Hugging Face's logo

MolGen
/
llama_ZINC_1B-raw_tokenizer_bpe_None_SAFE_500_0_0.1_f025f239

Model card Files Community
kmchiti commited on
Commit
57314e6
•
1 Parent(s): 3bb9262

Training in progress, step 35000, checkpoint

Browse files
Files changed (6) hide show
  1. tmp-spec-checkpoint-35000/config.json +30 -0
  2. tmp-spec-checkpoint-35000/pytorch_model.bin +3 -0
  3. tmp-spec-checkpoint-35000/special_tokens_map.json +30 -0
  4. tmp-spec-checkpoint-35000/tokenizer.json +1079 -0
  5. tmp-spec-checkpoint-35000/tokenizer_config.json +43 -0
  6. tmp-spec-checkpoint-35000/training_args.bin +3 -0
tmp-spec-checkpoint-35000/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attention_bias": false,
3
+ "attention_dropout": 0.0,
4
+ "bos_token_id": 2,
5
+ "eos_token_id": 3,
6
+ "fused_bias_fc": false,
7
+ "fused_dropout_add_ln": false,
8
+ "fused_mlp": false,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 512,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 1024,
13
+ "max_position_embeddings": 2048,
14
+ "max_seq_length": 64,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 8,
18
+ "num_hidden_layers": 12,
19
+ "num_key_value_heads": 8,
20
+ "pretraining_tp": 1,
21
+ "residual_in_fp32": true,
22
+ "rms_norm_eps": 1e-06,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "transformers_version": "4.44.2",
27
+ "use_cache": true,
28
+ "use_flash_attn": true,
29
+ "vocab_size": 500
30
+ }
tmp-spec-checkpoint-35000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7a305030119b8388766aefb7faa161b37d78a0be1d4873231b6b9daa3171637
3
+ size 63988303
tmp-spec-checkpoint-35000/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<bos>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<eos>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<pad>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tmp-spec-checkpoint-35000/tokenizer.json ADDED
@@ -0,0 +1,1079 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "<unk>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "<pad>",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "<bos>",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "<eos>",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ }
42
+ ],
43
+ "normalizer": null,
44
+ "pre_tokenizer": null,
45
+ "post_processor": {
46
+ "type": "TemplateProcessing",
47
+ "single": [
48
+ {
49
+ "SpecialToken": {
50
+ "id": "<bos>",
51
+ "type_id": 0
52
+ }
53
+ },
54
+ {
55
+ "Sequence": {
56
+ "id": "A",
57
+ "type_id": 0
58
+ }
59
+ },
60
+ {
61
+ "SpecialToken": {
62
+ "id": "<eos>",
63
+ "type_id": 0
64
+ }
