Xibanya commited on
Commit
6f8664c
·
1 Parent(s): 029ca94

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +1357 -0
trainer_state.json ADDED
@@ -0,0 +1,1357 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 3.459977149963379,
3
+ "best_model_checkpoint": "output/ds9_all\\checkpoint-8619",
4
+ "epoch": 30.0,
5
+ "global_step": 19890,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.15,
12
+ "learning_rate": 0.00013193067062605056,
13
+ "loss": 17.6929,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 0.3,
18
+ "learning_rate": 0.00011254053736637465,
19
+ "loss": 2.9597,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 0.45,
24
+ "learning_rate": 8.317464938377179e-05,
25
+ "loss": 2.8924,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 0.6,
30
+ "learning_rate": 5.04999004691124e-05,
31
+ "loss": 2.854,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 0.75,
36
+ "learning_rate": 2.1934390180170752e-05,
37
+ "loss": 2.7631,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 0.9,
42
+ "learning_rate": 3.963303752660306e-06,
43
+ "loss": 2.827,
44
+ "step": 600
45
+ },
46
+ {
47
+ "epoch": 1.06,
48
+ "learning_rate": 6.665901596802237e-07,
49
+ "loss": 2.7348,
50
+ "step": 700
51
+ },
52
+ {
53
+ "epoch": 1.21,
54
+ "learning_rate": 1.2792697375844372e-05,
55
+ "loss": 2.5285,
56
+ "step": 800
57
+ },
58
+ {
59
+ "epoch": 1.36,
60
+ "learning_rate": 3.758865334885973e-05,
61
+ "loss": 2.532,
62
+ "step": 900
63
+ },
64
+ {
65
+ "epoch": 1.51,
66
+ "learning_rate": 6.942506914238186e-05,
67
+ "loss": 2.5207,
68
+ "step": 1000
69
+ },
70
+ {
71
+ "epoch": 1.66,
72
+ "learning_rate": 0.00010107417071793643,
73
+ "loss": 2.5964,
74
+ "step": 1100
75
+ },
76
+ {
77
+ "epoch": 1.81,
78
+ "learning_rate": 0.00012535070970483918,
79
+ "loss": 2.651,
80
+ "step": 1200
81
+ },
82
+ {
83
+ "epoch": 1.96,
84
+ "learning_rate": 0.00013674321963112474,
85
+ "loss": 2.6082,
86
+ "step": 1300
87
+ },
88
+ {
89
+ "epoch": 2.11,
90
+ "learning_rate": 0.0001326652759440951,
91
+ "loss": 2.4929,
92
+ "step": 1400
93
+ },
94
+ {
95
+ "epoch": 2.26,
96
+ "learning_rate": 0.00011404268812251294,
97
+ "loss": 2.4281,
98
+ "step": 1500
99
+ },
100
+ {
101
+ "epoch": 2.41,
102
+ "learning_rate": 8.510331453017784e-05,
103
+ "loss": 2.4293,
104
+ "step": 1600
105
+ },
106
+ {
107
+ "epoch": 2.56,
108
+ "learning_rate": 5.2417218031482405e-05,
109
+ "loss": 2.4101,
110
+ "step": 1700
111
+ },
112
+ {
113
+ "epoch": 2.71,
114
+ "learning_rate": 2.3405074409295983e-05,
115
+ "loss": 2.3587,
116
+ "step": 1800
117
+ },
118
+ {
119
+ "epoch": 2.87,
120
+ "learning_rate": 4.653467399454465e-06,
121
+ "loss": 2.307,
122
+ "step": 1900
123
+ },
124
+ {
125
+ "epoch": 3.02,
126
+ "learning_rate": 4.1954639981762966e-07,
127
+ "loss": 2.3605,
128
+ "step": 2000
129
+ },
130
+ {
131
+ "epoch": 3.17,
132
+ "learning_rate": 1.1664532186009215e-05,
133
+ "loss": 2.137,
134
+ "step": 2100
135
+ },
136
+ {
137
+ "epoch": 3.32,
138
+ "learning_rate": 3.5835492391032385e-05,
139
+ "loss": 2.0627,
140
+ "step": 2200
141
+ },
142
+ {
143
+ "epoch": 3.47,
144
+ "learning_rate": 6.744492993645014e-05,
145
+ "loss": 2.1204,
146
+ "step": 2300
147
+ },
148
+ {
149
+ "epoch": 3.62,
150
+ "learning_rate": 9.931660131723702e-05,
151
+ "loss": 2.1823,
152
+ "step": 2400
153
+ },
154
+ {
155
+ "epoch": 3.77,
156
+ "learning_rate": 0.00012421472847147613,
157
+ "loss": 2.2277,
158
+ "step": 2500
159
+ },
160
+ {
161
+ "epoch": 3.92,
162
+ "learning_rate": 0.000136486726691794,
163
+ "loss": 2.2512,
164
+ "step": 2600
165
+ },
166
+ {
167
+ "epoch": 4.07,
168
+ "learning_rate": 0.00013334650250861535,
169
+ "loss": 2.0589,
170
+ "step": 2700
171
+ },
172
+ {
173
+ "epoch": 4.22,
174
