msanocki commited on
Commit
60a8798
·
verified ·
1 Parent(s): 68f584b

Training in progress, step 500

Browse files
Files changed (4) hide show
  1. config.json +2 -2
  2. model.safetensors +2 -2
  3. tokenizer.json +350 -70
  4. training_args.bin +1 -1
config.json CHANGED
@@ -40,7 +40,7 @@
40
  "tie_word_embeddings": true,
41
  "type_vocab_size": 2,
42
  "use_cache": true,
43
- "vocab_size": 51
44
  },
45
  "decoder_start_token_id": 2,
46
  "dtype": "float32",
@@ -82,7 +82,7 @@
82
  "tie_word_embeddings": true,
83
  "type_vocab_size": 2,
84
  "use_cache": true,
85
- "vocab_size": 51
86
  },
87
  "eos_token_id": 0,
88
  "is_encoder_decoder": true,
 
40
  "tie_word_embeddings": true,
41
  "type_vocab_size": 2,
42
  "use_cache": true,
43
+ "vocab_size": 107
44
  },
45
  "decoder_start_token_id": 2,
46
  "dtype": "float32",
 
82
  "tie_word_embeddings": true,
83
  "type_vocab_size": 2,
84
  "use_cache": true,
85
+ "vocab_size": 107
86
  },
87
  "eos_token_id": 0,
88
  "is_encoder_decoder": true,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:31867b879e4d27c355e5e225ea575470ebea30fd083d5d9de88ff08fa8ecdfb1
3
- size 31203500
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b01d61ac36b350768b46e4b0a856c39f9b65661201332b29aa28ae00a2495b32
3
+ size 31318412
tokenizer.json CHANGED
@@ -114,130 +114,346 @@
114
  "8": 14,
115
  "9": 15,
116
  "=": 16,
117
- "10": 17,
118
- "98": 18,
119
- "99": 19,
120
- "97": 20,
121
- "96": 21,
122
- "11": 22,
123
- "12": 23,
124
- "95": 24,
125
- "94": 25,
126
- "13": 26,
127
- "14": 27,
128
- "15": 28,
129
- "16": 29,
130
- "92": 30,
131
- "93": 31,
132
- "91": 32,
133
- "17": 33,
134
- "90": 34,
135
- "19": 35,
136
- "18": 36,
137
- "89": 37,
138
- "20": 38,
139
- "88": 39,
140
- "21": 40,
141
- "22": 41,
142
- "85": 42,
143
- "23": 43,
144
- "84": 44,
145
- "86": 45,
146
- "87": 46,
147
- "24": 47,
148
- "25": 48,
149
- "26": 49,
150
- "82": 50
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
151
  },
152
  "merges": [
 
 
 
 
153
  [
154
  "1",
155
- "0"
156
  ],
157
  [
158
- "9",
 
 
 
 
159
  "8"
160
  ],
161
  [
162
- "9",
163
- "9"
164
  ],
165
  [
166
- "9",
167
- "7"
168
  ],
169
  [
170
- "9",
171
- "6"
 
 
 
 
172
  ],
173
  [
174
  "1",
 
 
 
 
 
 
 
 
 
 
 
 
175
  "1"
176
  ],
177
  [
178
- "1",
179
- "2"
180
  ],
181
  [
182
- "9",
 
 
 
 
 
 
 
 
183
  "5"
184
  ],
185
  [
186
- "9",
187
  "4"
188
  ],
189
  [
190
- "1",
191
- "3"
192
  ],
193
  [
194
- "1",
 
 
 
 
 
 
 
 
 
 
 
 
195
  "4"
196
  ],
197
  [
198
  "1",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
199
  "5"
200
  ],
 
 
 
 
 
 
 
 
201
  [
202
  "1",
203
- "6"
204
  ],
205
  [
206
  "9",
207
- "2"
208
  ],
209
  [
210
- "9",
211
- "3"
 
 
 
 
 
 
 
 
212
  ],
213
  [
214
  "9",
 
 
 
 
 
 
 
 
 
 
 
 
215
  "1"
216
  ],
217
  [
218
- "1",
 
 
 
 
 
 
 
 
219
  "7"
220
  ],
 
 
 
 
 
 
 
 
 
 
 
 
221
  [
222
  "9",
223
- "0"
224
  ],
225
  [
226
  "1",
227
- "9"
228
  ],
229
  [
230
  "1",
231
- "8"
232
  ],
233
  [
234
- "8",
 
 
 
 
235
  "9"
236
  ],
 
 
 
 
 
 
 
 
237
  [
238
  "2",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
239
  "0"
240
  ],
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
241
  [
242
  "8",
243
  "8"
@@ -246,45 +462,109 @@
246
  "2",
247
  "1"
248
  ],
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
249
  [
250
  "2",
 
 
 
 
251
  "2"
252
  ],
253
  [
254
- "8",
255
- "5"
 