65
+ }
66
+ ],
67
+ "pair": [
68
+ {
69
+ "Sequence": {
70
+ "id": "A",
71
+ "type_id": 0
72
+ }
73
+ },
74
+ {
75
+ "Sequence": {
76
+ "id": "B",
77
+ "type_id": 1
78
+ }
79
+ }
80
+ ],
81
+ "special_tokens": {
82
+ "<bos>": {
83
+ "id": "<bos>",
84
+ "ids": [
85
+ 2
86
+ ],
87
+ "tokens": [
88
+ "<bos>"
89
+ ]
90
+ },
91
+ "<eos>": {
92
+ "id": "<eos>",
93
+ "ids": [
94
+ 3
95
+ ],
96
+ "tokens": [
97
+ "<eos>"
98
+ ]
99
+ }
100
+ }
101
+ },
102
+ "decoder": {
103
+ "type": "BPEDecoder",
104
+ "suffix": "</w>"
105
+ },
106
+ "model": {
107
+ "type": "BPE",
108
+ "dropout": 0.1,
109
+ "unk_token": "<unk>",
110
+ "continuing_subword_prefix": null,
111
+ "end_of_word_suffix": null,
112
+ "fuse_unk": false,
113
+ "byte_fallback": false,
114
+ "ignore_merges": false,
115
+ "vocab": {
116
+ "<unk>": 0,
117
+ "<pad>": 1,
118
+ "<bos>": 2,
119
+ "<eos>": 3,
120
+ "#": 4,
121
+ "%": 5,
122
+ "(": 6,
123
+ ")": 7,
124
+ "+": 8,
125
+ "-": 9,
126
+ ".": 10,
127
+ "0": 11,
128
+ "1": 12,
129
+ "2": 13,
130
+ "3": 14,
131
+ "4": 15,
132
+ "5": 16,
133
+ "6": 17,
134
+ "7": 18,
135
+ "8": 19,
136
+ "9": 20,
137
+ "=": 21,
138
+ "B": 22,
139
+ "C": 23,
140
+ "F": 24,
141
+ "H": 25,
142
+ "I": 26,
143
+ "N": 27,
144
+ "O": 28,
145
+ "P": 29,
146
+ "S": 30,
147
+ "[": 31,
148
+ "]": 32,
149
+ "c": 33,
150
+ "e": 34,
151
+ "i": 35,
152
+ "l": 36,
153
+ "n": 37,
154
+ "o": 38,
155
+ "r": 39,
156
+ "s": 40,
157
+ "CC": 41,
158
+ ".C": 42,
159
+ "=O": 43,
160
+ "cc": 44,
161
+ ".N": 45,
162
+ "c1": 46,
163
+ ")C": 47,
164
+ "(C": 48,
165
+ "1.C": 49,
166
+ "=O.C": 50,
167
+ "%1": 51,
168
+ "CCC": 52,
169
+ "(=O": 53,
170
+ ".c1": 54,
171
+ "%10": 55,
172
+ "CC1": 56,
173
+ ".N3": 57,
174
+ "(C)": 58,
175
+ "C1": 59,
176
+ ".N4": 60,
177
+ "(C)C": 61,
178
+ "(=O)C": 62,
179
+ ".C1": 63,
180
+ "(F": 64,
181
+ ".C4": 65,
182
+ ".O": 66,
183
+ "ccc": 67,
184
+ "(F)": 68,
185
+ ".CC": 69,
186
+ "N1": 70,
187
+ "(=O)": 71,
188
+ ".C5": 72,
189
+ "6C": 73,
190
+ ".C3": 74,
191
+ "cn": 75,
192
+ "CC2": 76,
193
+ "1.C1": 77,
194
+ "5C": 78,
195
+ "7C": 79,
196
+ ".N1": 80,
197
+ ".N5": 81,
198
+ "n1": 82,
199
+ ".N35": 83,
200
+ ".N46": 84,
201
+ "cccc": 85,
202
+ "CCN": 86,
203
+ ".N2": 87,
204
+ "(O": 88,
205
+ "1.C4": 89,
206
+ "8C": 90,
207
+ "%11": 91,
208
+ "7=O": 92,
209
+ "[n": 93,
210
+ "H]": 94,
211
+ "[nH]": 95,
212
+ "CCCC": 96,
213
+ "=O.C5": 97,
214
+ ".C6": 98,
215
+ "=O.C4": 99,
216
+ ".c17": 100,
217
+ "nc": 101,
218
+ "=O.C1": 102,
219
+ "4C": 103,
220
+ ".c18": 104,
221
+ "8=O": 105,
222
+ "=C": 106,
223
+ "cc1": 107,
224