+ "learning_rate": 0.00011550697633903009,
175
+ "loss": 2.0364,
176
+ "step": 2800
177
+ },
178
+ {
179
+ "epoch": 4.37,
180
+ "learning_rate": 8.701822922681755e-05,
181
+ "loss": 2.079,
182
+ "step": 2900
183
+ },
184
+ {
185
+ "epoch": 4.52,
186
+ "learning_rate": 5.4348018978061915e-05,
187
+ "loss": 2.0451,
188
+ "step": 3000
189
+ },
190
+ {
191
+ "epoch": 4.68,
192
+ "learning_rate": 2.4913414743980736e-05,
193
+ "loss": 2.0148,
194
+ "step": 3100
195
+ },
196
+ {
197
+ "epoch": 4.83,
198
+ "learning_rate": 5.396910863624203e-06,
199
+ "loss": 2.0322,
200
+ "step": 3200
201
+ },
202
+ {
203
+ "epoch": 4.98,
204
+ "learning_rate": 2.2931013159956905e-07,
205
+ "loss": 2.0037,
206
+ "step": 3300
207
+ },
208
+ {
209
+ "epoch": 5.13,
210
+ "learning_rate": 1.0583805241907789e-05,
211
+ "loss": 1.7858,
212
+ "step": 3400
213
+ },
214
+ {
215
+ "epoch": 5.28,
216
+ "learning_rate": 3.410963059878637e-05,
217
+ "loss": 1.7749,
218
+ "step": 3500
219
+ },
220
+ {
221
+ "epoch": 5.43,
222
+ "learning_rate": 6.546575312706341e-05,
223
+ "loss": 1.753,
224
+ "step": 3600
225
+ },
226
+ {
227
+ "epoch": 5.58,
228
+ "learning_rate": 9.753343905259311e-05,
229
+ "loss": 1.8774,
230
+ "step": 3700
231
+ },
232
+ {
233
+ "epoch": 5.73,
234
+ "learning_rate": 0.00012303240942344423,
235
+ "loss": 1.8664,
236
+ "step": 3800
237
+ },
238
+ {
239
+ "epoch": 5.88,
240
+ "learning_rate": 0.0001361736709920872,
241
+ "loss": 1.9253,
242
+ "step": 3900
243
+ },
244
+ {
245
+ "epoch": 6.03,
246
+ "learning_rate": 0.00013397378272625676,
247
+ "loss": 1.895,
248
+ "step": 4000
249
+ },
250
+ {
251
+ "epoch": 6.18,
252
+ "learning_rate": 0.00011693218198091279,
253
+ "loss": 1.6965,
254
+ "step": 4100
255
+ },
256
+ {
257
+ "epoch": 6.33,
258
+ "learning_rate": 8.891779797974934e-05,
259
+ "loss": 1.7647,
260
+ "step": 4200
261
+ },
262
+ {
263
+ "epoch": 6.49,
264
+ "learning_rate": 5.629069457859385e-05,
265
+ "loss": 1.7246,
266
+ "step": 4300
267
+ },
268
+ {
269
+ "epoch": 6.64,
270
+ "learning_rate": 2.645815444528567e-05,
271
+ "loss": 1.7252,
272
+ "step": 4400
273
+ },
274
+ {
275
+ "epoch": 6.79,
276
+ "learning_rate": 6.193014713115989e-06,
277
+ "loss": 1.7153,
278
+ "step": 4500
279
+ },
280
+ {
281
+ "epoch": 6.94,
282
+ "learning_rate": 9.603985859484616e-08,
283
+ "loss": 1.7007,
284
+ "step": 4600
285
+ },
286
+ {
287
+ "epoch": 7.09,
288
+ "learning_rate": 9.55141699790193e-06,
289
+ "loss": 1.5635,
290
+ "step": 4700
291
+ },
292
+ {
293
+ "epoch": 7.24,
294
+ "learning_rate": 3.2412505948464725e-05,
295
+ "loss": 1.4871,
296
+ "step": 4800
297
+ },
298
+ {
299
+ "epoch": 7.39,
300
+ "learning_rate": 6.348918775091967e-05,
301
+ "loss": 1.4965,
302
+ "step": 4900
303
+ },
304
+ {
305
+ "epoch": 7.54,
306
+ "learning_rate": 9.572616964271242e-05,
307
+ "loss": 1.5202,
308
+ "step": 5000
309
+ },
310
+ {
311
+ "epoch": 7.69,
312
+ "learning_rate": 0.00012180473766095787,
313
+ "loss": 1.5508,
314
+ "step": 5100
315
+ },
316
+ {
317
+ "epoch": 7.84,
318
+ "learning_rate": 0.0001358043133678906,
319
+ "loss": 1.658,
320
+ "step": 5200
321
+ },
322
+ {
323
+ "epoch": 7.99,
324
+ "learning_rate": 0.0001345465939513953,
325
+ "loss": 1.6614,
326
+ "step": 5300
327
+ },
328
+ {
329
+ "epoch": 8.14,
330
+ "learning_rate": 0.0001183171175764834,
331
+ "loss": 1.4421,
332
+ "step": 5400
333
+ },
334
+ {
335
+ "epoch": 8.3,
336
+ "learning_rate": 9.080043808116885e-05,
337
+ "loss": 1.4517,
338
+ "step": 5500
339
+ },
340
+ {
341
+ "epoch": 8.45,
342
+ "learning_rate": 5.824362620893997e-05,
343
+ "loss": 1.5004,
344
+ "step": 5600
345
+ },
346
+ {
347
+ "epoch": 8.6,
348
+ "learning_rate": 2.803800644656531e-05,
349
+ "loss": 1.4732,
350
+ "step": 5700
351
+ },
352
