terry69 commited on
Commit
2016948
·
verified ·
1 Parent(s): 2ffda52

Model save

Browse files
README.md CHANGED
@@ -1,14 +1,12 @@
1
  ---
2
  license: apache-2.0
3
- library_name: peft
4
  tags:
5
- - alignment-handbook
6
  - trl
7
  - sft
8
  - generated_from_trainer
9
- base_model: mistralai/Mistral-7B-Instruct-v0.2
10
  datasets:
11
- - preference-data
12
  model-index:
13
  - name: preference_p0.1_seed42_level2_style
14
  results: []
@@ -19,9 +17,9 @@ should probably proofread and complete it, then remove this comment. -->
19
 
20
  # preference_p0.1_seed42_level2_style
21
 
22
- This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the preference-data dataset.
23
  It achieves the following results on the evaluation set:
24
- - Loss: 0.9361
25
 
26
  ## Model description
27
 
@@ -41,13 +39,13 @@ More information needed
41
 
42
  The following hyperparameters were used during training:
43
  - learning_rate: 1e-05
44
- - train_batch_size: 8
45
  - eval_batch_size: 1
46
  - seed: 42
47
  - distributed_type: multi-GPU
48
  - num_devices: 4
49
- - gradient_accumulation_steps: 2
50
- - total_train_batch_size: 64
51
  - total_eval_batch_size: 4
52
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
53
  - lr_scheduler_type: cosine
@@ -58,13 +56,12 @@ The following hyperparameters were used during training:
58
 
59
  | Training Loss | Epoch | Step | Validation Loss |
60
  |:-------------:|:------:|:----:|:---------------:|
61
- | 0.9257 | 0.9991 | 551 | 0.9361 |
62
 
63
 
64
  ### Framework versions
65
 
66
- - PEFT 0.11.1
67
  - Transformers 4.43.4
68
  - Pytorch 2.3.1+cu121
69
  - Datasets 2.19.1
70
- - Tokenizers 0.19.1
 
1
  ---
2
  license: apache-2.0
3
+ base_model: mistralai/Mistral-7B-Instruct-v0.2
4
  tags:
 
5
  - trl
6
  - sft
7
  - generated_from_trainer
 
8
  datasets:
9
+ - generator
10
  model-index:
11
  - name: preference_p0.1_seed42_level2_style
12
  results: []
 
17
 
18
  # preference_p0.1_seed42_level2_style
19
 
20
+ This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the generator dataset.
21
  It achieves the following results on the evaluation set:
22
+ - Loss: 0.3182
23
 
24
  ## Model description
25
 
 
39
 
40
  The following hyperparameters were used during training:
41
  - learning_rate: 1e-05
42
+ - train_batch_size: 2
43
  - eval_batch_size: 1
44
  - seed: 42
45
  - distributed_type: multi-GPU
46
  - num_devices: 4
47
+ - gradient_accumulation_steps: 4
48
+ - total_train_batch_size: 32
49
  - total_eval_batch_size: 4
50
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
51
  - lr_scheduler_type: cosine
 
56
 
57
  | Training Loss | Epoch | Step | Validation Loss |
58
  |:-------------:|:------:|:----:|:---------------:|
59
+ | 0.3624 | 0.9994 | 1210 | 0.3182 |
60
 
61
 
62
  ### Framework versions
63
 
 
64
  - Transformers 4.43.4
65
  - Pytorch 2.3.1+cu121
66
  - Datasets 2.19.1
67
+ - Tokenizers 0.19.1
all_results.json CHANGED
@@ -1,14 +1,9 @@
1
  {
2
- "epoch": 0.99909338168631,
3
- "eval_loss": 0.9361373782157898,
4
- "eval_runtime": 1.9677,
5
- "eval_samples": 10,
6
- "eval_samples_per_second": 1.525,
7
- "eval_steps_per_second": 0.508,
8
- "total_flos": 2216760396742656.0,
9
- "train_loss": 0.9953486248715603,
10
- "train_runtime": 22334.5637,
11
- "train_samples": 89982,
12
- "train_samples_per_second": 1.579,
13
- "train_steps_per_second": 0.025
14
  }
 
1
  {
2
+ "epoch": 0.9993805492463349,
3
+ "total_flos": 253297038458880.0,
4
+ "train_loss": 0.5432274522860188,
5
+ "train_runtime": 27235.9072,
6
+ "train_samples": 98881,
7
+ "train_samples_per_second": 1.422,
8
+ "train_steps_per_second": 0.044
 
 
 
 
 
9
  }
runs/Sep05_22-24-45_COE-CS-sv003/events.out.tfevents.1725575476.COE-CS-sv003.17237.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:74f516ecffcfb261450cbd09b669e4609afab8d9a7c3dc9b045d5da3fbc81dec
3
- size 56672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e08629c9be20ebaf23c6359a27724f09d2f628450f8af817c1c7cd18c823873
3
+ size 57026
train_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
- "epoch": 0.99909338168631,
3
- "total_flos": 2216760396742656.0,
4
- "train_loss": 0.9953486248715603,
5
- "train_runtime": 22334.5637,
6
- "train_samples": 89982,
7
- "train_samples_per_second": 1.579,
8
- "train_steps_per_second": 0.025
9
  }
 