+ "CCC1": 108,
225
+ "1.C5": 109,
226
+ "9C": 110,
227
+ ".c16": 111,
228
+ "(F)F": 112,
229
+ "6=O": 113,
230
+ "CCO": 114,
231
+ "(Cl": 115,
232
+ "c2": 116,
233
+ "1.C3": 117,
234
+ "(Cl)": 118,
235
+ ".N36": 119,
236
+ "N14": 120,
237
+ ".N45": 121,
238
+ "c17": 122,
239
+ "(=O)CC": 123,
240
+ "CCC2": 124,
241
+ "(O)": 125,
242
+ "(C)(C)C": 126,
243
+ "nn": 127,
244
+ "3=O": 128,
245
+ "Cc1": 129,
246
+ "=O.CO": 130,
247
+ "c18": 131,
248
+ ".C2": 132,
249
+ "(O)C": 133,
250
+ ".C7": 134,
251
+ ".c19": 135,
252
+ "c1C": 136,
253
+ ".N47": 137,
254
+ "N13": 138,
255
+ "=O.C6": 139,
256
+ ".O=C": 140,
257
+ ".N57": 141,
258
+ ".C17": 142,
259
+ "C2": 143,
260
+ "c16": 144,
261
+ "N)": 145,
262
+ ".N6": 146,
263
+ "8CC": 147,
264
+ "6CC": 148,
265
+ ".n1": 149,
266
+ ".N24": 150,
267
+ ".N56": 151,
268
+ "9=O": 152,
269
+ ".N14": 153,
270
+ ".CCC": 154,
271
+ "(F)C": 155,
272
+ "ccc2": 156,
273
+ "Br": 157,
274
+ "(C)CC": 158,
275
+ "5CC": 159,
276
+ ".N13": 160,
277
+ "4=O": 161,
278
+ "CN": 162,
279
+ ".C=": 163,
280
+ ".C4(=O)C": 164,
281
+ "ccccc1": 165,
282
+ ".N34": 166,
283
+ "7CC": 167,
284
+ "ncc": 168,
285
+ "56": 169,
286
+ "c(": 170,
287
+ "c19": 171,
288
+ ".S": 172,
289
+ "%12": 173,
290
+ ".C18": 174,
291
+ ".C3(=O)C": 175,
292
+ ".N37": 176,
293
+ "1.C4(=O)C": 177,
294
+ "N4": 178,
295
+ "9CC": 179,
296
+ ".C16": 180,
297
+ "(F)(F)F": 181,
298
+ "nc1": 182,
299
+ "%10C": 183,
300
+ "=O.C3": 184,
301
+ ".O=C3": 185,
302
+ "c(C)": 186,
303
+ ".O56": 187,
304
+ ".C6C": 188,
305
+ "cs": 189,
306
+ ".C47": 190,
307
+ ".c15": 191,
308
+ ".C58": 192,
309
+ ".N35.N46": 193,
310
+ ".C8": 194,
311
+ ".CC1": 195,
312
+ "(N)": 196,
313
+ ".N25": 197,
314
+ "5=O": 198,
315
+ "(CC": 199,
316
+ "3=O.C4": 200,
317
+ ".C5C": 201,
318
+ "Br)": 202,
319
+ ".CO": 203,
320
+ ".C47=O": 204,
321
+ ".Cc1": 205,
322
+ "ccn": 206,
323
+ "CC12": 207,
324
+ "67": 208,
325
+ ".c1%10": 209,
326
+ "3=O.C1": 210,
327
+ "CCN4": 211,
328
+ "CCCN": 212,
329
+ "cc2": 213,
330
+ "(C1": 214,
331
+ "1.C17": 215,
332
+ "c15": 216,
333
+ "C4": 217,
334
+ ")C1": 218,
335
+ ".C48": 219,
336
+ ".O4": 220,
337
+ "1.C6": 221,
338
+ "N15": 222,
339
+ "C3": 223,
340
+ ".N48": 224,
341
+ "=CC": 225,
342
+ "4=O.C5": 226,
343
+ "C5": 227,
344
+ ".C69": 228,
345
+ "1.C18": 229,
346
+ ".C37": 230,
347
+ ".C19": 231,
348
+ "=N": 232,
349
+ "N3": 233,
350
+ ".C7C": 234,
351
+ "1.C3(=O)C": 235,
352
+ ".CC5": 236,
353
+ "1.C5(=O)C": 237,
354
+ "C17": 238,
355
+ "cnc": 239,
356
+ ".N58": 240,
357
+ "cnn": 241,
358
+ "%10CC": 242,
359
+ "(Br)": 243,
360
+ "5C2": 244,