+ {
353
+ "epoch": 8.75,
354
+ "learning_rate": 7.041115639600712e-06,
355
+ "loss": 1.3825,
356
+ "step": 5800
357
+ },
358
+ {
359
+ "epoch": 8.9,
360
+ "learning_rate": 1.9846620691268856e-08,
361
+ "loss": 1.3723,
362
+ "step": 5900
363
+ },
364
+ {
365
+ "epoch": 9.05,
366
+ "learning_rate": 8.568227632876734e-06,
367
+ "loss": 1.3401,
368
+ "step": 6000
369
+ },
370
+ {
371
+ "epoch": 9.2,
372
+ "learning_rate": 3.074553247281859e-05,
373
+ "loss": 1.2335,
374
+ "step": 6100
375
+ },
376
+ {
377
+ "epoch": 9.35,
378
+ "learning_rate": 6.151688066888818e-05,
379
+ "loss": 1.211,
380
+ "step": 6200
381
+ },
382
+ {
383
+ "epoch": 9.5,
384
+ "learning_rate": 9.389629889221367e-05,
385
+ "loss": 1.2744,
386
+ "step": 6300
387
+ },
388
+ {
389
+ "epoch": 9.65,
390
+ "learning_rate": 0.00012053273607180507,
391
+ "loss": 1.3462,
392
+ "step": 6400
393
+ },
394
+ {
395
+ "epoch": 9.8,
396
+ "learning_rate": 0.0001353789615654724,
397
+ "loss": 1.3833,
398
+ "step": 6500
399
+ },
400
+ {
401
+ "epoch": 9.95,
402
+ "learning_rate": 0.00013506445892160198,
403
+ "loss": 1.383,
404
+ "step": 6600
405
+ },
406
+ {
407
+ "epoch": 10.11,
408
+ "learning_rate": 1.3445492880161939e-06,
409
+ "loss": 1.2022,
410
+ "step": 6700
411
+ },
412
+ {
413
+ "epoch": 10.26,
414
+ "learning_rate": 1.1808914090066666e-06,
415
+ "loss": 1.1142,
416
+ "step": 6800
417
+ },
418
+ {
419
+ "epoch": 10.41,
420
+ "learning_rate": 9.048790708431603e-07,
421
+ "loss": 1.1743,
422
+ "step": 6900
423
+ },
424
+ {
425
+ "epoch": 10.56,
426
+ "learning_rate": 5.791749433756745e-07,
427
+ "loss": 1.1411,
428
+ "step": 7000
429
+ },
430
+ {
431
+ "epoch": 10.71,
432
+ "learning_rate": 2.7772314933464675e-07,
433
+ "loss": 1.2239,
434
+ "step": 7100
435
+ },
436
+ {
437
+ "epoch": 10.86,
438
+ "learning_rate": 6.896184033086834e-08,
439
+ "loss": 1.1843,
440
+ "step": 7200
441
+ },
442
+ {
443
+ "epoch": 11.01,
444
+ "learning_rate": 2.8578527430069923e-10,
445
+ "loss": 1.1689,
446
+ "step": 7300
447
+ },
448
+ {
449
+ "epoch": 11.16,
450
+ "learning_rate": 8.568227632876734e-08,
451
+ "loss": 1.105,
452
+ "step": 7400
453
+ },
454
+ {
455
+ "epoch": 11.31,
456
+ "learning_rate": 3.074553247281864e-07,
457
+ "loss": 1.1371,
458
+ "step": 7500
459
+ },
460
+ {
461
+ "epoch": 11.46,
462
+ "learning_rate": 6.151688066888805e-07,
463
+ "loss": 1.1963,
464
+ "step": 7600
465
+ },
466
+ {
467
+ "epoch": 11.61,
468
+ "learning_rate": 9.389629889221367e-07,
469
+ "loss": 1.153,
470
+ "step": 7700
471
+ },
472
+ {
473
+ "epoch": 11.76,
474
+ "learning_rate": 1.2053273607180523e-06,
475
+ "loss": 1.1951,
476
+ "step": 7800
477
+ },
478
+ {
479
+ "epoch": 11.92,
480
+ "learning_rate": 1.353789615654724e-06,
481
+ "loss": 1.1274,
482
+ "step": 7900
483
+ },
484
+ {
485
+ "epoch": 12.07,
486
+ "learning_rate": 1.363372539869449e-07,
487
+ "loss": 1.1804,
488
+ "step": 8000
489
+ },
490
+ {
491
+ "epoch": 12.22,
492
+ "learning_rate": 1.2362930050095064e-07,
493
+ "loss": 1.1455,
494
+ "step": 8100
495
+ },
496
+ {
497
+ "epoch": 12.37,
498
+ "learning_rate": 9.842812692084315e-08,
499
+ "loss": 1.1767,
500
+ "step": 8200
501
+ },
502
+ {
503
+ "epoch": 12.52,
504
+ "learning_rate": 6.645511813249419e-08,
505
+ "loss": 1.147,
506
+ "step": 8300
507
+ },
508
+ {
509
+ "epoch": 12.67,
510
+ "learning_rate": 3.496905867845774e-08,
511
+ "loss": 1.114,
512
+ "step": 8400
513
+ },
514
+ {
515
+ "epoch": 12.82,
516
+ "learning_rate": 1.1118181719517309e-08,
517
+ "loss": 1.1379,
518
+ "step": 8500
519
+ },
520
+ {
521
+ "epoch": 12.97,
522
+ "learning_rate": 3.173163117047257e-10,
523
+ "loss": 1.1261,
524
+ "step": 8600
525
+ },
526
+ {
527
+ "epoch": 13.0,
528
+ "eval_loss": 3.459977149963379,
529
+ "eval_runtime": 0.891,