1
  {
2
+ "epoch": 0.9993805492463349,
3
+ "total_flos": 253297038458880.0,
4
+ "train_loss": 0.5432274522860188,
5
+ "train_runtime": 27235.9072,
6
+ "train_samples": 98881,
7
+ "train_samples_per_second": 1.422,
8
+ "train_steps_per_second": 0.044
9
  }
trainer_state.json CHANGED
@@ -1,810 +1,1734 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 0.99909338168631,
5
  "eval_steps": 500,
6
- "global_step": 551,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
- "epoch": 0.001813236627379873,
13
- "grad_norm": 1.608370220567919,
14
- "learning_rate": 1.7857142857142858e-07,
15
- "loss": 1.4522,
16
  "step": 1
17
  },
18
  {
19
- "epoch": 0.009066183136899365,
20
- "grad_norm": 1.5297359658025094,
21
- "learning_rate": 8.928571428571429e-07,
22
- "loss": 1.4166,
23
  "step": 5
24
  },
25
  {
26
- "epoch": 0.01813236627379873,
27
- "grad_norm": 1.5839917103642143,
28
- "learning_rate": 1.7857142857142859e-06,
29
- "loss": 1.4271,
30
  "step": 10
31
  },
32
  {
33
- "epoch": 0.027198549410698096,
34
- "grad_norm": 1.4459847085261053,
35
- "learning_rate": 2.6785714285714285e-06,
36
- "loss": 1.4264,
37
  "step": 15
38
  },
39
  {
40
- "epoch": 0.03626473254759746,
41
- "grad_norm": 1.2521508384259885,
42
- "learning_rate": 3.5714285714285718e-06,
43
- "loss": 1.4045,
44
  "step": 20
45
  },
46
  {
47
- "epoch": 0.04533091568449683,
48
- "grad_norm": 1.0609715974133789,
49
- "learning_rate": 4.464285714285715e-06,
50
- "loss": 1.3712,
51
  "step": 25
52
  },
53
  {
54
- "epoch": 0.05439709882139619,
55
- "grad_norm": 0.8425419586295035,
56
- "learning_rate": 5.357142857142857e-06,
57
- "loss": 1.3389,
58
  "step": 30
59
  },
60
  {
61
- "epoch": 0.06346328195829556,
62
- "grad_norm": 0.6611927082349899,
63
- "learning_rate": 6.25e-06,
64
- "loss": 1.3212,
65
  "step": 35
66
  },
67
  {
68
- "epoch": 0.07252946509519492,
69
- "grad_norm": 0.5555400411042388,
70
- "learning_rate": 7.1428571428571436e-06,
71
- "loss": 1.2948,
72
  "step": 40
73
  },
74
  {
75
- "epoch": 0.0815956482320943,
76
- "grad_norm": 0.5418758139639963,
77
- "learning_rate": 8.035714285714286e-06,
78
- "loss": 1.2633,
79
  "step": 45
80
  },
81
  {
82
- "epoch": 0.09066183136899365,
83
- "grad_norm": 0.5238281972988632,
84
- "learning_rate": 8.92857142857143e-06,
85
- "loss": 1.2413,
86
  "step": 50
87
  },
88
  {
89
- "epoch": 0.09972801450589303,
90
- "grad_norm": 0.4543292624693451,
91
- "learning_rate": 9.821428571428573e-06,
92
- "loss": 1.2091,
93
  "step": 55
94
  },
95
  {
96
- "epoch": 0.10879419764279238,
97
- "grad_norm": 0.4372745406289742,
98
- "learning_rate": 9.998388886954546e-06,
99
- "loss": 1.1841,
100
  "step": 60
101
  },
102
  {
103
- "epoch": 0.11786038077969176,
104
- "grad_norm": 0.44362524920751634,
105
- "learning_rate": 9.991845519630679e-06,
106
- "loss": 1.1716,
107
  "step": 65
108
  },
109
  {
110
- "epoch": 0.12692656391659113,
111
- "grad_norm": 0.4607131462857648,
112
- "learning_rate": 9.98027578700917e-06,
113
- "loss": 1.1355,
114
  "step": 70
115
  },
116
  {
117
- "epoch": 0.1359927470534905,
118
- "grad_norm": 0.49049026812769303,
119
- "learning_rate": 9.963691338830045e-06,
120
- "loss": 1.1037,
121
  "step": 75
122
  },
123
  {
124
- "epoch": 0.14505893019038985,
125
- "grad_norm": 0.4405119219906117,
126
- "learning_rate": 9.942108874226812e-06,
127
- "loss": 1.0687,
128
  "step": 80
129
  },
130
  {
131
- "epoch": 0.1541251133272892,
132
- "grad_norm": 0.3404728933735868,
133
- "learning_rate": 9.915550124911866e-06,
134
- "loss": 1.0401,
135
  "step": 85
136
  },
137
  {
138
- "epoch": 0.1631912964641886,
139
- "grad_norm": 0.2944607543926559,
140
- "learning_rate": 9.884041833294477e-06,
141
- "loss": 1.026,
142
  "step": 90
143
  },
144
  {
145
- "epoch": 0.17225747960108795,
146
- "grad_norm": 0.2583435958941659,
147
- "learning_rate": 9.847615725553457e-06,
148
- "loss": 1.0279,
149
  "step": 95
150
  },
151
  {
152
- "epoch": 0.1813236627379873,
153
- "grad_norm": 0.24731705775803128,
154
- "learning_rate": 9.806308479691595e-06,
155
- "loss": 1.0249,
156
  "step": 100
157
  },
158
  {
159
- "epoch": 0.19038984587488667,
160
- "grad_norm": 0.24191833692118578,
161
- "learning_rate": 9.760161688604008e-06,
162
- "loss": 1.0105,
163
  "step": 105
164
  },
165
  {
166
- "epoch": 0.19945602901178605,
167
- "grad_norm": 0.24042046466972933,
168
- "learning_rate": 9.709221818197626e-06,
169
- "loss": 1.009,
170
  "step": 110
171
  },
172
  {
173
- "epoch": 0.2085222121486854,
174
- "grad_norm": 0.24389783391468597,
175
- "learning_rate": 9.653540160603956e-06,
176
- "loss": 0.9954,
177
  "step": 115
178
  },
179
  {
180
- "epoch": 0.21758839528558477,
181
- "grad_norm": 0.2399168792536427,
182
- "learning_rate": 9.59317278253227e-06,
183
- "loss": 0.9907,
184
  "step": 120
185
  },
186
  {
187
- "epoch": 0.22665457842248413,
188
- "grad_norm": 0.22813321731661593,
189
- "learning_rate": 9.528180468815155e-06,
190
- "loss": 0.9825,
191
  "step": 125
192
  },
193
  {
194
- "epoch": 0.2357207615593835,
195
- "grad_norm": 0.22409230838099056,
196
- "learning_rate": 9.458628661203368e-06,
197
- "loss": 0.986,
198
  "step": 130
199
  },
200
  {
201
- "epoch": 0.24478694469628287,
202
- "grad_norm": 0.22996377012206487,
203
- "learning_rate": 9.384587392471516e-06,
204
- "loss": 0.9751,
205
  "step": 135
206
  },
207
  {
208
- "epoch": 0.25385312783318226,
209
- "grad_norm": 0.23536306295912668,
210
- "learning_rate": 9.306131215901004e-06,
211
- "loss": 0.9864,
212
  "step": 140
213
  },
214
  {
215
- "epoch": 0.2629193109700816,
216
- "grad_norm": 0.22867770708215468,
217
- "learning_rate": 9.223339130211194e-06,
218
- "loss": 0.9756,
219
  "step": 145
220
  },
221
  {
222
- "epoch": 0.271985494106981,
223
- "grad_norm": 0.23364791744034372,
224
- "learning_rate": 9.136294500014387e-06,
225
- "loss": 0.9707,
226
  "step": 150
227
  },
228
  {
229
- "epoch": 0.2810516772438803,
230
- "grad_norm": 0.24574250376557552,
231
- "learning_rate": 9.045084971874738e-06,
232
- "loss": 0.9661,
233
  "step": 155
234
  },
235
  {
236
- "epoch": 0.2901178603807797,
237
- "grad_norm": 0.22924203105990873,
238
- "learning_rate": 8.949802386055582e-06,
239
- "loss": 0.9687,
240
  "step": 160
241
  },
242
  {
243
- "epoch": 0.2991840435176791,
244
- "grad_norm": 0.2316549750033851,
245
- "learning_rate": 8.850542684044078e-06,
246
- "loss": 0.9575,
247
  "step": 165
248
  },
249
  {
250
- "epoch": 0.3082502266545784,
251
- "grad_norm": 0.23845378563027272,
252
- "learning_rate": 8.747405811946272e-06,
253
- "loss": 0.9562,
254
  "step": 170
255
  },
256
  {
257
- "epoch": 0.3173164097914778,
258
- "grad_norm": 0.23837888017662803,
259
- "learning_rate": 8.640495619849821e-06,
260
- "loss": 0.9576,
261
  "step": 175
262
  },
263
  {
264
- "epoch": 0.3263825929283772,
265
- "grad_norm": 0.2383332569119122,
266
- "learning_rate": 8.529919757255783e-06,
267
- "loss": 0.957,
268
  "step": 180
269
  },
270
  {
271
- "epoch": 0.3354487760652765,
272
- "grad_norm": 0.23286580024364317,
273
- "learning_rate": 8.415789564684673e-06,
274
- "loss": 0.9612,
275
  "step": 185
276
  },
277
  {
278
- "epoch": 0.3445149592021759,
279
- "grad_norm": 0.22780730515845546,
280
- "learning_rate": 8.298219961566008e-06,
281
- "loss": 0.9553,
282
  "step": 190
283
  },
284
  {
285
- "epoch": 0.3535811423390752,
286
- "grad_norm": 0.2323208245333303,
287
- "learning_rate": 8.177329330524182e-06,
288
- "loss": 0.9529,
289
  "step": 195
290
  },
291
  {
292
- "epoch": 0.3626473254759746,
293
- "grad_norm": 0.24805785335563235,
294
- "learning_rate": 8.053239398177191e-06,
295
- "loss": 0.9517,
296
  "step": 200
297
  },
298
  {
299
- "epoch": 0.371713508612874,
300
- "grad_norm": 0.23665008354936803,
301
- "learning_rate": 7.92607511256826e-06,
302
- "loss": 0.9527,
303
  "step": 205
304
  },
305
  {
306
- "epoch": 0.38077969174977333,
307
- "grad_norm": 0.25231109855641104,
308
- "learning_rate": 7.795964517353734e-06,
309
- "loss": 0.9475,
310
  "step": 210
311
  },
312
  {
313
- "epoch": 0.3898458748866727,
314
- "grad_norm": 0.2640919317704817,
315
- "learning_rate": 7.663038622873999e-06,
316
- "loss": 0.956,
317
  "step": 215