361
+ ".N24.N35": 245,
362
+ "[N": 246,
363
+ "(=O)C1": 247,
364
+ "1.Cc1": 248,
365
+ "nc2": 249,
366
+ ")CC": 250,
367
+ ")C2": 251,
368
+ "=O.CO5": 252,
369
+ ".C59": 253,
370
+ ".O45": 254,
371
+ ".C57": 255,
372
+ "C18": 256,
373
+ "3=O.C": 257,
374
+ ".O3": 258,
375
+ "6=O.N35": 259,
376
+ "C16": 260,
377
+ "(=O)N": 261,
378
+ "CCCO": 262,
379
+ "cc8": 263,
380
+ ".c17ccc": 264,
381
+ ".C36": 265,
382
+ "cc(C)": 266,
383
+ "C12": 267,
384
+ "c1Cl": 268,
385
+ "cc1C": 269,
386
+ "1.C16": 270,
387
+ "=7": 271,
388
+ "+]": 272,
389
+ "(C2": 273,
390
+ "-]": 274,
391
+ "cc(Cl)": 275,
392
+ "4C2": 276,
393
+ "=O.C7": 277,
394
+ "5CC1": 278,
395
+ ".C6%10": 279,
396
+ "nn1": 280,
397
+ ".N15": 281,
398
+ "c(F)": 282,
399
+ ".O67": 283,
400
+ "cc(F)": 284,
401
+ ".O5C": 285,
402
+ ".C9": 286,
403
+ ".C37=O": 287,
404
+ ".c18ccc": 288,
405
+ "c[nH]": 289,
406
+ ".N26": 290,
407
+ "[N+]": 291,
408
+ ".CC6": 292,
409
+ "c1%10": 293,
410
+ "(=O)CCC": 294,
411
+ "cn1": 295,
412
+ "=8": 296,
413
+ "cc7": 297,
414
+ "8=O.N46": 298,
415
+ "(C#": 299,
416
+ "1.C19": 300,
417
+ "no": 301,
418
+ ".N67": 302,
419
+ "=O.CO6": 303,
420
+ ".C68": 304,
421
+ "CCOC": 305,
422
+ "c(Cl)": 306,
423
+ "OC": 307,
424
+ "4=O.C1": 308,
425
+ ".N14CC": 309,
426
+ "5C1": 310,
427
+ ".N13CC": 311,
428
+ ".C46": 312,
429
+ "(F)(F)C": 313,
430
+ "n[nH]": 314,
431
+ "2CC": 315,
432
+ ".C8C": 316,
433
+ "CC17": 317,
434
+ "1.C2": 318,
435
+ ".C5(=O)C": 319,
436
+ "(C)C1": 320,
437
+ "cc9": 321,
438
+ ".c16ccc": 322,
439
+ ".CC8": 323,
440
+ "ccc1": 324,
441
+ ".CC4": 325,
442
+ ".C38": 326,
443
+ "CC1C": 327,
444
+ "CC2(": 328,
445
+ "S(=O)": 329,
446
+ "=O.C58": 330,
447
+ "[O": 331,
448
+ "[O-]": 332,
449
+ "6C2": 333,
450
+ "%10=O": 334,
451
+ "78": 335,
452
+ ".C48=O": 336,
453
+ "cccc1": 337,
454
+ "C=": 338,
455
+ "s1": 339,
456
+ "1.C=": 340,
457
+ ".CC9": 341,
458
+ ".C15": 342,
459
+ "(=O)=O": 343,
460
+ "co": 344,
461
+ "(C)O": 345,
462
+ "ccccc2": 346,
463
+ ".CC7": 347,
464
+ ".C36=O": 348,
465
+ ".C2(=O)C": 349,
466
+ "CC16": 350,
467
+ "C19": 351,
468
+ "#CC": 352,
469
+ "(=O)[O-]": 353,
470
+ "[N+](=O)[O-]": 354,
471
+ ".C7%10": 355,
472
+ "CC18": 356,
473
+ ".C37=O.C4": 357,
474
+ "=O.C47=O": 358,
475
+ "6C1": 359,
476
+ "c8": 360,
477
+ "ccc(F)": 361,
478
+ "1.C6C": 362,
479
+ "2C": 363,
480
+ "7C1": 364,
481
+ ".C1%10": 365,
482
+ "7=O.N46": 366,
483
+ "(C#N)": 367,
484
+ ".N35C": 368,
485
+ ".O4C": 369,
486
+ "1.C48": 370,
487
+ "c7": 371,
488
+ ".N36.N47": 372,
489
+ ".N14CCC": 373,
490
+ "(O)CC": 374,
491
+ "1.C47": 375,
492