530
+ "eval_samples_per_second": 76.319,
531
+ "eval_steps_per_second": 76.319,
532
+ "step": 8619
533
+ },
534
+ {
535
+ "epoch": 13.12,
536
+ "learning_rate": 5.0185668342927494e-09,
537
+ "loss": 1.1202,
538
+ "step": 8700
539
+ },
540
+ {
541
+ "epoch": 13.27,
542
+ "learning_rate": 2.415461538642524e-08,
543
+ "loss": 1.1387,
544
+ "step": 8800
545
+ },
546
+ {
547
+ "epoch": 13.42,
548
+ "learning_rate": 5.338103337974863e-08,
549
+ "loss": 1.1257,
550
+ "step": 8900
551
+ },
552
+ {
553
+ "epoch": 13.57,
554
+ "learning_rate": 8.606259068723572e-08,
555
+ "loss": 1.1419,
556
+ "step": 9000
557
+ },
558
+ {
559
+ "epoch": 13.73,
560
+ "learning_rate": 1.1477964205183373e-07,
561
+ "loss": 1.1491,
562
+ "step": 9100
563
+ },
564
+ {
565
+ "epoch": 13.88,
566
+ "learning_rate": 1.330125980932328e-07,
567
+ "loss": 1.141,
568
+ "step": 9200
569
+ },
570
+ {
571
+ "epoch": 14.0,
572
+ "eval_loss": 3.4634451866149902,
573
+ "eval_runtime": 0.8581,
574
+ "eval_samples_per_second": 79.247,
575
+ "eval_steps_per_second": 79.247,
576
+ "step": 9282
577
+ },
578
+ {
579
+ "epoch": 14.03,
580
+ "learning_rate": 1.3662205797010195e-07,
581
+ "loss": 1.1135,
582
+ "step": 9300
583
+ },
584
+ {
585
+ "epoch": 14.18,
586
+ "learning_rate": 1.2497725716841956e-07,
587
+ "loss": 1.1462,
588
+ "step": 9400
589
+ },
590
+ {
591
+ "epoch": 14.33,
592
+ "learning_rate": 1.0049124874956241e-07,
593
+ "loss": 1.1353,
594
+ "step": 9500
595
+ },
596
+ {
597
+ "epoch": 14.48,
598
+ "learning_rate": 6.876501764792728e-08,
599
+ "loss": 1.1573,
600
+ "step": 9600
601
+ },
602
+ {
603
+ "epoch": 14.63,
604
+ "learning_rate": 3.700132283557427e-08,
605
+ "loss": 1.1477,
606
+ "step": 9700
607
+ },
608
+ {
609
+ "epoch": 14.78,
610
+ "learning_rate": 1.2411428614983945e-08,
611
+ "loss": 1.0973,
612
+ "step": 9800
613
+ },
614
+ {
615
+ "epoch": 14.93,
616
+ "learning_rate": 5.779420298980625e-10,
617
+ "loss": 1.1278,
618
+ "step": 9900
619
+ },
620
+ {
621
+ "epoch": 15.0,
622
+ "eval_loss": 3.4664907455444336,
623
+ "eval_runtime": 0.883,
624
+ "eval_samples_per_second": 77.009,
625
+ "eval_steps_per_second": 77.009,
626
+ "step": 9945
627
+ },
628
+ {
629
+ "epoch": 15.08,
630
+ "learning_rate": 4.187401906767171e-09,
631
+ "loss": 1.125,
632
+ "step": 10000
633
+ },
634
+ {
635
+ "epoch": 15.23,
636
+ "learning_rate": 2.2420357948166246e-08,
637
+ "loss": 1.1384,
638
+ "step": 10100
639
+ },
640
+ {
641
+ "epoch": 15.38,
642
+ "learning_rate": 5.1137409312764306e-08,
643
+ "loss": 1.1209,
644
+ "step": 10200
645
+ },
646
+ {
647
+ "epoch": 15.54,
648
+ "learning_rate": 8.381896662025135e-08,
649
+ "loss": 1.1044,
650
+ "step": 10300
651
+ },
652
+ {
653
+ "epoch": 15.69,
654
+ "learning_rate": 1.1304538461357478e-07,
655
+ "loss": 1.1637,
656
+ "step": 10400
657
+ },
658
+ {
659
+ "epoch": 15.84,
660
+ "learning_rate": 1.3218143316570722e-07,
661
+ "loss": 1.1661,
662
+ "step": 10500
663
+ },
664
+ {
665
+ "epoch": 15.99,
666
+ "learning_rate": 1.368826836882953e-07,
667
+ "loss": 1.1183,
668
+ "step": 10600
669
+ },
670
+ {
671
+ "epoch": 16.0,
672
+ "eval_loss": 3.4696619510650635,
673
+ "eval_runtime": 0.872,
674
+ "eval_samples_per_second": 77.979,
675
+ "eval_steps_per_second": 77.979,
676
+ "step": 10608
677
+ },
678
+ {
679
+ "epoch": 16.14,
680
+ "learning_rate": 1.2608181828048274e-07,
681
+ "loss": 1.1561,
682
+ "step": 10700
683
+ },
684
+ {
685
+ "epoch": 16.29,
686
+ "learning_rate": 1.0223094132154226e-07,
687
+ "loss": 1.1467,
688
+ "step": 10800
689
+ },
690
+ {
691
+ "epoch": 16.44,
692
+ "learning_rate": 7.074488186750591e-08,
693
+ "loss": 1.1874,
694
+ "step": 10900
695
+ },
696
+ {
697
+ "epoch": 16.59,
698
+ "learning_rate": 3.8771873079156884e-08,