318
  },
319
  {
320
- "epoch": 0.3989120580235721,
321
- "grad_norm": 0.2728546324917129,
322
- "learning_rate": 7.527431274237149e-06,
323
- "loss": 0.9519,
324
  "step": 220
325
  },
326
  {
327
- "epoch": 0.40797824116047143,
328
- "grad_norm": 0.24572462484094396,
329
- "learning_rate": 7.3892790165483164e-06,
330
- "loss": 0.9489,
331
  "step": 225
332
  },
333
  {
334
- "epoch": 0.4170444242973708,
335
- "grad_norm": 0.2580500734658738,
336
- "learning_rate": 7.24872095742033e-06,
337
- "loss": 0.9516,
338
  "step": 230
339
  },
340
  {
341
- "epoch": 0.42611060743427015,
342
- "grad_norm": 0.24803902317298773,
343
- "learning_rate": 7.105898626904134e-06,
344
- "loss": 0.951,
345
  "step": 235
346
  },
347
  {
348
- "epoch": 0.43517679057116954,
349
- "grad_norm": 0.25527768819790775,
350
- "learning_rate": 6.960955834980028e-06,
351
- "loss": 0.9524,
352
  "step": 240
353
  },
354
  {
355
- "epoch": 0.4442429737080689,
356
- "grad_norm": 0.25074573855319177,
357
- "learning_rate": 6.814038526753205e-06,
358
- "loss": 0.9429,
359
  "step": 245
360
  },
361
  {
362
- "epoch": 0.45330915684496825,
363
- "grad_norm": 0.2892217050535589,
364
- "learning_rate": 6.665294635499404e-06,
365
- "loss": 0.9387,
366
  "step": 250
367
  },
368
  {
369
- "epoch": 0.46237533998186764,
370
- "grad_norm": 0.253783191309419,
371
- "learning_rate": 6.514873933708637e-06,
372
- "loss": 0.9431,
373
  "step": 255
374
  },
375
  {
376
- "epoch": 0.471441523118767,
377
- "grad_norm": 0.2955490720646012,
378
- "learning_rate": 6.362927882276991e-06,
379
- "loss": 0.9439,
380
  "step": 260
381
  },
382
  {
383
- "epoch": 0.48050770625566636,
384
- "grad_norm": 0.25718021395217194,
385
- "learning_rate": 6.209609477998339e-06,
386
- "loss": 0.945,
387
  "step": 265
388
  },
389
  {
390
- "epoch": 0.48957388939256574,
391
- "grad_norm": 0.2525565505597455,
392
- "learning_rate": 6.055073099509549e-06,
393
- "loss": 0.942,
394
  "step": 270
395
  },
396
  {
397
- "epoch": 0.4986400725294651,
398
- "grad_norm": 0.2607701707964316,
399
- "learning_rate": 5.89947435184427e-06,
400
- "loss": 0.9354,
401
  "step": 275
402
  },
403
  {
404
- "epoch": 0.5077062556663645,
405
- "grad_norm": 0.2632871574380947,
406
- "learning_rate": 5.7429699097518585e-06,
407
- "loss": 0.9367,
408
  "step": 280
409
  },
410
  {
411
- "epoch": 0.5167724388032638,
412
- "grad_norm": 0.26054567909822246,
413
- "learning_rate": 5.585717359939192e-06,
414
- "loss": 0.9408,
415
  "step": 285
416
  },
417
  {
418
- "epoch": 0.5258386219401632,
419
- "grad_norm": 0.261273626471609,
420
- "learning_rate": 5.4278750423942e-06,
421
- "loss": 0.9411,
422
  "step": 290
423
  },
424
  {
425
- "epoch": 0.5349048050770625,
426
- "grad_norm": 0.267892663759971,
427
- "learning_rate": 5.26960189095093e-06,
428
- "loss": 0.9347,
429
  "step": 295
430
  },
431
  {
432
- "epoch": 0.543970988213962,
433
- "grad_norm": 0.26789779552055176,
434
- "learning_rate": 5.111057273256648e-06,
435
- "loss": 0.9307,
436
  "step": 300
437
  },
438
  {
439
- "epoch": 0.5530371713508613,
440
- "grad_norm": 0.2568984132631926,
441
- "learning_rate": 4.952400830302117e-06,
442
- "loss": 0.9386,
443
  "step": 305
444
  },
445
  {
446
- "epoch": 0.5621033544877606,
447
- "grad_norm": 0.2752971863164433,
448
- "learning_rate": 4.793792315676665e-06,
449
- "loss": 0.9447,
450
  "step": 310
451
  },
452
  {
453
- "epoch": 0.57116953762466,
454
- "grad_norm": 0.28286656586411274,
455
- "learning_rate": 4.635391434709847e-06,
456
- "loss": 0.9411,
457
  "step": 315
458
  },
459
  {
460
- "epoch": 0.5802357207615594,
461
- "grad_norm": 0.27251868450404326,
462
- "learning_rate": 4.477357683661734e-06,
463
- "loss": 0.9333,
464
  "step": 320
465
  },
466
  {
467
- "epoch": 0.5893019038984587,
468
- "grad_norm": 0.26134497842639987,
469
- "learning_rate": 4.319850189123681e-06,
470
- "loss": 0.9362,
471
  "step": 325
472
  },
473
  {
474
- "epoch": 0.5983680870353582,
475
- "grad_norm": 0.28846220700452374,
476
- "learning_rate": 4.163027547791347e-06,
477
- "loss": 0.9353,
478
  "step": 330
479
  },
480
  {
481
- "epoch": 0.6074342701722575,
482
- "grad_norm": 0.2791534046074814,
483
- "learning_rate": 4.007047666771274e-06,
484
- "loss": 0.9442,
485
  "step": 335
486
  },
487
  {
488
- "epoch": 0.6165004533091568,
489
- "grad_norm": 0.2649936484712171,
490
- "learning_rate": 3.8520676045817945e-06,
491
- "loss": 0.9362,
492
  "step": 340
493
  },
494
  {
495
- "epoch": 0.6255666364460563,
496
- "grad_norm": 0.32310076173900937,
497
- "learning_rate": 3.69824341300844e-06,
498
- "loss": 0.9355,
499
  "step": 345
500
  },
501
  {
502
- "epoch": 0.6346328195829556,
503
- "grad_norm": 0.2973560173396777,
504
- "learning_rate": 3.5457299799730047e-06,
505
- "loss": 0.9361,
506
  "step": 350
507
  },
508
  {
509
- "epoch": 0.6436990027198549,
510
- "grad_norm": 0.2943763714466763,
511
- "learning_rate": 3.394680873574546e-06,
512
- "loss": 0.9352,
513
  "step": 355
514
  },
515
  {
516
- "epoch": 0.6527651858567544,
517
- "grad_norm": 0.2664148416975679,
518
- "learning_rate": 3.2452481874593234e-06,
519
- "loss": 0.9317,
520
  "step": 360
521
  },
522
  {
523
- "epoch": 0.6618313689936537,
524
- "grad_norm": 0.2751770240187752,
525
- "learning_rate": 3.097582387675385e-06,
526
- "loss": 0.936,
527
  "step": 365
528
  },
529
  {
530
- "epoch": 0.670897552130553,
531
- "grad_norm": 0.27382749760131525,
532
- "learning_rate": 2.951832161166024e-06,
533
- "loss": 0.9357,
534
  "step": 370
535
  },
536
  {
537
- "epoch": 0.6799637352674524,
538
- "grad_norm": 0.2746042204403535,
539
- "learning_rate": 2.8081442660546126e-06,
540
- "loss": 0.933,
541
  "step": 375
542
  },
543
  {
544
- "epoch": 0.6890299184043518,
545
- "grad_norm": 0.2672719630547927,
546
- "learning_rate": 2.6666633838716317e-06,
547
- "loss": 0.9312,
548
  "step": 380
549
  },
550
  {
551
- "epoch": 0.6980961015412511,
552
- "grad_norm": 0.2710211886898411,
553
- "learning_rate": 2.527531973872617e-06,
554
- "loss": 0.929,
555
  "step": 385
556
  },
557
  {
558
- "epoch": 0.7071622846781505,
559
- "grad_norm": 0.2836603128928457,
560
- "learning_rate": 2.3908901295937713e-06,
561
- "loss": 0.9366,
562
  "step": 390
563
  },
564
  {
565
- "epoch": 0.7162284678150499,
566
- "grad_norm": 0.26008430614825967,
567
- "learning_rate": 2.2568754377896516e-06,
568
- "loss": 0.932,
569
  "step": 395
570
  },
571
  {
572
- "epoch": 0.7252946509519492,
573
- "grad_norm": 0.2810813209490739,
574
- "learning_rate": 2.125622839894964e-06,
575
- "loss": 0.9279,
576
  "step": 400
577
  },
578
  {
579
- "epoch": 0.7343608340888486,
580
- "grad_norm": 0.28237517955456193,
581
- "learning_rate": 1.9972644961499853e-06,
582
- "loss": 0.9307,
583
  "step": 405
584
  },
585
  {
586
- "epoch": 0.743427017225748,
587
- "grad_norm": 0.26852889419057147,
588
- "learning_rate": 1.8719296525263925e-06,
589
- "loss": 0.9287,
590
  "step": 410
591
  },
592
  {
593
- "epoch": 0.7524932003626473,
594
- "grad_norm": 0.275008082951018,
595
- "learning_rate": 1.7497445105875377e-06,
596
- "loss": 0.9265,
597
  "step": 415
598
  },
599
  {
600
- "epoch": 0.7615593834995467,
601
- "grad_norm": 0.2683518983451818,
602
- "learning_rate": 1.6308321004141609e-06,
603
- "loss": 0.9305,
604
  "step": 420
605
  },
606
  {
607
- "epoch": 0.7706255666364461,
608
- "grad_norm": 0.27766234829147984,
609
- "learning_rate": 1.5153121567235334e-06,
610
- "loss": 0.9293,
611
  "step": 425
612
  },
613
  {
614
- "epoch": 0.7796917497733454,
615
- "grad_norm": 0.2551135015387663,
616
- "learning_rate": 1.4033009983067454e-06,
617
- "loss": 0.9293,
618
  "step": 430
619
  },
620
  {
621
- "epoch": 0.7887579329102448,
622
- "grad_norm": 0.26684387240821533,
623
- "learning_rate": 1.2949114109055417e-06,
624
- "loss": 0.9326,
625
  "step": 435
626
  },
627
  {
628
- "epoch": 0.7978241160471442,
629
- "grad_norm": 0.263475545535637,
630
- "learning_rate": 1.1902525336466465e-06,
631
- "loss": 0.9336,
632
  "step": 440
633
  },
634
  {
635
- "epoch": 0.8068902991840435,
636
- "grad_norm": 0.26984207139853317,
637
- "learning_rate": 1.0894297491479044e-06,
638
- "loss": 0.9398,
639
  "step": 445
