG TC",
1318
+ "CCC G",
1319
+ "AAG TAA",
1320
+ "TAG TG",
1321
+ "TTTTTTTT TTTTTTTT",
1322
+ "AGC ATT",
1323
+ "ATC TGC",
1324
+ "TCTC AC",
1325
+ "AA ATTG",
1326
+ "TT TAGG",
1327
+ "AG ACCC",
1328
+ "GGG CC",
1329
+ "TCC TTC",
1330
+ "AT AGGG",
1331
+ "AA TATG",
1332
+ "TT ATAC",
1333
+ "TAG AAG",
1334
+ "AA AGTG",
1335
+ "AA ATCC",
1336
+ "TTCC TC",
1337
+ "TTTC AC",
1338
+ "AG TATG",
1339
+ "TACTAAAA ATAC",
1340
+ "ATG TGC",
1341
+ "AGG AGGC",
1342
+ "TAT ATC",
1343
+ "TTC TAA",
1344
+ "TG AGGC",
1345
+ "ACAC AC",
1346
+ "TCC CCC",
1347
+ "AAC ATC",
1348
+ "AAGC G",
1349
+ "AA TGGC",
1350
+ "ACC CCC",
1351
+ "AG ATAC",
1352
+ "AT AAAAG",
1353
+ "ATG ATT",
1354
+ "TGG AGG",
1355
+ "AG TTAA"
1356
+ ]
1357
+ }
1358
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "clean_up_tokenization_spaces": true,
3
+ "cls_token": "[CLS]",
4
+ "do_basic_tokenize": true,
5
+ "do_lower_case": true,
6
+ "mask_token": "[MASK]",
7
+ "max_len": 512,
8
+ "model_max_length": 512,
9
+ "never_split": null,
10
+ "pad_token": "[PAD]",
11
+ "sep_token": "[SEP]",
12
+ "strip_accents": null,
13
+ "tokenize_chinese_chars": true,
14
+ "tokenizer_class": "BertTokenizer",
15
+ "unk_token": "[UNK]"
16
+ }
17
+