699
+ "loss": 1.1281,
700
+ "step": 11000
701
+ },
702
+ {
703
+ "epoch": 16.74,
704
+ "learning_rate": 1.3570699499049336e-08,
705
+ "loss": 1.0814,
706
+ "step": 11100
707
+ },
708
+ {
709
+ "epoch": 16.89,
710
+ "learning_rate": 8.627460130551071e-10,
711
+ "loss": 1.1048,
712
+ "step": 11200
713
+ },
714
+ {
715
+ "epoch": 17.0,
716
+ "eval_loss": 3.4714086055755615,
717
+ "eval_runtime": 0.874,
718
+ "eval_samples_per_second": 77.801,
719
+ "eval_steps_per_second": 77.801,
720
+ "step": 11271
721
+ },
722
+ {
723
+ "epoch": 17.04,
724
+ "learning_rate": 3.5330803651032965e-09,
725
+ "loss": 1.1379,
726
+ "step": 11300
727
+ },
728
+ {
729
+ "epoch": 17.19,
730
+ "learning_rate": 2.0975460526287258e-08,
731
+ "loss": 1.1625,
732
+ "step": 11400
733
+ },
734
+ {
735
+ "epoch": 17.35,
736
+ "learning_rate": 4.922996891933192e-08,
737
+ "loss": 1.1556,
738
+ "step": 11500
739
+ },
740
+ {
741
+ "epoch": 17.5,
742
+ "learning_rate": 8.188202660552875e-08,
743
+ "loss": 1.1082,
744
+ "step": 11600
745
+ },
746
+ {
747
+ "epoch": 17.65,
748
+ "learning_rate": 1.115186855828889e-07,
749
+ "loss": 1.1591,
750
+ "step": 11700
751
+ },
752
+ {
753
+ "epoch": 17.8,
754
+ "learning_rate": 1.3141157932507567e-07,
755
+ "loss": 1.119,
756
+ "step": 11800
757
+ },
758
+ {
759
+ "epoch": 17.95,
760
+ "learning_rate": 1.3704445380800497e-07,
761
+ "loss": 1.1061,
762
+ "step": 11900
763
+ },
764
+ {
765
+ "epoch": 18.0,
766
+ "eval_loss": 3.4751551151275635,
767
+ "eval_runtime": 0.87,
768
+ "eval_samples_per_second": 78.161,
769
+ "eval_steps_per_second": 78.161,
770
+ "step": 11934
771
+ },
772
+ {
773
+ "epoch": 18.1,
774
+ "learning_rate": 1.2713848593130804e-07,
775
+ "loss": 1.1182,
776
+ "step": 12000
777
+ },
778
+ {
779
+ "epoch": 18.25,
780
+ "learning_rate": 1.0394261282124642e-07,
781
+ "loss": 1.1469,
782
+ "step": 12100
783
+ },
784
+ {
785
+ "epoch": 18.4,
786
+ "learning_rate": 7.272295898605559e-08,
787
+ "loss": 1.1398,
788
+ "step": 12200
789
+ },
790
+ {
791
+ "epoch": 18.55,
792
+ "learning_rate": 4.05672759158415e-08,
793
+ "loss": 1.0718,
794
+ "step": 12300
795
+ },
796
+ {
797
+ "epoch": 18.7,
798
+ "learning_rate": 1.4775820423167982e-08,
799
+ "loss": 1.1535,
800
+ "step": 12400
801
+ },
802
+ {
803
+ "epoch": 18.85,
804
+ "learning_rate": 1.2039882169440381e-09,
805
+ "loss": 1.1471,
806
+ "step": 12500
807
+ },
808
+ {
809
+ "epoch": 19.0,
810
+ "eval_loss": 3.4772818088531494,
811
+ "eval_runtime": 0.871,
812
+ "eval_samples_per_second": 78.07,
813
+ "eval_steps_per_second": 78.07,
814
+ "step": 12597
815
+ },
816
+ {
817
+ "epoch": 19.0,
818
+ "learning_rate": 2.9329721396952097e-09,
819
+ "loss": 1.1334,
820
+ "step": 12600
821
+ },
822
+ {
823
+ "epoch": 19.16,
824
+ "learning_rate": 1.9570243547781353e-08,
825
+ "loss": 1.1311,
826
+ "step": 12700
827
+ },
828
+ {
829
+ "epoch": 19.31,
830
+ "learning_rate": 4.733866750445838e-08,
831
+ "loss": 1.129,
832
+ "step": 12800
833
+ },
834
+ {
835
+ "epoch": 19.46,
836
+ "learning_rate": 7.993402009629649e-08,
837
+ "loss": 1.0708,
838
+ "step": 12900
839
+ },
840
+ {
841
+ "epoch": 19.61,
842
+ "learning_rate": 1.0995622699438409e-07,
843
+ "loss": 1.1266,
844
+ "step": 13000
845
+ },
846
+ {
847
+ "epoch": 19.76,
848
+ "learning_rate": 1.3058939130244216e-07,
849
+ "loss": 1.146,
850
+ "step": 13100
851
+ },
852
+ {
853
+ "epoch": 19.91,
854
+ "learning_rate": 1.3716110242414843e-07,
855
+ "loss": 1.1402,
856
+ "step": 13200
857
+ },
858
+ {
859
+ "epoch": 20.0,
860
+ "eval_loss": 3.4797611236572266,
861
+ "eval_runtime": 0.848,
862
+ "eval_samples_per_second": 80.192,
863
+ "eval_steps_per_second": 80.192,
864
+ "step": 13260
865
+ },
866
+ {
867
+ "epoch": 20.06,
868