640
  },
641
  {
642
- "epoch": 0.8159564823209429,
643
- "grad_norm": 0.2566066507678282,
644
- "learning_rate": 9.925445774069232e-07,
645
- "loss": 0.933,
646
  "step": 450
647
  },
648
  {
649
- "epoch": 0.8250226654578422,
650
- "grad_norm": 0.2735206272690916,
651
- "learning_rate": 8.996945735790447e-07,
652
- "loss": 0.9289,
653
  "step": 455
654
  },
655
  {
656
- "epoch": 0.8340888485947416,
657
- "grad_norm": 0.2752286092204346,
658
- "learning_rate": 8.109732297475637e-07,
659
- "loss": 0.9376,
660
  "step": 460
661
  },
662
  {
663
- "epoch": 0.843155031731641,
664
- "grad_norm": 0.2578854396417924,
665
- "learning_rate": 7.264698807851328e-07,
666
- "loss": 0.9269,
667
  "step": 465
668
  },
669
  {
670
- "epoch": 0.8522212148685403,
671
- "grad_norm": 0.279822688586382,
672
- "learning_rate": 6.462696144011149e-07,
673
- "loss": 0.9307,
674
  "step": 470
675
  },
676
  {
677
- "epoch": 0.8612873980054397,
678
- "grad_norm": 0.267972480985642,
679
- "learning_rate": 5.704531854654721e-07,
680
- "loss": 0.9301,
681
  "step": 475
682
  },
683
  {
684
- "epoch": 0.8703535811423391,
685
- "grad_norm": 0.27447811669013067,
686
- "learning_rate": 4.990969346954611e-07,
687
- "loss": 0.9336,
688
  "step": 480
689
  },
690
  {
691
- "epoch": 0.8794197642792384,
692
- "grad_norm": 0.26428274302055943,
693
- "learning_rate": 4.322727117869951e-07,
694
- "loss": 0.9232,
695
  "step": 485
696
  },
697
  {
698
- "epoch": 0.8884859474161378,
699
- "grad_norm": 0.2819091291557529,
700
- "learning_rate": 3.7004780306809873e-07,
701
- "loss": 0.9329,
702
  "step": 490
703
  },
704
  {
705
- "epoch": 0.8975521305530372,
706
- "grad_norm": 0.26368048329492144,
707
- "learning_rate": 3.1248486374726884e-07,
708
- "loss": 0.933,
709
  "step": 495
710
  },
711
  {
712
- "epoch": 0.9066183136899365,
713
- "grad_norm": 0.27276688187915327,
714
- "learning_rate": 2.596418548250029e-07,
715
- "loss": 0.9306,
716
  "step": 500
717
  },
718
  {
719
- "epoch": 0.915684496826836,
720
- "grad_norm": 0.25969692879390444,
721
- "learning_rate": 2.1157198473197417e-07,
722
- "loss": 0.9324,
723
  "step": 505
724
  },
725
  {
726
- "epoch": 0.9247506799637353,
727
- "grad_norm": 0.2526365087770105,
728
- "learning_rate": 1.6832365575265742e-07,
729
- "loss": 0.9402,
730
  "step": 510
731
  },
732
  {
733
- "epoch": 0.9338168631006346,
734
- "grad_norm": 0.26187899010526317,
735
- "learning_rate": 1.2994041528833267e-07,
736
- "loss": 0.9343,
737
  "step": 515
738
  },
739
  {
740
- "epoch": 0.942883046237534,
741
- "grad_norm": 0.275531542176856,
742
- "learning_rate": 9.646091200853802e-08,
743
- "loss": 0.9303,
744
  "step": 520
745
  },
746
  {
747
- "epoch": 0.9519492293744334,
748
- "grad_norm": 0.2829534195567479,
749
- "learning_rate": 6.791885693514134e-08,
750
- "loss": 0.9281,
751
  "step": 525
752
  },
753
  {
754
- "epoch": 0.9610154125113327,
755
- "grad_norm": 0.2673494615735909,
756
- "learning_rate": 4.434298949819449e-08,
757
- "loss": 0.9272,
758
  "step": 530
759
  },
760
  {
761
- "epoch": 0.970081595648232,
762
- "grad_norm": 0.2651047754241703,
763
- "learning_rate": 2.57570485977654e-08,
764
- "loss": 0.9318,
765
  "step": 535
766
  },
767
  {
768
- "epoch": 0.9791477787851315,
769
- "grad_norm": 0.2613626592280882,
770
- "learning_rate": 1.2179748700879013e-08,
771
- "loss": 0.934,
772
  "step": 540
773
  },
774
  {
775
- "epoch": 0.9882139619220308,
776
- "grad_norm": 0.2677931475414294,
777
- "learning_rate": 3.6247609976319818e-09,
778
- "loss": 0.932,
779
  "step": 545
780
  },
781
  {
782
- "epoch": 0.9972801450589301,
783
- "grad_norm": 0.2602021958898807,
784
- "learning_rate": 1.0069963546743833e-10,
785
- "loss": 0.9257,
786
  "step": 550
787
  },
788
  {
789
- "epoch": 0.99909338168631,
790
- "eval_loss": 0.9361373782157898,
791
- "eval_runtime": 1.4957,
792
- "eval_samples_per_second": 2.006,
793
- "eval_steps_per_second": 0.669,
794
- "step": 551
795
  },
796
  {
797
- "epoch": 0.99909338168631,
798
- "step": 551,
799
- "total_flos": 2216760396742656.0,
800
- "train_loss": 0.9953486248715603,
801
- "train_runtime": 22334.5637,
802
- "train_samples_per_second": 1.579,
803
- "train_steps_per_second": 0.025
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
804
  }
805
  ],
806
  "logging_steps": 5,
807
- "max_steps": 551,
808
  "num_input_tokens_seen": 0,
809
  "num_train_epochs": 1,
810
  "save_steps": 100,
@@ -820,8 +1744,8 @@
820
  "attributes": {}
821
  }
822
  },
823
- "total_flos": 2216760396742656.0,
824
- "train_batch_size": 8,
825
  "trial_name": null,
826
  "trial_params": null
827
  }
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.9993805492463349,
5
  "eval_steps": 500,
6
+ "global_step": 1210,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
+ "epoch": 0.0008259343382201115,
13
+ "grad_norm": 23.24616747351659,
14
+ "learning_rate": 8.264462809917357e-08,
15
+ "loss": 1.4263,
16
  "step": 1
17
  },
18
  {
19
+ "epoch": 0.0041296716911005575,
20
+ "grad_norm": 24.213699866017716,
21
+ "learning_rate": 4.132231404958678e-07,
22
+ "loss": 1.4129,
23
  "step": 5
24
  },
25
  {
26
+ "epoch": 0.008259343382201115,
27
+ "grad_norm": 8.707304164936993,
28
+ "learning_rate": 8.264462809917356e-07,
29
+ "loss": 1.3289,
30
  "step": 10
31
  },
32
  {
33
+ "epoch": 0.012389015073301672,
34
+ "grad_norm": 10.918691302163335,
35
+ "learning_rate": 1.2396694214876035e-06,
36
+ "loss": 1.1721,
37
  "step": 15
38
  },
39
  {
40
+ "epoch": 0.01651868676440223,
41
+ "grad_norm": 2.87174085951705,
42
+ "learning_rate": 1.6528925619834712e-06,
43
+ "loss": 1.0219,
44
  "step": 20
45
  },
46
  {
47
+ "epoch": 0.020648358455502787,
48
+ "grad_norm": 3.071874166303048,
49
+ "learning_rate": 2.066115702479339e-06,
50
+ "loss": 0.9724,
51
  "step": 25
52
  },
53
  {
54
+ "epoch": 0.024778030146603344,
55
+ "grad_norm": 2.4133173682706213,
56
+ "learning_rate": 2.479338842975207e-06,
57
+ "loss": 0.9426,
58
  "step": 30
59
  },
60
  {
61
+ "epoch": 0.028907701837703904,
62
+ "grad_norm": 2.236585592513117,
63
+ "learning_rate": 2.8925619834710743e-06,
64
+ "loss": 0.921,
65
  "step": 35
66
  },
67
  {
68
+ "epoch": 0.03303737352880446,
69
+ "grad_norm": 2.385522667127509,
70
+ "learning_rate": 3.3057851239669424e-06,
71
+ "loss": 0.9121,
72
  "step": 40
73
  },
74
  {
75
+ "epoch": 0.03716704521990502,
76
+ "grad_norm": 2.380746269584164,
77
+ "learning_rate": 3.71900826446281e-06,
78
+ "loss": 0.9001,
79
  "step": 45
80
  },
81
  {
82
+ "epoch": 0.041296716911005574,
83
+ "grad_norm": 2.569197332815026,
84
+ "learning_rate": 4.132231404958678e-06,
85
+ "loss": 0.8867,
86
  "step": 50
87
  },
88
  {
89
+ "epoch": 0.045426388602106134,
90
+ "grad_norm": 2.3723293989848213,
91
+ "learning_rate": 4.5454545454545455e-06,
92
+ "loss": 0.8707,
93
  "step": 55
94
  },
95
  {
96
+ "epoch": 0.04955606029320669,
97
+ "grad_norm": 2.3355520193379142,
98
+ "learning_rate": 4.958677685950414e-06,
99
+ "loss": 0.8698,
100
  "step": 60
101
  },
102
  {
103
+ "epoch": 0.05368573198430725,
104
+ "grad_norm": 2.375730929200025,
105
+ "learning_rate": 5.371900826446281e-06,
106
+ "loss": 0.859,
107
  "step": 65
108
  },
109
  {
110
+ "epoch": 0.05781540367540781,
111
+ "grad_norm": 2.3481213761418838,
112
+ "learning_rate": 5.785123966942149e-06,
113
+ "loss": 0.8632,
114
  "step": 70
115
  },
116
  {
117
+ "epoch": 0.06194507536650836,
118
+ "grad_norm": 2.3161171161194254,
119
+ "learning_rate": 6.198347107438017e-06,
120
+ "loss": 0.8462,
121
  "step": 75
122
  },
123
  {
124
+ "epoch": 0.06607474705760892,
125
+ "grad_norm": 2.273903126512892,
126
+ "learning_rate": 6.611570247933885e-06,
127
+ "loss": 0.8593,
128
  "step": 80
129
  },
130
  {
131
+ "epoch": 0.07020441874870947,
132
+ "grad_norm": 2.3651082830117787,
133
+ "learning_rate": 7.0247933884297525e-06,
134
+ "loss": 0.8388,
135
  "step": 85
136
  },
137
  {
138
+ "epoch": 0.07433409043981004,
139
+ "grad_norm": 2.5780580351650584,
140
+ "learning_rate": 7.43801652892562e-06,
141
+ "loss": 0.8354,
142
  "step": 90
143
  },
144
  {
145
+ "epoch": 0.0784637621309106,
146
+ "grad_norm": 2.5881003929304165,
147
+ "learning_rate": 7.851239669421489e-06,