+ ".N68": 376,
493
+ "5C6": 377,
494
+ "n1C": 378,
495
+ ".C49": 379,
496
+ "ncn": 380,
497
+ ".CCCC": 381,
498
+ ".O6C": 382,
499
+ "C6": 383,
500
+ "C15": 384,
501
+ "[nH]1": 385,
502
+ "CC1CN": 386,
503
+ "cccc2": 387,
504
+ "4=O.C": 388,
505
+ "nc(C)": 389,
506
+ "CCOCC": 390,
507
+ "ccc8": 391,
508
+ "N=": 392,
509
+ ".n18": 393,
510
+ ".C17CC": 394,
511
+ "CCN4C": 395,
512
+ "7(C)C": 396,
513
+ ".n17": 397,
514
+ "CCC2(": 398,
515
+ ".C79": 399,
516
+ ".N46C": 400,
517
+ "N14CC": 401,
518
+ "ccs": 402,
519
+ "N2": 403,
520
+ "ccc(Cl)": 404,
521
+ "=O.C8": 405,
522
+ "N4C": 406,
523
+ "1.C7C": 407,
524
+ "=O.CO4": 408,
525
+ "8C1": 409,
526
+ "CC2C": 410,
527
+ "(C)CCC": 411,
528
+ "N13CC": 412,
529
+ ".N38": 413,
530
+ "3=O.C4(=O)C": 414,
531
+ "ccc7": 415,
532
+ "%11C": 416,
533
+ ".N13CCC": 417,
534
+ "(N)=O": 418,
535
+ "%11CC": 419,
536
+ ".c14": 420,
537
+ "cc1F": 421,
538
+ "c1.C1": 422,
539
+ "1.C1%10": 423,
540
+ ".C48=O.C5": 424,
541
+ ".N16": 425,
542
+ "8CC1": 426,
543
+ "c14": 427,
544
+ "c2ccccc2": 428,
545
+ "1.C5C": 429,
546
+ "=O.C69": 430,
547
+ "9CC1": 431,
548
+ "CCN3": 432,
549
+ "O1": 433,
550
+ "7C2": 434,
551
+ ".C46=O": 435,
552
+ ".C9C": 436,
553
+ "N5C": 437,
554
+ ".c17cn": 438,
555
+ ".C38=O": 439,
556
+ "CC2CC": 440,
557
+ "(C1)": 441,
558
+ "cc1Cl": 442,
559
+ "1.CO": 443,
560
+ "c9": 444,
561
+ "NC": 445,
562
+ "#N": 446,
563
+ ".O=C4": 447,
564
+ "C3(=O)C": 448,
565
+ "(C)c1": 449,
566
+ "CCN5C": 450,
567
+ "1.C8C": 451,
568
+ "N4C1": 452,
569
+ ".CO5": 453,
570
+ ".c18cn": 454,
571
+ "7CC1": 455,
572
+ ".C18CC": 456,
573
+ ".C16CC": 457,
574
+ "o1": 458,
575
+ "[N+](=O)[O-])": 459,
576
+ "cc(Br)": 460,
577
+ ".CC(C)C": 461,
578
+ ".c1%11": 462,
579
+ "n2": 463,
580
+ "ccnc": 464,
581
+ ".N46.N57": 465,
582
+ ".C%10": 466,
583
+ "1.C7": 467,
584
+ "=6": 468,
585
+ "S(C)": 469,
586
+ "cc6": 470,
587
+ "1.C4(=O)CC": 471,
588
+ "c1F": 472,
589
+ "C#": 473,
590
+ ")C(C)C": 474,
591
+ ".C4(=O)CC": 475,
592
+ "ccc9": 476,
593
+ "CC19": 477,
594
+ "=O.C47": 478,
595
+ "1.C48=O.C5": 479,
596
+ "=O.CS": 480,
597
+ "N5": 481,
598
+ "N13CCC": 482,
599
+ "=O.C6C": 483,
600
+ "5=O.C6": 484,
601
+ "C17C": 485,
602
+ "c2c1": 486,
603
+ ".C4C": 487,
604
+ "=O.C9": 488,
605
+ "(O)C1": 489,
606
+ "c(O)": 490,
607
+ "CC1(C)C": 491,
608
+ ".N59": 492,
609
+ ".N36.N45": 493,
610
+ ".C3(=O)CC": 494,
611
+ "CC2)C": 495,
612
+ ".N15CC": 496,
613
+ "1.C37": 497,
614
+ ".N23": 498,
615
+ ".CC%10": 499
616
+ },
617
+ "merges": [
618
+ "C C",
619
+ ". C",
620
+ "= O",
621
+ "c c",
622
+ ". N",
623
+ "c 1",
624