+ "learning_rate": 1.2830956053906914e-07,
869
+ "loss": 1.1516,
870
+ "step": 13300
871
+ },
872
+ {
873
+ "epoch": 20.21,
874
+ "learning_rate": 1.0590224609743171e-07,
875
+ "loss": 1.1088,
876
+ "step": 13400
877
+ },
878
+ {
879
+ "epoch": 20.36,
880
+ "learning_rate": 7.502625830800328e-08,
881
+ "loss": 1.1331,
882
+ "step": 13500
883
+ },
884
+ {
885
+ "epoch": 20.51,
886
+ "learning_rate": 4.269132667061271e-08,
887
+ "loss": 1.157,
888
+ "step": 13600
889
+ },
890
+ {
891
+ "epoch": 20.66,
892
+ "learning_rate": 1.6238402532934746e-08,
893
+ "loss": 1.1155,
894
+ "step": 13700
895
+ },
896
+ {
897
+ "epoch": 20.81,
898
+ "learning_rate": 1.673053844704988e-09,
899
+ "loss": 1.1693,
900
+ "step": 13800
901
+ },
902
+ {
903
+ "epoch": 20.97,
904
+ "learning_rate": 2.3020300149297553e-09,
905
+ "loss": 1.0847,
906
+ "step": 13900
907
+ },
908
+ {
909
+ "epoch": 21.0,
910
+ "eval_loss": 3.4811391830444336,
911
+ "eval_runtime": 0.8641,
912
+ "eval_samples_per_second": 78.699,
913
+ "eval_steps_per_second": 78.699,
914
+ "step": 13923
915
+ },
916
+ {
917
+ "epoch": 21.12,
918
+ "learning_rate": 1.7982535520914407e-08,
919
+ "loss": 1.1415,
920
+ "step": 14000
921
+ },
922
+ {
923
+ "epoch": 21.27,
924
+ "learning_rate": 4.515464856300149e-08,
925
+ "loss": 1.1699,
926
+ "step": 14100
927
+ },
928
+ {
929
+ "epoch": 21.42,
930
+ "learning_rate": 7.764952480614377e-08,
931
+ "loss": 1.0972,
932
+ "step": 14200
933
+ },
934
+ {
935
+ "epoch": 21.57,
936
+ "learning_rate": 1.0808990091715443e-07,
937
+ "loss": 1.1351,
938
+ "step": 14300
939
+ },
940
+ {
941
+ "epoch": 21.72,
942
+ "learning_rate": 1.295649436664554e-07,
943
+ "loss": 1.1021,
944
+ "step": 14400
945
+ },
946
+ {
947
+ "epoch": 21.87,
948
+ "learning_rate": 1.3719920609842196e-07,
949
+ "loss": 1.1462,
950
+ "step": 14500
951
+ },
952
+ {
953
+ "epoch": 22.0,
954
+ "eval_loss": 3.48413348197937,
955
+ "eval_runtime": 0.887,
956
+ "eval_samples_per_second": 76.663,
957
+ "eval_steps_per_second": 76.663,
958
+ "step": 14586
959
+ },
960
+ {
961
+ "epoch": 22.02,
962
+ "learning_rate": 1.292594929888637e-07,
963
+ "loss": 1.1323,
964
+ "step": 14600
965
+ },
966
+ {
967
+ "epoch": 22.17,
968
+ "learning_rate": 1.0754834557415543e-07,
969
+ "loss": 1.1255,
970
+ "step": 14700
971
+ },
972
+ {
973
+ "epoch": 22.32,
974
+ "learning_rate": 7.699481330029191e-08,
975
+ "loss": 1.1254,
976
+ "step": 14800
977
+ },
978
+ {
979
+ "epoch": 22.47,
980
+ "learning_rate": 4.453541907189201e-08,
981
+ "loss": 1.1578,
982
+ "step": 14900
983
+ },
984
+ {
985
+ "epoch": 22.62,
986
+ "learning_rate": 1.7539370793205484e-08,
987
+ "loss": 1.1509,
988
+ "step": 15000
989
+ },
990
+ {
991
+ "epoch": 22.78,
992
+ "learning_rate": 2.135541078398113e-09,
993
+ "loss": 1.096,
994
+ "step": 15100
995
+ },
996
+ {
997
+ "epoch": 22.93,
998
+ "learning_rate": 1.821038434527568e-09,
999
+ "loss": 1.1107,
1000
+ "step": 15200
1001
+ },
1002
+ {
1003
+ "epoch": 23.0,
1004
+ "eval_loss": 3.485179901123047,
1005
+ "eval_runtime": 0.876,
1006
+ "eval_samples_per_second": 77.624,
1007
+ "eval_steps_per_second": 77.624,
1008
+ "step": 15249
1009
+ },
1010
+ {
1011
+ "epoch": 23.08,
1012
+ "learning_rate": 1.6667263928194815e-08,
1013
+ "loss": 1.1112,
1014
+ "step": 15300
1015
+ },
1016
+ {
1017
+ "epoch": 23.23,
1018
+ "learning_rate": 4.330370110778617e-08,
1019
+ "loss": 1.0894,
1020
+ "step": 15400
1021
+ },
1022
+ {
1023
+ "epoch": 23.38,
1024
+ "learning_rate": 7.568311933111189e-08,
1025
+ "loss": 1.1571,
1026
+ "step": 15500
1027
+ },
1028
+ {
1029
+ "epoch": 23.53,
1030
+ "learning_rate": 1.0645446752718106e-07,
1031
+ "loss": 1.0807,
1032
+ "step": 15600
1033
+ },
1034
+ {
1035
+ "epoch": 23.68,
1036
+ "learning_rate": 1.286317723671232e-07,