148
+ "loss": 0.8177,
149
  "step": 95
150
  },
151
  {
152
+ "epoch": 0.08259343382201115,
153
+ "grad_norm": 2.3662580429191262,
154
+ "learning_rate": 8.264462809917356e-06,
155
+ "loss": 0.8397,
156
  "step": 100
157
  },
158
  {
159
+ "epoch": 0.0867231055131117,
160
+ "grad_norm": 2.4549245207596173,
161
+ "learning_rate": 8.677685950413224e-06,
162
+ "loss": 0.8209,
163
  "step": 105
164
  },
165
  {
166
+ "epoch": 0.09085277720421227,
167
+ "grad_norm": 2.6986718605229454,
168
+ "learning_rate": 9.090909090909091e-06,
169
+ "loss": 0.8076,
170
  "step": 110
171
  },
172
  {
173
+ "epoch": 0.09498244889531282,
174
+ "grad_norm": 2.6597400001383136,
175
+ "learning_rate": 9.50413223140496e-06,
176
+ "loss": 0.812,
177
  "step": 115
178
  },
179
  {
180
+ "epoch": 0.09911212058641337,
181
+ "grad_norm": 2.6969489551453703,
182
+ "learning_rate": 9.917355371900828e-06,
183
+ "loss": 0.8064,
184
  "step": 120
185
  },
186
  {
187
+ "epoch": 0.10324179227751394,
188
+ "grad_norm": 2.5801588887536204,
189
+ "learning_rate": 9.999667111219573e-06,
190
+ "loss": 0.7993,
191
  "step": 125
192
  },
193
  {
194
+ "epoch": 0.1073714639686145,
195
+ "grad_norm": 2.5177823470227314,
196
+ "learning_rate": 9.998314826517564e-06,
197
+ "loss": 0.8123,
198
  "step": 130
199
  },
200
  {
201
+ "epoch": 0.11150113565971505,
202
+ "grad_norm": 2.3846369879591767,
203
+ "learning_rate": 9.995922621477252e-06,
204
+ "loss": 0.8064,
205
  "step": 135
206
  },
207
  {
208
+ "epoch": 0.11563080735081561,
209
+ "grad_norm": 2.37259551970217,
210
+ "learning_rate": 9.99249099380692e-06,
211
+ "loss": 0.7961,
212
  "step": 140
213
  },
214
  {
215
+ "epoch": 0.11976047904191617,
216
+ "grad_norm": 2.1621998588898346,
217
+ "learning_rate": 9.988020657471078e-06,
218
+ "loss": 0.7924,
219
  "step": 145
220
  },
221
  {
222
+ "epoch": 0.12389015073301672,
223
+ "grad_norm": 2.3342111944519623,
224
+ "learning_rate": 9.98251254254193e-06,
225
+ "loss": 0.7785,
226
  "step": 150
227
  },
228
  {
229
+ "epoch": 0.12801982242411727,
230
+ "grad_norm": 2.3550905722953823,
231
+ "learning_rate": 9.97596779500586e-06,
232
+ "loss": 0.7695,
233
  "step": 155
234
  },
235
  {
236
+ "epoch": 0.13214949411521784,
237
+ "grad_norm": 2.4721774789219344,
238
+ "learning_rate": 9.968387776525009e-06,
239
+ "loss": 0.7777,
240
  "step": 160
241
  },
242
  {
243
+ "epoch": 0.1362791658063184,
244
+ "grad_norm": 2.390765566890305,
245
+ "learning_rate": 9.959774064153977e-06,
246
+ "loss": 0.7865,
247
  "step": 165
248
  },
249
  {
250
+ "epoch": 0.14040883749741895,
251
+ "grad_norm": 2.392413139550164,
252
+ "learning_rate": 9.950128450011706e-06,
253
+ "loss": 0.7723,
254
  "step": 170
255
  },
256
  {
257
+ "epoch": 0.14453850918851952,
258
+ "grad_norm": 2.24515023866806,
259
+ "learning_rate": 9.939452940908627e-06,
260
+ "loss": 0.755,
261
  "step": 175
262
  },
263
  {
264
+ "epoch": 0.14866818087962008,
265
+ "grad_norm": 2.546933963480071,
266
+ "learning_rate": 9.927749757929125e-06,
267
+ "loss": 0.7538,
268
  "step": 180
269
  },
270
  {
271
+ "epoch": 0.15279785257072062,
272
+ "grad_norm": 2.5390977331021136,
273
+ "learning_rate": 9.915021335969452e-06,
274
+ "loss": 0.7436,
275
  "step": 185
276
  },
277
  {
278
+ "epoch": 0.1569275242618212,
279
+ "grad_norm": 2.435644472976208,
280
+ "learning_rate": 9.901270323231114e-06,
281
+ "loss": 0.7423,
282
  "step": 190
283
  },
284
  {
285
+ "epoch": 0.16105719595292176,
286
+ "grad_norm": 2.3962427953474794,
287
+ "learning_rate": 9.886499580669917e-06,
288
+ "loss": 0.7402,
289
  "step": 195
290
  },
291
  {
292
+ "epoch": 0.1651868676440223,
293
+ "grad_norm": 2.293192651317426,
294
+ "learning_rate": 9.870712181400726e-06,
295
+ "loss": 0.7328,
296
  "step": 200
297
  },
298
  {
299
+ "epoch": 0.16931653933512286,
300
+ "grad_norm": 2.3625939065789523,
301
+ "learning_rate": 9.853911410058097e-06,
302
+ "loss": 0.7414,
303
  "step": 205
304
  },
305
  {
306
+ "epoch": 0.1734462110262234,
307
+ "grad_norm": 2.1188762141799633,
308
+ "learning_rate": 9.836100762112887e-06,
309
+ "loss": 0.7324,
310
  "step": 210
311
  },
312
  {
313
+ "epoch": 0.17757588271732397,
314
+ "grad_norm": 2.3459257676868885,
315
+ "learning_rate": 9.817283943145014e-06,
316
+ "loss": 0.7216,
317
  "step": 215
318
  },
319
  {
320
+ "epoch": 0.18170555440842454,
321
+ "grad_norm": 2.353468899573234,
322
+ "learning_rate": 9.797464868072489e-06,
323
+ "loss": 0.7244,
324
  "step": 220
325
  },
326
  {
327
+ "epoch": 0.18583522609952507,
328
+ "grad_norm": 2.204966096546786,
329
+ "learning_rate": 9.776647660336905e-06,
330
+ "loss": 0.7193,
331
  "step": 225
332
  },
333
  {
334
+ "epoch": 0.18996489779062564,
335
+ "grad_norm": 2.310006498983248,
336
+ "learning_rate": 9.754836651045538e-06,
337
+ "loss": 0.7161,
338
  "step": 230
339
  },
340
  {
341
+ "epoch": 0.1940945694817262,
342
+ "grad_norm": 2.5104871936969277,
343
+ "learning_rate": 9.732036378070243e-06,
344
+ "loss": 0.7044,
345
  "step": 235
346
  },
347
  {
348
+ "epoch": 0.19822424117282675,
349
+ "grad_norm": 2.366571213288955,
350
+ "learning_rate": 9.708251585103324e-06,
351
+ "loss": 0.7004,
352
  "step": 240
353
  },
354
  {
355
+ "epoch": 0.20235391286392732,
356
+ "grad_norm": 2.570331809825131,
357
+ "learning_rate": 9.683487220670595e-06,
358
+ "loss": 0.7158,
359
  "step": 245
360
  },
361
  {
362
+ "epoch": 0.20648358455502788,
363
+ "grad_norm": 2.4620380694723147,
364
+ "learning_rate": 9.657748437101819e-06,
365
+ "loss": 0.7019,
366
  "step": 250
367
  },
368
  {
369
+ "epoch": 0.21061325624612842,
370
+ "grad_norm": 2.3314886422942513,
371
+ "learning_rate": 9.631040589458742e-06,
372
+ "loss": 0.6989,
373
  "step": 255
374
  },
375
  {
376
+ "epoch": 0.214742927937229,
377
+ "grad_norm": 2.2564618217324743,
378
+ "learning_rate": 9.603369234420944e-06,
379
+ "loss": 0.6947,
380
  "step": 260
381
  },
382
  {
383
+ "epoch": 0.21887259962832956,
384
+ "grad_norm": 2.2447803221031903,
385
+ "learning_rate": 9.574740129129767e-06,
386
+ "loss": 0.6756,
387
  "step": 265
388
  },
389
  {
390
+ "epoch": 0.2230022713194301,
391
+ "grad_norm": 2.590402933040932,
392
+ "learning_rate": 9.545159229990493e-06,
393
+ "loss": 0.6795,
394
  "step": 270
395
  },
396
  {
397
+ "epoch": 0.22713194301053066,
398
+ "grad_norm": 2.3017458472761634,
399
+ "learning_rate": 9.514632691433108e-06,
400
+ "loss": 0.6905,
401
  "step": 275
402
  },
403
  {
404
+ "epoch": 0.23126161470163123,
405
+ "grad_norm": 2.363184146884425,
406
+ "learning_rate": 9.483166864631837e-06,
407
+ "loss": 0.6572,
408
  "step": 280
409
  },
410
  {
411
+ "epoch": 0.23539128639273177,
412
+ "grad_norm": 2.531694092570608,
413
+ "learning_rate": 9.450768296183764e-06,
414
+ "loss": 0.6694,
415
  "step": 285
416
  },
417
  {
418
+ "epoch": 0.23952095808383234,
419
+ "grad_norm": 2.3202550588633017,
420
+ "learning_rate": 9.417443726746776e-06,
421
+ "loss": 0.6704,
422
  "step": 290
423
  },
424
  {
425
+ "epoch": 0.2436506297749329,
426
+ "grad_norm": 2.3120309970976685,
427
+ "learning_rate": 9.383200089637143e-06,
428
+ "loss": 0.6572,
429
  "step": 295
430
  },
431
  {
432
+ "epoch": 0.24778030146603344,
433
+ "grad_norm": 2.3498499961336687,
434
+ "learning_rate": 9.348044509387022e-06,
435
+ "loss": 0.6655,
436
  "step": 300
437
  },
438
  {
439
+ "epoch": 0.251909973157134,
440
+ "grad_norm": 2.5440021629079665,
441
+ "learning_rate": 9.311984300262151e-06,
442
+ "loss": 0.6612,
443
  "step": 305
444
  },
445
  {
446
+ "epoch": 0.25603964484823455,
447
+ "grad_norm": 2.358901339481344,
448
+ "learning_rate": 9.275026964740101e-06,
449
+ "loss": 0.6444,
450
  "step": 310
451
  },
452
  {
453
+ "epoch": 0.2601693165393351,
454
+ "grad_norm": 2.206937338630286,
455
+ "learning_rate": 9.237180191949347e-06,
456
+ "loss": 0.6445,
457
  "step": 315
458
  },
459
  {
460
+ "epoch": 0.2642989882304357,
461
+ "grad_norm": 2.226748345548228,
462
+ "learning_rate": 9.198451856069514e-06,
463
+ "loss": 0.6474,
464
  "step": 320
465
  },
466
  {
467
+ "epoch": 0.26842865992153625,
468
+ "grad_norm": 2.366603978699467,
469
+ "learning_rate": 9.158850014693123e-06,
470