+ ") C",
625
+ "( C",
626
+ "1 .C",
627
+ "=O .C",
628
+ "% 1",
629
+ "CC C",
630
+ "( =O",
631
+ ". c1",
632
+ "%1 0",
633
+ "CC 1",
634
+ ".N 3",
635
+ "(C )",
636
+ "C 1",
637
+ ".N 4",
638
+ "(C )C",
639
+ "(=O )C",
640
+ ".C 1",
641
+ "( F",
642
+ ".C 4",
643
+ ". O",
644
+ "cc c",
645
+ "(F )",
646
+ ". CC",
647
+ "N 1",
648
+ "(=O )",
649
+ ".C 5",
650
+ "6 C",
651
+ ".C 3",
652
+ "c n",
653
+ "CC 2",
654
+ "1.C 1",
655
+ "5 C",
656
+ "7 C",
657
+ ".N 1",
658
+ ".N 5",
659
+ "n 1",
660
+ ".N3 5",
661
+ ".N4 6",
662
+ "cc cc",
663
+ "CC N",
664
+ ".N 2",
665
+ "( O",
666
+ "1.C 4",
667
+ "8 C",
668
+ "%1 1",
669
+ "7 =O",
670
+ "[ n",
671
+ "H ]",
672
+ "[n H]",
673
+ "CC CC",
674
+ "=O.C 5",
675
+ ".C 6",
676
+ "=O.C 4",
677
+ ".c1 7",
678
+ "n c",
679
+ "=O.C 1",
680
+ "4 C",
681
+ ".c1 8",
682
+ "8 =O",
683
+ "= C",
684
+ "cc 1",
685
+ "CCC 1",
686
+ "1.C 5",
687
+ "9 C",
688
+ ".c1 6",
689
+ "(F) F",
690
+ "6 =O",
691
+ "CC O",
692
+ "(C l",
693
+ "c 2",
694
+ "1.C 3",
695
+ "(Cl )",
696
+ ".N3 6",
697
+ "N1 4",
698
+ ".N4 5",
699
+ "c1 7",
700
+ "(=O) CC",
701
+ "CCC 2",
702
+ "(O )",
703
+ "(C) (C)C",
704
+ "n n",
705
+ "3 =O",
706
+ "C c1",
707
+ "=O.C O",
708
+ "c1 8",
709
+ ".C 2",
710
+ "(O )C",
711
+ ".C 7",
712
+ ".c1 9",
713
+ "c1 C",
714
+ ".N4 7",
715
+ "N1 3",
716
+ "=O.C 6",
717
+ ".O =C",
718
+ ".N5 7",
719
+ ".C1 7",
720
+ "C 2",
721
+ "c1 6",
722
+ "N )",
723
+ ".N 6",
724
+ "8 CC",
725
+ "6 CC",
726
+ ". n1",
727
+ ".N2 4",
728
+ ".N5 6",
729
+ "9 =O",
730
+ ".N1 4",
731
+ ". CCC",
732
+ "(F )C",
733
+ "ccc 2",
734
+ "B r",
735
+ "(C) CC",
736
+ "5 CC",
737
+ ".N1 3",
738
+ "4 =O",
739
+ "C N",
740
+ ".C =",
741
+ ".C4 (=O)C",
742
+ "cccc c1",
743
+ ".N3 4",
744
+ "7 CC",
745
+ "n cc",
746
+ "5 6",
747
+ "c (",
748
+ "c1 9",
749
+ ". S",
750
+ "%1 2",
751
+ ".C1 8",
752
+ ".C3 (=O)C",
753
+ ".N3 7",
754
+ "1.C4 (=O)C",
755
+ "N 4",
756
+ "9 CC",
757
+ ".C1 6",
758
+ "(F) (F)F",
759
+ "n c1",
760
+ "%10 C",
761
+ "=O.C 3",
762
+ ".O=C 3",
763
+ "c (C)",
764
+ ".O 56",
765
+ ".C 6C",
766
+ "c s",
767
+ ".C4 7",
768
+ ".c1 5",
769
+ ".C5 8",
770
+ ".N35 .N46",
771
+ ".C 8",
772
+ ". CC1",
773
+ "( N)",
774
+ ".N2 5",
775
+ "5 =O",
776
+ "( CC",
777
+ "3 =O.C4",
778
+ ".C5 C",
779
+ "Br )",
780
+ ".C O",
781
+ ".C4 7=O",
782
+ ".C c1",
783
+ "cc n",
784
+ "CC1 2",
785
+ "6 7",
786
+ ".c1 %10",
787
+ "3 =O.C1",
788
+ "CCN 4",
789
+ "CCC N",
790
+ "cc 2",
791
+ "(C 1",
792
+ "1.C1 7",
793
+ "c1 5",
794
+ "C 4",
795
+ ")C 1",
796
+ ".C4 8",
797
+ ".O 4",
798
+ "1.C 6",
799
+ "N1 5",
800
+ "C 3",
801