1037
+ "loss": 1.1122,
1038
+ "step": 15700
1039
+ },
1040
+ {
1041
+ "epoch": 23.83,
1042
+ "learning_rate": 1.3718015337930875e-07,
1043
+ "loss": 1.2016,
1044
+ "step": 15800
1045
+ },
1046
+ {
1047
+ "epoch": 23.98,
1048
+ "learning_rate": 1.301588843603994e-07,
1049
+ "loss": 1.1192,
1050
+ "step": 15900
1051
+ },
1052
+ {
1053
+ "epoch": 24.0,
1054
+ "eval_loss": 3.487265110015869,
1055
+ "eval_runtime": 1.004,
1056
+ "eval_samples_per_second": 67.732,
1057
+ "eval_steps_per_second": 67.732,
1058
+ "step": 15912
1059
+ },
1060
+ {
1061
+ "epoch": 24.13,
1062
+ "learning_rate": 1.0916199355343483e-07,
1063
+ "loss": 1.1713,
1064
+ "step": 16000
1065
+ },
1066
+ {
1067
+ "epoch": 24.28,
1068
+ "learning_rate": 7.895637379105998e-08,
1069
+ "loss": 1.1412,
1070
+ "step": 16100
1071
+ },
1072
+ {
1073
+ "epoch": 24.43,
1074
+ "learning_rate": 4.639956191883156e-08,
1075
+ "loss": 1.1021,
1076
+ "step": 16200
1077
+ },
1078
+ {
1079
+ "epoch": 24.59,
1080
+ "learning_rate": 1.8882882423516715e-08,
1081
+ "loss": 1.1353,
1082
+ "step": 16300
1083
+ },
1084
+ {
1085
+ "epoch": 24.74,
1086
+ "learning_rate": 2.653406048604756e-09,
1087
+ "loss": 1.0801,
1088
+ "step": 16400
1089
+ },
1090
+ {
1091
+ "epoch": 24.89,
1092
+ "learning_rate": 1.3956866321093081e-09,
1093
+ "loss": 1.0868,
1094
+ "step": 16500
1095
+ },
1096
+ {
1097
+ "epoch": 25.0,
1098
+ "eval_loss": 3.487858295440674,
1099
+ "eval_runtime": 0.854,
1100
+ "eval_samples_per_second": 79.625,
1101
+ "eval_steps_per_second": 79.625,
1102
+ "step": 16575
1103
+ },
1104
+ {
1105
+ "epoch": 25.04,
1106
+ "learning_rate": 1.539526233904202e-08,
1107
+ "loss": 1.0812,
1108
+ "step": 16600
1109
+ },
1110
+ {
1111
+ "epoch": 25.19,
1112
+ "learning_rate": 4.1473830357287196e-08,
1113
+ "loss": 1.104,
1114
+ "step": 16700
1115
+ },
1116
+ {
1117
+ "epoch": 25.34,
1118
+ "learning_rate": 7.371081224908017e-08,
1119
+ "loss": 1.0943,
1120
+ "step": 16800
1121
+ },
1122
+ {
1123
+ "epoch": 25.49,
1124
+ "learning_rate": 1.0478749405153535e-07,
1125
+ "loss": 1.097,
1126
+ "step": 16900
1127
+ },
1128
+ {
1129
+ "epoch": 25.64,
1130
+ "learning_rate": 1.2764858300209813e-07,
1131
+ "loss": 1.1316,
1132
+ "step": 17000
1133
+ },
1134
+ {
1135
+ "epoch": 25.79,
1136
+ "learning_rate": 1.3710396014140516e-07,
1137
+ "loss": 1.1679,
1138
+ "step": 17100
1139
+ },
1140
+ {
1141
+ "epoch": 25.94,
1142
+ "learning_rate": 1.31006985286884e-07,
1143
+ "loss": 1.1313,
1144
+ "step": 17200
1145
+ },
1146
+ {
1147
+ "epoch": 26.0,
1148
+ "eval_loss": 3.489774703979492,
1149
+ "eval_runtime": 0.859,
1150
+ "eval_samples_per_second": 79.161,
1151
+ "eval_steps_per_second": 79.161,
1152
+ "step": 17238
1153
+ },
1154
+ {
1155
+ "epoch": 26.09,
1156
+ "learning_rate": 1.1100177565244e-07,
1157
+ "loss": 1.1454,
1158
+ "step": 17300
1159
+ },
1160
+ {
1161
+ "epoch": 26.24,
1162
+ "learning_rate": 8.123384073832201e-08,
1163
+ "loss": 1.0682,
1164
+ "step": 17400
1165
+ },
1166
+ {
1167
+ "epoch": 26.4,
1168
+ "learning_rate": 4.859766378694414e-08,
1169
+ "loss": 1.1414,
1170
+ "step": 17500
1171
+ },
1172
+ {
1173
+ "epoch": 26.55,
1174
+ "learning_rate": 2.050258742073295e-08,
1175
+ "loss": 1.146,
1176
+ "step": 17600
1177
+ },
1178
+ {
1179
+ "epoch": 26.7,
1180
+ "learning_rate": 3.326994962912756e-09,
1181
+ "loss": 1.145,
1182
+ "step": 17700
1183
+ },
1184
+ {
1185
+ "epoch": 26.85,
1186
+ "learning_rate": 9.702355760016704e-10,
1187
+ "loss": 1.1254,
1188
+ "step": 17800
1189
+ },
1190
+ {
1191
+ "epoch": 27.0,
1192
+ "learning_rate": 1.3967360831826574e-08,
1193
+ "loss": 1.1033,
1194
+ "step": 17900
1195
+ },
1196
+ {
1197
+ "epoch": 27.0,
1198
+ "eval_loss": 3.491461992263794,
1199
+ "eval_runtime": 0.878,