+ "loss": 0.6361,
471
  "step": 325
472
  },
473
  {
474
+ "epoch": 0.2725583316126368,
475
+ "grad_norm": 2.2830136404441657,
476
+ "learning_rate": 9.118382907149164e-06,
477
+ "loss": 0.6395,
478
  "step": 330
479
  },
480
  {
481
+ "epoch": 0.27668800330373733,
482
+ "grad_norm": 2.4029340921426727,
483
+ "learning_rate": 9.077058952788888e-06,
484
+ "loss": 0.6379,
485
  "step": 335
486
  },
487
  {
488
+ "epoch": 0.2808176749948379,
489
+ "grad_norm": 2.4477206203283512,
490
+ "learning_rate": 9.034886749234112e-06,
491
+ "loss": 0.6272,
492
  "step": 340
493
  },
494
  {
495
+ "epoch": 0.28494734668593846,
496
+ "grad_norm": 2.3054439086007243,
497
+ "learning_rate": 8.991875070588449e-06,
498
+ "loss": 0.6279,
499
  "step": 345
500
  },
501
  {
502
+ "epoch": 0.28907701837703903,
503
+ "grad_norm": 2.2372770463589196,
504
+ "learning_rate": 8.948032865611823e-06,
505
+ "loss": 0.6276,
506
  "step": 350
507
  },
508
  {
509
+ "epoch": 0.2932066900681396,
510
+ "grad_norm": 2.2567729960946803,
511
+ "learning_rate": 8.90336925585864e-06,
512
+ "loss": 0.6107,
513
  "step": 355
514
  },
515
  {
516
+ "epoch": 0.29733636175924016,
517
+ "grad_norm": 2.2634696566360697,
518
+ "learning_rate": 8.857893533780015e-06,
519
+ "loss": 0.6222,
520
  "step": 360
521
  },
522
  {
523
+ "epoch": 0.3014660334503407,
524
+ "grad_norm": 2.2158775605445573,
525
+ "learning_rate": 8.811615160790428e-06,
526
+ "loss": 0.6099,
527
  "step": 365
528
  },
529
  {
530
+ "epoch": 0.30559570514144124,
531
+ "grad_norm": 2.359375453911521,
532
+ "learning_rate": 8.764543765299245e-06,
533
+ "loss": 0.6088,
534
  "step": 370
535
  },
536
  {
537
+ "epoch": 0.3097253768325418,
538
+ "grad_norm": 2.2574003131115017,
539
+ "learning_rate": 8.716689140707488e-06,
540
+ "loss": 0.5978,
541
  "step": 375
542
  },
543
  {
544
+ "epoch": 0.3138550485236424,
545
+ "grad_norm": 2.305475140301284,
546
+ "learning_rate": 8.668061243370273e-06,
547
+ "loss": 0.5955,
548
  "step": 380
549
  },
550
  {
551
+ "epoch": 0.31798472021474294,
552
+ "grad_norm": 2.512743372685998,
553
+ "learning_rate": 8.61867019052535e-06,
554
+ "loss": 0.6035,
555
  "step": 385
556
  },
557
  {
558
+ "epoch": 0.3221143919058435,
559
+ "grad_norm": 2.407216931553165,
560
+ "learning_rate": 8.568526258188172e-06,
561
+ "loss": 0.5836,
562
  "step": 390
563
  },
564
  {
565
+ "epoch": 0.326244063596944,
566
+ "grad_norm": 2.2192688797123066,
567
+ "learning_rate": 8.517639879013918e-06,
568
+ "loss": 0.5888,
569
  "step": 395
570
  },
571
  {
572
+ "epoch": 0.3303737352880446,
573
+ "grad_norm": 2.1358480481396267,
574
+ "learning_rate": 8.466021640126946e-06,
575
+ "loss": 0.5916,
576
  "step": 400
577
  },
578
  {
579
+ "epoch": 0.33450340697914516,
580
+ "grad_norm": 2.1925271239115034,
581
+ "learning_rate": 8.413682280918094e-06,
582
+ "loss": 0.577,
583
  "step": 405
584
  },
585
  {
586
+ "epoch": 0.3386330786702457,
587
+ "grad_norm": 2.2104260296514613,
588
+ "learning_rate": 8.3606326908103e-06,
589
+ "loss": 0.5921,
590
  "step": 410
591
  },
592
  {
593
+ "epoch": 0.3427627503613463,
594
+ "grad_norm": 2.3351708112925946,
595
+ "learning_rate": 8.306883906993022e-06,
596
+ "loss": 0.5664,
597
  "step": 415
598
  },
599
  {
600
+ "epoch": 0.3468924220524468,
601
+ "grad_norm": 2.129763886062084,
602
+ "learning_rate": 8.25244711212589e-06,
603
+ "loss": 0.575,
604
  "step": 420
605
  },
606
  {
607
+ "epoch": 0.35102209374354737,
608
+ "grad_norm": 2.210580574947652,
609
+ "learning_rate": 8.197333632012123e-06,
610
+ "loss": 0.5797,
611
  "step": 425
612
  },
613
  {
614
+ "epoch": 0.35515176543464794,
615
+ "grad_norm": 2.3295885927377387,
616
+ "learning_rate": 8.141554933242135e-06,
617
+ "loss": 0.569,
618
  "step": 430
619
  },
620
  {
621
+ "epoch": 0.3592814371257485,
622
+ "grad_norm": 2.332203260771466,
623
+ "learning_rate": 8.08512262080787e-06,
624
+ "loss": 0.573,
625
  "step": 435
626
  },
627
  {
628
+ "epoch": 0.36341110881684907,
629
+ "grad_norm": 2.13902847391916,
630
+ "learning_rate": 8.028048435688333e-06,
631
+ "loss": 0.5788,
632
  "step": 440
633
  },
634
  {
635
+ "epoch": 0.36754078050794964,
636
+ "grad_norm": 2.3592904701751145,
637
+ "learning_rate": 7.970344252406832e-06,
638
+ "loss": 0.569,
639
  "step": 445
640
  },
641
  {
642
+ "epoch": 0.37167045219905015,
643
+ "grad_norm": 2.4900138472719617,
644
+ "learning_rate": 7.912022076560426e-06,
645
+ "loss": 0.5697,
646
  "step": 450
647
  },
648
  {
649
+ "epoch": 0.3758001238901507,
650
+ "grad_norm": 2.1966947004630826,
651
+ "learning_rate": 7.853094042322121e-06,
652
+ "loss": 0.5577,
653
  "step": 455
654
  },
655
  {
656
+ "epoch": 0.3799297955812513,
657
+ "grad_norm": 2.4924992867127,
658
+ "learning_rate": 7.7935724099163e-06,
659
+ "loss": 0.5564,
660
  "step": 460
661
  },
662
  {
663
+ "epoch": 0.38405946727235185,
664
+ "grad_norm": 2.3168755091791775,
665
+ "learning_rate": 7.733469563067928e-06,
666
+ "loss": 0.558,
667
  "step": 465
668
  },
669
  {
670
+ "epoch": 0.3881891389634524,
671
+ "grad_norm": 2.090377992450065,
672
+ "learning_rate": 7.67279800642607e-06,
673
+ "loss": 0.5472,
674
  "step": 470
675
  },
676
  {
677
+ "epoch": 0.392318810654553,
678
+ "grad_norm": 2.19832848896829,
679
+ "learning_rate": 7.611570362962247e-06,
680
+ "loss": 0.547,
681
  "step": 475
682
  },
683
  {
684
+ "epoch": 0.3964484823456535,
685
+ "grad_norm": 2.139436532548822,
686
+ "learning_rate": 7.549799371344175e-06,
687
+ "loss": 0.5376,
688
  "step": 480
689
  },
690
  {
691
+ "epoch": 0.40057815403675406,
692
+ "grad_norm": 2.1648898845336304,
693
+ "learning_rate": 7.487497883285428e-06,
694
+ "loss": 0.5435,
695
  "step": 485
696
  },
697
  {
698
+ "epoch": 0.40470782572785463,
699
+ "grad_norm": 2.1434615200482403,
700
+ "learning_rate": 7.424678860871584e-06,
701
+ "loss": 0.5347,
702
  "step": 490
703
  },
704
  {
705
+ "epoch": 0.4088374974189552,
706
+ "grad_norm": 2.028970353253071,
707
+ "learning_rate": 7.361355373863415e-06,
708
+ "loss": 0.5349,
709
  "step": 495
710
  },
711
  {
712
+ "epoch": 0.41296716911005577,
713
+ "grad_norm": 2.1772432202945624,
714
+ "learning_rate": 7.297540596977663e-06,
715
+ "loss": 0.532,
716
  "step": 500
717
  },
718
  {
719
+ "epoch": 0.41709684080115633,
720
+ "grad_norm": 2.2430530701967832,
721
+ "learning_rate": 7.233247807145989e-06,
722
+ "loss": 0.5385,
723
  "step": 505
724
  },
725
  {
726
+ "epoch": 0.42122651249225684,
727
+ "grad_norm": 2.0619754497698035,
728
+ "learning_rate": 7.168490380752648e-06,
729
+ "loss": 0.5349,
730
  "step": 510
731
  },
732
  {
733
+ "epoch": 0.4253561841833574,
734
+ "grad_norm": 2.1855473382543993,
735
+ "learning_rate": 7.103281790851483e-06,
736
+ "loss": 0.5233,
737
  "step": 515
738
  },
739
  {
740
+ "epoch": 0.429485855874458,
741
+ "grad_norm": 2.2538669231669712,
742
+ "learning_rate": 7.037635604362786e-06,
743
+ "loss": 0.5283,
744
  "step": 520
745
  },
746
  {
747
+ "epoch": 0.43361552756555855,
748
+ "grad_norm": 2.2135151111234794,
749
+ "learning_rate": 6.971565479250659e-06,
750
+ "loss": 0.5167,
751
  "step": 525
752
  },
753
  {
754
+ "epoch": 0.4377451992566591,
755
+ "grad_norm": 2.383590881907924,
756
+ "learning_rate": 6.905085161681408e-06,
757
+ "loss": 0.5219,
758
  "step": 530
759
  },
760
  {
761
+ "epoch": 0.4418748709477597,
762
+ "grad_norm": 2.3721647738464258,
763
+ "learning_rate": 6.838208483163601e-06,
764
+ "loss": 0.5174,
765
  "step": 535
766
  },
767
  {
768
+ "epoch": 0.4460045426388602,
769
+ "grad_norm": 2.338909227744761,
770
+ "learning_rate": 6.770949357670358e-06,
771
+ "loss": 0.5273,
772
  "step": 540
773
  },
774
  {
775
+ "epoch": 0.45013421432996076,
776
+ "grad_norm": 2.1495131685935163,
777
+ "learning_rate": 6.703321778744495e-06,
778
+ "loss": 0.5186,
779
  "step": 545
780
  },
781
  {
782
+ "epoch": 0.4542638860210613,
783
+ "grad_norm": 2.010655198436141,
784
+ "learning_rate": 6.635339816587109e-06,
785
+ "loss": 0.5164,
786
  "step": 550
787
  },
788
  {
789
+ "epoch": 0.4583935577121619,
790
+ "grad_norm": 2.1095925263837385,
791
+ "learning_rate": 6.5670176151302136e-06,
792
+ "loss": 0.4991,
793
+ "step": 555
 