 
 
 
256
  ],
257
  [
258
  "2",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
259
  "3"
260
  ],
261
  [
262
- "8",
263
- "4"
264
  ],
265
  [
266
- "8",
267
  "6"
268
  ],
269
  [
270
  "8",
271
- "7"
272
  ],
273
  [
274
- "2",
275
  "4"
276
  ],
277
  [
278
- "2",
 
 
 
 
279
  "5"
280
  ],
281
  [
282
- "2",
283
- "6"
284
  ],
285
  [
286
- "8",
287
- "2"
 
 
 
 
288
  ]
289
  ]
290
  }
 
114
  "8": 14,
115
  "9": 15,
116
  "=": 16,
117
+ "51": 17,
118
+ "16": 18,
119
+ "35": 19,
120
+ "28": 20,
121
+ "25": 21,
122
+ "12": 22,
123
+ "58": 23,
124
+ "22": 24,
125
+ "14": 25,
126
+ "20": 26,
127
+ "86": 27,
128
+ "31": 28,
129
+ "57": 29,
130
+ "53": 30,
131
+ "60": 31,
132
+ "65": 32,
133
+ "54": 33,
134
+ "70": 34,
135
+ "62": 35,
136
+ "32": 36,
137
+ "76": 37,
138
+ "24": 38,
139
+ "18": 39,
140
+ "46": 40,
141
+ "87": 41,
142
+ "73": 42,
143
+ "85": 43,
144
+ "50": 44,
145
+ "23": 45,
146
+ "15": 46,
147
+ "94": 47,
148
+ "68": 48,
149
+ "61": 49,
150
+ "49": 50,
151
+ "92": 51,
152
+ "36": 52,
153
+ "44": 53,
154
+ "81": 54,
155
+ "37": 55,
156
+ "39": 56,
157
+ "47": 57,
158
+ "71": 58,
159
+ "59": 59,
160
+ "66": 60,
161
+ "93": 61,
162
+ "11": 62,
163
+ "13": 63,
164
+ "41": 64,
165
+ "99": 65,
166
+ "75": 66,
167
+ "83": 67,
168
+ "27": 68,
169
+ "52": 69,
170
+ "84": 70,
171
+ "40": 71,
172
+ "78": 72,
173
+ "80": 73,
174
+ "10": 74,
175
+ "43": 75,
176
+ "95": 76,
177
+ "64": 77,
178
+ "74": 78,
179
+ "88": 79,
180
+ "21": 80,
181
+ "48": 81,
182
+ "55": 82,
183
+ "77": 83,
184
+ "89": 84,
185
+ "26": 85,
186
+ "42": 86,
187
+ "91": 87,
188
+ "79": 88,
189
+ "29": 89,
190
+ "19": 90,
191
+ "38": 91,
192
+ "67": 92,
193
+ "17": 93,
194
+ "30": 94,
195
+ "72": 95,
196
+ "96": 96,
197
+ "33": 97,
198
+ "97": 98,
199
+ "56": 99,
200
+ "82": 100,
201
+ "34": 101,
202
+ "63": 102,
203
+ "45": 103,
204
+ "69": 104,
205
+ "90": 105,
206
+ "98": 106
207
  },
208
  "merges": [
209
+ [
210
+ "5",
211
+ "1"
212
+ ],
213
  [
214
  "1",
215
+ "6"
216
  ],
217
  [
218
+ "3",
219
+ "5"
220
+ ],
221
+ [
222
+ "2",
223
  "8"
224
  ],
225
  [
226
+ "2",
227
+ "5"
228
  ],
229
  [
230
+ "1",
231
+ "2"
232
  ],
233
  [
234
+ "5",
235
+ "8"
236
+ ],
237
+ [
238
+ "2",
239
+ "2"
240
  ],
241
  [
242
  "1",
243
+ "4"
244
+ ],
245
+ [
246
+ "2",
247
+ "0"
248
+ ],
249
+ [
250
+ "8",
251
+ "6"
252
+ ],
253
+ [
254
+ "3",
255
  "1"
256
  ],
257
  [
258
+ "5",
259
+ "7"
260
  ],
261
  [
262
+ "5",
263
+ "3"
264
+ ],
265
+ [
266
+ "6",
267
+ "0"
268
+ ],
269
+ [
270
+ "6",
271
  "5"
272
  ],
273
  [
274
+ "5",
275
  "4"
276
  ],
277
  [
278
+ "7",
279
+ "0"
280
  ],
281
  [
282
+ "6",
283
+ "2"
284
+ ],
285
+ [
286
+ "3",
287
+ "2"
288
+ ],
289
+ [
290
+ "7",
291
+ "6"
292
+ ],
293
+ [
294
+ "2",
295
  "4"
296
  ],
297
  [
298
  "1",
299
+ "8"
300
+ ],
301
+ [
302
+ "4",
303
+ "6"
304
+ ],
305
+ [
306
+ "8",
307
+ "7"
308
+ ],
309
+ [
310
+ "7",
311
+ "3"
312
+ ],
313
+ [
314
+ "8",
315
  "5"
316
  ],
317
+ [
318
+ "5",
319
+ "0"
320
+ ],
321
+ [
322
+ "2",
323
+ "3"
324
+ ],
325
  [
326
  "1",
327
+ "5"
328
  ],
329
  [
330
  "9",
331
+ "4"
332
  ],
333
  [
334
+ "6",
335
+ "8"
336
+ ],
337
+ [
338
+ "6",
339
+ "1"
340
+ ],
341
+ [
342
+ "4",
343
+ "9"
344
  ],
345
  [
346
  "9",
347
+ "2"
348
+ ],
349
+ [
350
+ "3",
351
+ "6"
352
+ ],
353
+ [
354
+ "4",
355
+ "4"
356
+ ],
357
+ [
358
+ "8",
359
  "1"
360
  ],
361
  [
362
+ "3",
363
+ "7"
364
+ ],
365
+ [
366
+ "3",
367
+ "9"
368
+ ],
369
+ [
370
+ "4",
371
  "7"
372
  ],
373
+ [
374
+ "7",
375
+ "1"
376
+ ],
377
+ [
378
+ "5",
379
+ "9"
380
+ ],
381
+ [
382
+ "6",
383
+ "6"
384
+ ],
385
  [
386
  "9",
387
+ "3"
388
  ],
389
  [
390
  "1",
391
+ "1"
392
  ],
393
  [
394
  "1",
395
+ "3"
396
  ],
397
  [
398
+ "4",
399
+ "1"
400
+ ],
401
+ [
402
+ "9",
403
  "9"
404
  ],
405
+ [
406
+ "7",
407
+ "5"
408
+ ],
409
+ [
410
+ "8",
411
+ "3"
412
+ ],
413
  [
414
  "2",
415
+ "7"
416
+ ],
417
+ [
418
+ "5",
419
+ "2"
420
+ ],
421
+ [
422
+ "8",
423
+ "4"
424
+ ],
425
+ [
426
+ "4",
427
+ "0"
428
+ ],
429
+ [
430
+ "7",
431
+ "8"
432
+ ],
433
+ [
434
+ "8",
435
+ "0"
436
+ ],
437
+ [
438
+ "1",
439
  "0"
440
  ],
441
+ [
442
+ "4",
443
+ "3"
444
+ ],
445
+ [
446
+ "9",
447
+ "5"
448
+ ],
449
+ [
450
+ "6",
451
+ "4"
452
+ ],
453
+ [
454
+ "7",
455
+ "4"
456
+ ],
457
  [
458
  "8",
459
  "8"
 