+ ".N4 8",
802
+ "= CC",
803
+ "4 =O.C5",
804
+ "C 5",
805
+ ".C6 9",
806
+ "1.C1 8",
807
+ ".C3 7",
808
+ ".C1 9",
809
+ "= N",
810
+ "N 3",
811
+ ".C 7C",
812
+ "1.C3 (=O)C",
813
+ ".CC 5",
814
+ "1.C5 (=O)C",
815
+ "C1 7",
816
+ "cn c",
817
+ ".N5 8",
818
+ "cn n",
819
+ "%10 CC",
820
+ "( Br)",
821
+ "5C 2",
822
+ ".N24 .N35",
823
+ "[ N",
824
+ "(=O)C 1",
825
+ "1.C c1",
826
+ "nc 2",
827
+ ") CC",
828
+ ")C 2",
829
+ "=O.CO 5",
830
+ ".C5 9",
831
+ ".O4 5",
832
+ ".C5 7",
833
+ "C1 8",
834
+ "3 =O.C",
835
+ ".O 3",
836
+ "6=O .N35",
837
+ "C1 6",
838
+ "(=O) N",
839
+ "CCC O",
840
+ "cc 8",
841
+ ".c17 ccc",
842
+ ".C3 6",
843
+ "cc (C)",
844
+ "C1 2",
845
+ "c1C l",
846
+ "cc1 C",
847
+ "1.C1 6",
848
+ "= 7",
849
+ "+ ]",
850
+ "(C 2",
851
+ "- ]",
852
+ "cc (Cl)",
853
+ "4C 2",
854
+ "=O.C 7",
855
+ "5 CC1",
856
+ ".C6 %10",
857
+ "n n1",
858
+ ".N1 5",
859
+ "c (F)",
860
+ ".O 67",
861
+ "cc (F)",
862
+ ".O 5C",
863
+ ".C 9",
864
+ ".C3 7=O",
865
+ ".c18 ccc",
866
+ "c [nH]",
867
+ ".N2 6",
868
+ "[N +]",
869
+ ".CC 6",
870
+ "c1 %10",
871
+ "(=O) CCC",
872
+ "cn 1",
873
+ "= 8",
874
+ "cc 7",
875
+ "8=O .N46",
876
+ "(C #",
877
+ "1.C1 9",
878
+ "n o",
879
+ ".N6 7",
880
+ "=O.CO 6",
881
+ ".C6 8",
882
+ "CCO C",
883
+ "c (Cl)",
884
+ "O C",
885
+ "4 =O.C1",
886
+ ".N14 CC",
887
+ "5 C1",
888
+ ".N13 CC",
889
+ ".C4 6",
890
+ "(F) (F)C",
891
+ "n [nH]",
892
+ "2 CC",
893
+ ".C 8C",
894
+ "CC1 7",
895
+ "1.C 2",
896
+ ".C5 (=O)C",
897
+ "(C)C 1",
898
+ "cc 9",
899
+ ".c16 ccc",
900
+ ".CC 8",
901
+ "cc c1",
902
+ ".CC 4",
903
+ ".C3 8",
904
+ "CC1 C",
905
+ "CC2 (",
906
+ "S (=O)",
907
+ "=O.C5 8",
908
+ "[ O",
909
+ "[O -]",
910
+ "6C 2",
911
+ "%10 =O",
912
+ "7 8",
913
+ ".C4 8=O",
914
+ "cccc 1",
915
+ "C =",
916
+ "s 1",
917
+ "1.C =",
918
+ ".CC 9",
919
+ ".C1 5",
920
+ "(=O) =O",
921
+ "c o",
922
+ "(C) O",
923
+ "cc ccc2",
924
+ ".CC 7",
925
+ ".C3 6=O",
926
+ ".C2 (=O)C",
927
+ "CC1 6",
928
+ "C1 9",
929
+ "# CC",
930
+ "(=O) [O-]",
931
+ "[N+] (=O)[O-]",
932
+ ".C7 %10",
933
+ "CC1 8",
934
+ ".C37 =O.C4",
935
+ "=O.C4 7=O",
936
+ "6 C1",
937
+ "c 8",
938
+ "ccc (F)",
939
+ "1.C 6C",
940
+ "2 C",
941
+ "7 C1",
942
+ ".C1 %10",
943
+ "7=O .N46",
944
+ "(C# N)",
945
+ ".N3 5C",
946
+ ".O 4C",
947
+ "1.C4 8",
948
+ "c 7",
949
+ ".N36 .N47",
950
+ ".N14 CCC",
951
+ "(O) CC",
952
+ "1.C4 7",
953
+ ".N6 8",
954
+ "5C 6",
955
+ "n1 C",
956
+ ".C4 9",
957
+ "n cn",
958
+ ".CC CC",
959
+ ".O 6C",
960
+ "C 6",
961
+ "C1 5",
962
+ "[nH] 1",
963
+ "CC1 CN",
964
+ "cccc 2",
965
+ "4 =O.C",
966
+ "nc (C)",
967