1200
+ "eval_samples_per_second": 77.447,
1201
+ "eval_steps_per_second": 77.447,
1202
+ "step": 17901
1203
+ },
1204
+ {
1205
+ "epoch": 27.15,
1206
+ "learning_rate": 3.936765275044558e-08,
1207
+ "loss": 1.1389,
1208
+ "step": 18000
1209
+ },
1210
+ {
1211
+ "epoch": 27.3,
1212
+ "learning_rate": 7.140452090629836e-08,
1213
+ "loss": 1.1195,
1214
+ "step": 18100
1215
+ },
1216
+ {
1217
+ "epoch": 27.45,
1218
+ "learning_rate": 1.0280468285978317e-07,
1219
+ "loss": 1.0567,
1220
+ "step": 18200
1221
+ },
1222
+ {
1223
+ "epoch": 27.6,
1224
+ "learning_rate": 1.2643940663167218e-07,
1225
+ "loss": 1.1261,
1226
+ "step": 18300
1227
+ },
1228
+ {
1229
+ "epoch": 27.75,
1230
+ "learning_rate": 1.369429361019015e-07,
1231
+ "loss": 1.1393,
1232
+ "step": 18400
1233
+ },
1234
+ {
1235
+ "epoch": 27.9,
1236
+ "learning_rate": 1.3193067062605075e-07,
1237
+ "loss": 1.1578,
1238
+ "step": 18500
1239
+ },
1240
+ {
1241
+ "epoch": 28.0,
1242
+ "eval_loss": 3.4938805103302,
1243
+ "eval_runtime": 0.864,
1244
+ "eval_samples_per_second": 78.704,
1245
+ "eval_steps_per_second": 78.704,
1246
+ "step": 18564
1247
+ },
1248
+ {
1249
+ "epoch": 28.05,
1250
+ "learning_rate": 1.1254053736637487e-07,
1251
+ "loss": 1.136,
1252
+ "step": 18600
1253
+ },
1254
+ {
1255
+ "epoch": 28.21,
1256
+ "learning_rate": 8.317464938377186e-08,
1257
+ "loss": 1.102,
1258
+ "step": 18700
1259
+ },
1260
+ {
1261
+ "epoch": 28.36,
1262
+ "learning_rate": 5.0499900469112783e-08,
1263
+ "loss": 1.1325,
1264
+ "step": 18800
1265
+ },
1266
+ {
1267
+ "epoch": 28.51,
1268
+ "learning_rate": 2.193439018017092e-08,
1269
+ "loss": 1.1511,
1270
+ "step": 18900
1271
+ },
1272
+ {
1273
+ "epoch": 28.66,
1274
+ "learning_rate": 3.963303752660322e-09,
1275
+ "loss": 1.118,
1276
+ "step": 19000
1277
+ },
1278
+ {
1279
+ "epoch": 28.81,
1280
+ "learning_rate": 6.66590159680239e-10,
1281
+ "loss": 1.1024,
1282
+ "step": 19100
1283
+ },
1284
+ {
1285
+ "epoch": 28.96,
1286
+ "learning_rate": 1.2792697375844275e-08,
1287
+ "loss": 1.0987,
1288
+ "step": 19200
1289
+ },
1290
+ {
1291
+ "epoch": 29.0,
1292
+ "eval_loss": 3.4947092533111572,
1293
+ "eval_runtime": 0.879,
1294
+ "eval_samples_per_second": 77.363,
1295
+ "eval_steps_per_second": 77.363,
1296
+ "step": 19227
1297
+ },
1298
+ {
1299
+ "epoch": 29.11,
1300
+ "learning_rate": 3.7294625797960914e-08,
1301
+ "loss": 1.135,
1302
+ "step": 19300
1303
+ },
1304
+ {
1305
+ "epoch": 29.26,
1306
+ "learning_rate": 6.909504912431355e-08,
1307
+ "loss": 1.1308,
1308
+ "step": 19400
1309
+ },
1310
+ {
1311
+ "epoch": 29.41,
1312
+ "learning_rate": 1.0078308218564023e-07,
1313
+ "loss": 1.0927,
1314
+ "step": 19500
1315
+ },
1316
+ {
1317
+ "epoch": 29.56,
1318
+ "learning_rate": 1.2516463805405508e-07,
1319
+ "loss": 1.1707,
1320
+ "step": 19600
1321
+ },
1322
+ {
1323
+ "epoch": 29.71,
1324
+ "learning_rate": 1.367044084314665e-07,
1325
+ "loss": 1.1217,
1326
+ "step": 19700
1327
+ },
1328
+ {
1329
+ "epoch": 29.86,
1330
+ "learning_rate": 1.3278253644438908e-07,
1331
+ "loss": 1.0779,
1332
+ "step": 19800
1333
+ },
1334
+ {
1335
+ "epoch": 30.0,
1336
+ "eval_loss": 3.497239112854004,
1337
+ "eval_runtime": 0.879,
1338
+ "eval_samples_per_second": 77.362,
1339
+ "eval_steps_per_second": 77.362,
1340
+ "step": 19890
1341
+ },
1342
+ {
1343
+ "epoch": 30.0,
1344
+ "step": 19890,
1345
+ "total_flos": 5197098516480000.0,
1346
+ "train_loss": 0.14960067772038202,
1347
+ "train_runtime": 189.9317,
1348
+ "train_samples_per_second": 104.722,
1349
+ "train_steps_per_second": 104.722
1350
+ }
1351
+ ],
1352
+ "max_steps": 19890,
1353
+ "num_train_epochs": 30,
1354
+ "total_flos": 5197098516480000.0,
1355
+ "trial_name": null,
1356
+ "trial_params": null
1357
+ }