794
  },
795
  {
796
+ "epoch": 0.46252322940326246,
797
+ "grad_norm": 2.2264319027764383,
798
+ "learning_rate": 6.4983693890940335e-06,
799
+ "loss": 0.5195,
800
+ "step": 560
801
+ },
802
+ {
803
+ "epoch": 0.46665290109436297,
804
+ "grad_norm": 2.1349760132456246,
805
+ "learning_rate": 6.4294094210295725e-06,
806
+ "loss": 0.504,
807
+ "step": 565
808
+ },
809
+ {
810
+ "epoch": 0.47078257278546354,
811
+ "grad_norm": 2.2896775214795033,
812
+ "learning_rate": 6.360152058347068e-06,
813
+ "loss": 0.5078,
814
+ "step": 570
815
+ },
816
+ {
817
+ "epoch": 0.4749122444765641,
818
+ "grad_norm": 2.1221923536777747,
819
+ "learning_rate": 6.290611710330957e-06,
820
+ "loss": 0.5059,
821
+ "step": 575
822
+ },
823
+ {
824
+ "epoch": 0.47904191616766467,
825
+ "grad_norm": 2.1068836258621406,
826
+ "learning_rate": 6.2208028451419575e-06,
827
+ "loss": 0.5083,
828
+ "step": 580
829
+ },
830
+ {
831
+ "epoch": 0.48317158785876524,
832
+ "grad_norm": 2.0972952063729475,
833
+ "learning_rate": 6.150739986806911e-06,
834
+ "loss": 0.4923,
835
+ "step": 585
836
+ },
837
+ {
838
+ "epoch": 0.4873012595498658,
839
+ "grad_norm": 2.233487162462348,
840
+ "learning_rate": 6.0804377121969985e-06,
841
+ "loss": 0.4953,
842
+ "step": 590
843
+ },
844
+ {
845
+ "epoch": 0.4914309312409663,
846
+ "grad_norm": 2.108166108174955,
847
+ "learning_rate": 6.009910647994956e-06,
848
+ "loss": 0.4918,
849
+ "step": 595
850
+ },
851
+ {
852
+ "epoch": 0.4955606029320669,
853
+ "grad_norm": 2.055777194538188,
854
+ "learning_rate": 5.939173467651942e-06,
855
+ "loss": 0.483,
856
+ "step": 600
857
+ },
858
+ {
859
+ "epoch": 0.49969027462316745,
860
+ "grad_norm": 2.1437965393063374,
861
+ "learning_rate": 5.8682408883346535e-06,
862
+ "loss": 0.4847,
863
+ "step": 605
864
+ },
865
+ {
866
+ "epoch": 0.503819946314268,
867
+ "grad_norm": 2.019446515641709,
868
+ "learning_rate": 5.7971276678633625e-06,
869
+ "loss": 0.4935,
870
+ "step": 610
871
+ },
872
+ {
873
+ "epoch": 0.5079496180053685,
874
+ "grad_norm": 2.1084072043678894,
875
+ "learning_rate": 5.725848601641492e-06,
876
+ "loss": 0.4736,
877
+ "step": 615
878
+ },
879
+ {
880
+ "epoch": 0.5120792896964691,
881
+ "grad_norm": 2.185992748799692,
882
+ "learning_rate": 5.654418519577369e-06,
883
+ "loss": 0.4743,
884
+ "step": 620
885
+ },
886
+ {
887
+ "epoch": 0.5162089613875697,
888
+ "grad_norm": 2.0909394304010416,
889
+ "learning_rate": 5.5828522829987965e-06,
890
+ "loss": 0.4808,
891
+ "step": 625
892
+ },
893
+ {
894
+ "epoch": 0.5203386330786702,
895
+ "grad_norm": 2.058129072039577,
896
+ "learning_rate": 5.511164781561096e-06,
897
+ "loss": 0.4727,
898
+ "step": 630
899
+ },
900
+ {
901
+ "epoch": 0.5244683047697708,
902
+ "grad_norm": 2.1781635883341837,
903
+ "learning_rate": 5.439370930149252e-06,
904
+ "loss": 0.4665,
905
+ "step": 635
906
+ },
907
+ {
908
+ "epoch": 0.5285979764608714,
909
+ "grad_norm": 2.132119418871688,
910
+ "learning_rate": 5.367485665774802e-06,
911
+ "loss": 0.4647,
912
+ "step": 640
913
+ },
914
+ {
915
+ "epoch": 0.5327276481519719,
916
+ "grad_norm": 2.0923424452918953,
917
+ "learning_rate": 5.295523944468137e-06,
918
+ "loss": 0.4752,
919
+ "step": 645
920
+ },
921
+ {
922
+ "epoch": 0.5368573198430725,
923
+ "grad_norm": 2.1500817775070704,
924
+ "learning_rate": 5.223500738166837e-06,
925
+ "loss": 0.4653,
926
+ "step": 650
927
+ },
928
+ {
929
+ "epoch": 0.5409869915341731,
930
+ "grad_norm": 2.3057702591417764,
931
+ "learning_rate": 5.1514310316006835e-06,
932
+ "loss": 0.459,
933
+ "step": 655
934
+ },
935
+ {
936
+ "epoch": 0.5451166632252736,
937
+ "grad_norm": 2.027442727638895,
938
+ "learning_rate": 5.07932981917404e-06,
939
+ "loss": 0.467,
940
+ "step": 660
941
+ },
942
+ {
943
+ "epoch": 0.5492463349163742,
944
+ "grad_norm": 2.003781400613107,
945
+ "learning_rate": 5.007212101846194e-06,
946
+ "loss": 0.4534,
947
+ "step": 665
948
+ },
949
+ {
950
+ "epoch": 0.5533760066074747,
951
+ "grad_norm": 2.1173198648186236,
952
+ "learning_rate": 4.935092884010347e-06,
953
+ "loss": 0.4582,
954
+ "step": 670
955
+ },
956
+ {
957
+ "epoch": 0.5575056782985752,
958
+ "grad_norm": 2.1197258530702303,
959
+ "learning_rate": 4.8629871703718844e-06,
960
+ "loss": 0.4484,
961
+ "step": 675
962
+ },
963
+ {
964
+ "epoch": 0.5616353499896758,
965
+ "grad_norm": 2.112938054969618,
966
+ "learning_rate": 4.7909099628265946e-06,
967
+ "loss": 0.4504,
968
+ "step": 680
969
+ },
970
+ {
971
+ "epoch": 0.5657650216807764,
972
+ "grad_norm": 2.1008936035245056,
973
+ "learning_rate": 4.718876257339444e-06,
974
+ "loss": 0.4485,
975
+ "step": 685
976
+ },
977
+ {
978
+ "epoch": 0.5698946933718769,
979
+ "grad_norm": 2.925269693235509,
980
+ "learning_rate": 4.646901040824622e-06,
981
+ "loss": 0.4523,
982
+ "step": 690
983
+ },
984
+ {
985
+ "epoch": 0.5740243650629775,
986
+ "grad_norm": 2.013226878250545,
987
+ "learning_rate": 4.574999288027437e-06,
988
+ "loss": 0.4544,
989
+ "step": 695
990
+ },
991
+ {
992
+ "epoch": 0.5781540367540781,
993
+ "grad_norm": 2.059493099572322,
994
+ "learning_rate": 4.503185958408767e-06,
995
+ "loss": 0.4467,
996
+ "step": 700
997
+ },
998
+ {
999
+ "epoch": 0.5822837084451786,
1000
+ "grad_norm": 2.0049953659179716,
1001
+ "learning_rate": 4.431475993032673e-06,
1002
+ "loss": 0.441,
1003
+ "step": 705
1004
+ },
1005
+ {
1006
+ "epoch": 0.5864133801362792,
1007
+ "grad_norm": 2.1512212806706255,
1008
+ "learning_rate": 4.359884311457857e-06,
1009
+ "loss": 0.4337,
1010
+ "step": 710
1011
+ },
1012
+ {
1013
+ "epoch": 0.5905430518273798,
1014
+ "grad_norm": 2.1177615430048755,
1015
+ "learning_rate": 4.2884258086335755e-06,
1016
+ "loss": 0.4458,
1017
+ "step": 715
1018
+ },
1019
+ {
1020
+ "epoch": 0.5946727235184803,
1021
+ "grad_norm": 2.008192166381139,
1022
+ "learning_rate": 4.217115351800693e-06,
1023
+ "loss": 0.4413,
1024
+ "step": 720
1025
+ },
1026
+ {
1027
+ "epoch": 0.5988023952095808,
1028
+ "grad_norm": 2.063297374873059,
1029
+ "learning_rate": 4.145967777398481e-06,
1030
+ "loss": 0.4344,
1031
+ "step": 725
1032
+ },
1033
+ {
1034
+ "epoch": 0.6029320669006814,
1035
+ "grad_norm": 2.004714733933455,
1036
+ "learning_rate": 4.074997887977843e-06,
1037
+ "loss": 0.4358,
1038
+ "step": 730
1039
+ },
1040
+ {
1041
+ "epoch": 0.6070617385917819,
1042
+ "grad_norm": 2.025598614826924,
1043
+ "learning_rate": 4.004220449121574e-06,
1044
+ "loss": 0.4482,
1045
+ "step": 735
1046
+ },
1047
+ {
1048
+ "epoch": 0.6111914102828825,
1049
+ "grad_norm": 1.938185383839704,
1050
+ "learning_rate": 3.933650186372329e-06,
1051
+ "loss": 0.4337,
1052
+ "step": 740
1053
+ },
1054
+ {
1055
+ "epoch": 0.615321081973983,
1056
+ "grad_norm": 2.0495762719124104,
1057
+ "learning_rate": 3.863301782168896e-06,
1058
+ "loss": 0.4412,
1059
+ "step": 745
1060
+ },
1061
+ {
1062
+ "epoch": 0.6194507536650836,
1063
+ "grad_norm": 2.028634933360134,
1064
+ "learning_rate": 3.7931898727914723e-06,
1065
+ "loss": 0.4293,
1066
+ "step": 750
1067
+ },
1068
+ {
1069
+ "epoch": 0.6235804253561842,
1070
+ "grad_norm": 2.0029138326822404,
1071
+ "learning_rate": 3.7233290453165127e-06,
1072
+ "loss": 0.4304,
1073
+ "step": 755
1074
+ },
1075
+ {
1076
+ "epoch": 0.6277100970472848,
1077
+ "grad_norm": 2.048375432552439,
1078
+ "learning_rate": 3.6537338345818273e-06,
1079
+ "loss": 0.4213,
1080
+ "step": 760
1081
+ },
1082
+ {
1083
+ "epoch": 0.6318397687383853,
1084
+ "grad_norm": 2.0326809490611852,
1085
+ "learning_rate": 3.5844187201625567e-06,
1086
+ "loss": 0.424,
1087
+ "step": 765
1088
+ },
1089
+ {
1090
+ "epoch": 0.6359694404294859,
1091
+ "grad_norm": 2.019159008104501,
1092
+ "learning_rate": 3.5153981233586277e-06,
1093
+ "loss": 0.4228,
1094
+ "step": 770
1095
+ },
1096
+ {
1097
+ "epoch": 0.6400991121205865,
1098
+ "grad_norm": 2.09985714316485,
1099
+ "learning_rate": 3.446686404194337e-06,
1100
+ "loss": 0.4206,
1101
+ "step": 775
1102
+ },
1103
+ {
1104
+ "epoch": 0.644228783811687,
1105
+ "grad_norm": 1.9414886074513862,
1106
+ "learning_rate": 3.3782978584307035e-06,
1107
+ "loss": 0.4173,
1108
+ "step": 780
1109
+ },
1110
+ {
1111
+ "epoch": 0.6483584555027875,
1112
+ "grad_norm": 2.3066100655713755,
1113
+ "learning_rate": 3.310246714591162e-06,
1114
+ "loss": 0.4248,
1115
+ "step": 785
1116
+ },
1117
+ {
1118
+ "epoch": 0.652488127193888,
1119
+ "grad_norm": 2.106953928933072,
1120
+ "learning_rate": 3.2425471310012645e-06,
1121
+ "loss": 0.4159,
1122
+ "step": 790
1123
+ },
1124
+ {
1125
+ "epoch": 0.6566177988849886,
1126
+ "grad_norm": 2.197357084281967,
1127
+ "learning_rate": 3.1752131928429787e-06,
1128
+ "loss": 0.425,
1129
+ "step": 795
1130
+ },
1131
+ {
1132
+ "epoch": 0.6607474705760892,
1133
+ "grad_norm": 1.9729477591043592,
1134
+ "learning_rate": 3.1082589092242116e-06,
1135
+ "loss": 0.4241,
1136
+ "step": 800
1137
+ },
1138
+ {
1139
+ "epoch": 0.6648771422671897,
1140
+ "grad_norm": 2.0442009757093875,
1141
+ "learning_rate": 3.041698210264149e-06,
1142
+ "loss": 0.4238,
1143
+ "step": 805
1144
+ },
1145
+ {
1146
+ "epoch": 0.6690068139582903,
1147
+ "grad_norm": 2.031799210389413,
1148
+ "learning_rate": 2.9755449441950434e-06,
1149
+ "loss": 0.4131,
1150
+ "step": 810
1151
+ },
1152
+ {
1153
+ "epoch": 0.6731364856493909,
1154
+ "grad_norm": 2.070029206474213,
1155
+ "learning_rate": 2.9098128744810245e-06,
1156
+ "loss": 0.4159,
1157
+ "step": 815
1158
+ },
1159
+ {
1160
+ "epoch": 0.6772661573404914,
1161
+ "grad_norm": 2.0394445933724414,
1162
+ "learning_rate": 2.84451567695456e-06,
1163
+ "loss": 0.42,
1164
+ "step": 820
1165
+ },
1166
+ {
1167
+ "epoch": 0.681395829031592,
1168
+ "grad_norm": 2.0377377910207657,
1169
+ "learning_rate": 2.7796669369711294e-06,
1170
+ "loss": 0.4095,
1171
+ "step": 825
1172
+ },
1173
+ {
1174
+ "epoch": 0.6855255007226926,
1175
+ "grad_norm": 2.0626337564803965,
1176
+ "learning_rate": 2.715280146582752e-06,
1177
+ "loss": 0.4069,
1178
+ "step": 830
1179
+ },
1180
+ {
1181
+ "epoch": 0.6896551724137931,
1182
+ "grad_norm": 2.0040834447649454,
1183
+ "learning_rate": 2.651368701730889e-06,
1184
+ "loss": 0.41,
1185
+ "step": 835
1186
+ },
1187
+ {
1188
+ "epoch": 0.6937848441048936,
1189