462
  "2",
463
  "1"
464
  ],
465
+ [
466
+ "4",
467
+ "8"
468
+ ],
469
+ [
470
+ "5",
471
+ "5"
472
+ ],
473
+ [
474
+ "7",
475
+ "7"
476
+ ],
477
+ [
478
+ "8",
479
+ "9"
480
+ ],
481
  [
482
  "2",
483
+ "6"
484
+ ],
485
+ [
486
+ "4",
487
  "2"
488
  ],
489
  [
490
+ "9",
491
+ "1"
492
+ ],
493
+ [
494
+ "7",
495
+ "9"
496
  ],
497
  [
498
  "2",
499
+ "9"
500
+ ],
501
+ [
502
+ "1",
503
+ "9"
504
+ ],
505
+ [
506
+ "3",
507
+ "8"
508
+ ],
509
+ [
510
+ "6",
511
+ "7"
512
+ ],
513
+ [
514
+ "1",
515
+ "7"
516
+ ],
517
+ [
518
+ "3",
519
+ "0"
520
+ ],
521
+ [
522
+ "7",
523
+ "2"
524
+ ],
525
+ [
526
+ "9",
527
+ "6"
528
+ ],
529
+ [
530
+ "3",
531
  "3"
532
  ],
533
  [
534
+ "9",
535
+ "7"
536
  ],
537
  [
538
+ "5",
539
  "6"
540
  ],
541
  [
542
  "8",
543
+ "2"
544
  ],
545
  [
546
+ "3",
547
  "4"
548
  ],
549
  [
550
+ "6",
551
+ "3"
552
+ ],
553
+ [
554
+ "4",
555
  "5"
556
  ],
557
  [
558
+ "6",
559
+ "9"
560
  ],
561
  [
562
+ "9",
563
+ "0"
564
+ ],
565
+ [
566
+ "9",
567
+ "8"
568
  ]
569
  ]
570
  }
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:94b3002be46938dc49fc5b6a676604c9c87b4b8630a03c6f897cee4a3e54acfe
3
  size 5329
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41720a433a47f91c6331d71cd8db9723ab4af2380127620e11673d408948259a
3
  size 5329