+ "CCO CC",
968
+ "ccc 8",
969
+ "N =",
970
+ ".n1 8",
971
+ ".C17 CC",
972
+ "CCN 4C",
973
+ "7 (C)C",
974
+ ".n1 7",
975
+ "CCC2 (",
976
+ ".C7 9",
977
+ ".N4 6C",
978
+ "N14 CC",
979
+ "cc s",
980
+ "N 2",
981
+ "ccc (Cl)",
982
+ "=O.C 8",
983
+ "N 4C",
984
+ "1.C 7C",
985
+ "=O.CO 4",
986
+ "8 C1",
987
+ "CC2 C",
988
+ "(C) CCC",
989
+ "N13 CC",
990
+ ".N3 8",
991
+ "3=O.C4 (=O)C",
992
+ "ccc 7",
993
+ "%11 C",
994
+ ".N13 CCC",
995
+ "(N) =O",
996
+ "%11 CC",
997
+ ".c1 4",
998
+ "cc1 F",
999
+ "c1 .C1",
1000
+ "1.C1 %10",
1001
+ ".C48 =O.C5",
1002
+ ".N1 6",
1003
+ "8 CC1",
1004
+ "c1 4",
1005
+ "c2 ccccc2",
1006
+ "1.C 5C",
1007
+ "=O.C6 9",
1008
+ "9 CC1",
1009
+ "CCN 3",
1010
+ "O 1",
1011
+ "7C 2",
1012
+ ".C4 6=O",
1013
+ ".C 9C",
1014
+ "N 5C",
1015
+ ".c17 cn",
1016
+ ".C3 8=O",
1017
+ "CC2 CC",
1018
+ "(C1 )",
1019
+ "cc1C l",
1020
+ "1.C O",
1021
+ "c 9",
1022
+ "N C",
1023
+ "# N",
1024
+ ".O=C 4",
1025
+ "C3 (=O)C",
1026
+ "(C) c1",
1027
+ "CCN 5C",
1028
+ "1.C 8C",
1029
+ "N4 C1",
1030
+ ".CO 5",
1031
+ ".c18 cn",
1032
+ "7 CC1",
1033
+ ".C1 8CC",
1034
+ ".C1 6CC",
1035
+ "o 1",
1036
+ "[N+](=O)[O-] )",
1037
+ "cc (Br)",
1038
+ ".CC (C)C",
1039
+ ".c1 %11",
1040
+ "n 2",
1041
+ "cc nc",
1042
+ ".N46 .N57",
1043
+ ".C %10",
1044
+ "1.C 7",
1045
+ "= 6",
1046
+ "S (C)",
1047
+ "cc 6",
1048
+ "1.C4 (=O)CC",
1049
+ "c1 F",
1050
+ "C #",
1051
+ ")C (C)C",
1052
+ ".C4 (=O)CC",
1053
+ "ccc 9",
1054
+ "CC1 9",
1055
+ "=O.C4 7",
1056
+ "1.C48 =O.C5",
1057
+ "=O.C S",
1058
+ "N 5",
1059
+ "N13 CCC",
1060
+ "=O.C 6C",
1061
+ "5 =O.C6",
1062
+ "C1 7C",
1063
+ "c2 c1",
1064
+ ".C4 C",
1065
+ "=O.C 9",
1066
+ "(O)C 1",
1067
+ "c (O)",
1068
+ "CC1 (C)C",
1069
+ ".N5 9",
1070
+ ".N36 .N45",
1071
+ ".C3 (=O)CC",
1072
+ "CC2 )C",
1073
+ ".N1 5CC",
1074
+ "1.C3 7",
1075
+ ".N2 3",
1076
+ ".CC %10"
1077
+ ]
1078
+ }
1079
+ }
tmp-spec-checkpoint-35000/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<unk>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<pad>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<bos>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<eos>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<bos>",
37
+ "clean_up_tokenization_spaces": true,
38
+ "eos_token": "<eos>",
39
+ "model_max_length": 1000000000000000019884624838656,
40
+ "pad_token": "<pad>",
41
+ "tokenizer_class": "PreTrainedTokenizerFast",
42
+ "unk_token": "<unk>"
43
+ }
tmp-spec-checkpoint-35000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4a2d9385265875a0ddfc74277788484fe87d0926d7a0785a0fcc25aec01805fc
3
+ size 6584