+ "grad_norm": 1.917041220777535,
1190
+ "learning_rate": 2.5879458994593786e-06,
1191
+ "loss": 0.4074,
1192
+ "step": 840
1193
+ },
1194
+ {
1195
+ "epoch": 0.6979145157959942,
1196
+ "grad_norm": 2.0269668889490515,
1197
+ "learning_rate": 2.5250249351479206e-06,
1198
+ "loss": 0.4081,
1199
+ "step": 845
1200
+ },
1201
+ {
1202
+ "epoch": 0.7020441874870947,
1203
+ "grad_norm": 1.9753046952791828,
1204
+ "learning_rate": 2.4626188997667224e-06,
1205
+ "loss": 0.4018,
1206
+ "step": 850
1207
+ },
1208
+ {
1209
+ "epoch": 0.7061738591781953,
1210
+ "grad_norm": 2.014280374102551,
1211
+ "learning_rate": 2.400740777152874e-06,
1212
+ "loss": 0.4063,
1213
+ "step": 855
1214
+ },
1215
+ {
1216
+ "epoch": 0.7103035308692959,
1217
+ "grad_norm": 2.026632120733398,
1218
+ "learning_rate": 2.3394034413090015e-06,
1219
+ "loss": 0.3955,
1220
+ "step": 860
1221
+ },
1222
+ {
1223
+ "epoch": 0.7144332025603964,
1224
+ "grad_norm": 2.088872002071019,
1225
+ "learning_rate": 2.278619653724781e-06,
1226
+ "loss": 0.4038,
1227
+ "step": 865
1228
+ },
1229
+ {
1230
+ "epoch": 0.718562874251497,
1231
+ "grad_norm": 2.096103572784026,
1232
+ "learning_rate": 2.218402060721845e-06,
1233
+ "loss": 0.3921,
1234
+ "step": 870
1235
+ },
1236
+ {
1237
+ "epoch": 0.7226925459425976,
1238
+ "grad_norm": 1.9266065409655262,
1239
+ "learning_rate": 2.1587631908226812e-06,
1240
+ "loss": 0.4085,
1241
+ "step": 875
1242
+ },
1243
+ {
1244
+ "epoch": 0.7268222176336981,
1245
+ "grad_norm": 1.9536655952796587,
1246
+ "learning_rate": 2.09971545214401e-06,
1247
+ "loss": 0.3965,
1248
+ "step": 880
1249
+ },
1250
+ {
1251
+ "epoch": 0.7309518893247987,
1252
+ "grad_norm": 2.061222200669705,
1253
+ "learning_rate": 2.04127112981522e-06,
1254
+ "loss": 0.3941,
1255
+ "step": 885
1256
+ },
1257
+ {
1258
+ "epoch": 0.7350815610158993,
1259
+ "grad_norm": 2.010203957849096,
1260
+ "learning_rate": 1.9834423834224014e-06,
1261
+ "loss": 0.3982,
1262
+ "step": 890
1263
+ },
1264
+ {
1265
+ "epoch": 0.7392112327069998,
1266
+ "grad_norm": 1.9380026785117144,
1267
+ "learning_rate": 1.926241244478496e-06,
1268
+ "loss": 0.3965,
1269
+ "step": 895
1270
+ },
1271
+ {
1272
+ "epoch": 0.7433409043981003,
1273
+ "grad_norm": 2.0416911208044244,
1274
+ "learning_rate": 1.8696796139200852e-06,
1275
+ "loss": 0.4015,
1276
+ "step": 900
1277
+ },
1278
+ {
1279
+ "epoch": 0.7474705760892009,
1280
+ "grad_norm": 2.040772561659114,
1281
+ "learning_rate": 1.8137692596313528e-06,
1282
+ "loss": 0.3827,
1283
+ "step": 905
1284
+ },
1285
+ {
1286
+ "epoch": 0.7516002477803014,
1287
+ "grad_norm": 1.9118935892662992,
1288
+ "learning_rate": 1.7585218139957205e-06,
1289
+ "loss": 0.3895,
1290
+ "step": 910
1291
+ },
1292
+ {
1293
+ "epoch": 0.755729919471402,
1294
+ "grad_norm": 1.927575457582737,
1295
+ "learning_rate": 1.7039487714756953e-06,
1296
+ "loss": 0.3965,
1297
+ "step": 915
1298
+ },
1299
+ {
1300
+ "epoch": 0.7598595911625026,
1301
+ "grad_norm": 1.9797748596618283,
1302
+ "learning_rate": 1.6500614862213866e-06,
1303
+ "loss": 0.3937,
1304
+ "step": 920
1305
+ },
1306
+ {
1307
+ "epoch": 0.7639892628536031,
1308
+ "grad_norm": 2.05432833461262,
1309
+ "learning_rate": 1.596871169708235e-06,
1310
+ "loss": 0.3891,
1311
+ "step": 925
1312
+ },
1313
+ {
1314
+ "epoch": 0.7681189345447037,
1315
+ "grad_norm": 1.9501310111794177,
1316
+ "learning_rate": 1.54438888840441e-06,
1317
+ "loss": 0.3801,
1318
+ "step": 930
1319
+ },
1320
+ {
1321
+ "epoch": 0.7722486062358043,
1322
+ "grad_norm": 1.970421056222239,
1323
+ "learning_rate": 1.4926255614683931e-06,
1324
+ "loss": 0.3806,
1325
+ "step": 935
1326
+ },
1327
+ {
1328
+ "epoch": 0.7763782779269048,
1329
+ "grad_norm": 2.1277167725190598,
1330
+ "learning_rate": 1.4415919584771999e-06,
1331
+ "loss": 0.3843,
1332
+ "step": 940
1333
+ },
1334
+ {
1335
+ "epoch": 0.7805079496180054,
1336
+ "grad_norm": 1.9784981959186678,
1337
+ "learning_rate": 1.3912986971857168e-06,
1338
+ "loss": 0.3828,
1339
+ "step": 945
1340
+ },
1341
+ {
1342
+ "epoch": 0.784637621309106,
1343
+ "grad_norm": 2.0003697611347975,
1344
+ "learning_rate": 1.3417562413176405e-06,
1345
+ "loss": 0.3859,
1346
+ "step": 950
1347
+ },
1348
+ {
1349
+ "epoch": 0.7887672930002065,
1350
+ "grad_norm": 2.0785112893490485,
1351
+ "learning_rate": 1.292974898388456e-06,
1352
+ "loss": 0.3812,
1353
+ "step": 955
1354
+ },
1355
+ {
1356
+ "epoch": 0.792896964691307,
1357
+ "grad_norm": 2.045681460940008,
1358
+ "learning_rate": 1.2449648175609115e-06,
1359
+ "loss": 0.3794,
1360
+ "step": 960
1361
+ },
1362
+ {
1363
+ "epoch": 0.7970266363824076,
1364
+ "grad_norm": 2.006461021136274,
1365
+ "learning_rate": 1.1977359875334454e-06,
1366
+ "loss": 0.3867,
1367
+ "step": 965
1368
+ },
1369
+ {
1370
+ "epoch": 0.8011563080735081,
1371
+ "grad_norm": 2.0404335899757187,
1372
+ "learning_rate": 1.1512982344619904e-06,
1373
+ "loss": 0.3845,
1374
+ "step": 970
1375
+ },
1376
+ {
1377
+ "epoch": 0.8052859797646087,
1378
+ "grad_norm": 2.1247685897806186,
1379
+ "learning_rate": 1.1056612199156093e-06,
1380
+ "loss": 0.3838,
1381
+ "step": 975
1382
+ },
1383
+ {
1384
+ "epoch": 0.8094156514557093,
1385
+ "grad_norm": 2.059455987291636,
1386
+ "learning_rate": 1.0608344388663576e-06,
1387
+ "loss": 0.3769,
1388
+ "step": 980
1389
+ },
1390
+ {
1391
+ "epoch": 0.8135453231468098,
1392
+ "grad_norm": 2.010190102670995,
1393
+ "learning_rate": 1.01682721771382e-06,
1394
+ "loss": 0.38,
1395
+ "step": 985
1396
+ },
1397
+ {
1398
+ "epoch": 0.8176749948379104,
1399
+ "grad_norm": 2.1310463662484755,
1400
+ "learning_rate": 9.73648712344707e-07,
1401
+ "loss": 0.3727,
1402
+ "step": 990
1403
+ },
1404
+ {
1405
+ "epoch": 0.821804666529011,
1406
+ "grad_norm": 2.026722500514339,
1407
+ "learning_rate": 9.313079062279429e-07,
1408
+ "loss": 0.3775,
1409
+ "step": 995
1410
+ },
1411
+ {
1412
+ "epoch": 0.8259343382201115,
1413
+ "grad_norm": 1.98494077041973,
1414
+ "learning_rate": 8.898136085456127e-07,
1415
+ "loss": 0.3795,
1416
+ "step": 1000
1417
+ },
1418
+ {
1419
+ "epoch": 0.8300640099112121,
1420
+ "grad_norm": 2.0710248547928267,
1421
+ "learning_rate": 8.491744523601741e-07,
1422
+ "loss": 0.3789,
1423
+ "step": 1005
1424
+ },
1425
+ {
1426
+ "epoch": 0.8341936816023127,
1427
+ "grad_norm": 2.052056544033471,
1428
+ "learning_rate": 8.093988928183127e-07,
1429
+ "loss": 0.376,
1430
+ "step": 1010
1431
+ },
1432
+ {
1433
+ "epoch": 0.8383233532934131,
1434
+ "grad_norm": 2.0064833870293612,
1435
+ "learning_rate": 7.70495205391818e-07,
1436
+ "loss": 0.3728,
1437
+ "step": 1015
1438
+ },
1439
+ {
1440
+ "epoch": 0.8424530249845137,
1441
+ "grad_norm": 2.032309328318966,
1442
+ "learning_rate": 7.324714841558322e-07,
1443
+ "loss": 0.3701,
1444
+ "step": 1020
1445
+ },
1446
+ {
1447
+ "epoch": 0.8465826966756143,
1448
+ "grad_norm": 1.976437679829882,
1449
+ "learning_rate": 6.953356401048466e-07,
1450
+ "loss": 0.3796,
1451
+ "step": 1025
1452
+ },
1453
+ {
1454
+ "epoch": 0.8507123683667148,
1455
+ "grad_norm": 2.0095046825383927,
1456
+ "learning_rate": 6.590953995067812e-07,
1457
+ "loss": 0.3718,
1458
+ "step": 1030
1459
+ },
1460
+ {
1461
+ "epoch": 0.8548420400578154,
1462
+ "grad_norm": 2.1362491401601598,
1463
+ "learning_rate": 6.237583022955079e-07,
1464
+ "loss": 0.3669,
1465
+ "step": 1035
1466
+ },
1467
+ {
1468
+ "epoch": 0.858971711748916,
1469
+ "grad_norm": 2.030103317522303,
1470
+ "learning_rate": 5.89331700502136e-07,
1471
+ "loss": 0.3674,
1472
+ "step": 1040
1473
+ },
1474
+ {
1475
+ "epoch": 0.8631013834400165,
1476
+ "grad_norm": 2.0320527842019316,
1477
+ "learning_rate": 5.558227567253832e-07,
1478
+ "loss": 0.3679,
1479
+ "step": 1045
1480
+ },
1481
+ {
1482
+ "epoch": 0.8672310551311171,
1483
+ "grad_norm": 2.1203806417251214,
1484
+ "learning_rate": 5.232384426413772e-07,
1485
+ "loss": 0.3697,
1486
+ "step": 1050
1487
+ },
1488
+ {
1489
+ "epoch": 0.8713607268222177,
1490
+ "grad_norm": 1.9029660119177207,
1491
+ "learning_rate": 4.915855375531642e-07,
1492
+ "loss": 0.3708,
1493
+ "step": 1055
1494
+ },
1495
+ {
1496
+ "epoch": 0.8754903985133182,
1497
+ "grad_norm": 1.9598889356044642,
1498
+ "learning_rate": 4.608706269802471e-07,
1499
+ "loss": 0.3641,
1500
+ "step": 1060
1501
+ },
1502
+ {
1503
+ "epoch": 0.8796200702044188,
1504
+ "grad_norm": 2.015130495239755,
1505
+ "learning_rate": 4.3110010128843747e-07,
1506
+ "loss": 0.3659,
1507
+ "step": 1065
1508
+ },
1509
+ {
1510
+ "epoch": 0.8837497418955194,
1511
+ "grad_norm": 1.9290620704097432,
1512
+ "learning_rate": 4.022801543603194e-07,
1513
+ "loss": 0.3648,
1514
+ "step": 1070
1515
+ },
1516
+ {
1517
+ "epoch": 0.8878794135866198,
1518
+ "grad_norm": 2.131063629467497,
1519
+ "learning_rate": 3.744167823065814e-07,
1520
+ "loss": 0.3619,
1521
+ "step": 1075
1522
+ },
1523
+ {
1524
+ "epoch": 0.8920090852777204,
1525
+ "grad_norm": 2.007895538821572,
1526
+ "learning_rate": 3.4751578221850377e-07,
1527
+ "loss": 0.3729,
1528
+ "step": 1080
1529
+ },
1530
+ {
1531
+ "epoch": 0.896138756968821,
1532
+ "grad_norm": 2.0860597695484113,
1533
+ "learning_rate": 3.2158275096184744e-07,
1534
+ "loss": 0.3674,
1535
+ "step": 1085
1536
+ },
1537
+ {
1538
+ "epoch": 0.9002684286599215,
1539
+ "grad_norm": 2.0200760244926044,
1540
+ "learning_rate": 2.966230840124007e-07,
1541
+ "loss": 0.3677,
1542
+ "step": 1090
1543
+ },
1544
+ {
1545
+ "epoch": 0.9043981003510221,
1546
+ "grad_norm": 2.0288061050580053,
1547
+ "learning_rate": 2.72641974333433e-07,
1548
+ "loss": 0.3698,
1549
+ "step": 1095
1550
+ },
1551
+ {
1552
+ "epoch": 0.9085277720421226,
1553
+ "grad_norm": 2.0237726744767,
1554
+ "learning_rate": 2.4964441129527337e-07,
1555
+ "loss": 0.3622,
1556
+ "step": 1100
1557
+ },
1558
+ {
1559
+ "epoch": 0.9126574437332232,
1560
+ "grad_norm": 1.9537324233727318,
1561
+ "learning_rate": 2.2763517963725169e-07,
1562
+ "loss": 0.3678,
1563
+ "step": 1105
1564
+ },
1565
+ {
1566
+ "epoch": 0.9167871154243238,
1567
+ "grad_norm": 2.0620765585117837,
1568
+ "learning_rate": 2.0661885847221398e-07,
1569
+ "loss": 0.3698,
1570
+ "step": 1110
1571
+ },
1572
+ {
1573
+ "epoch": 0.9209167871154244,
1574
+ "grad_norm": 1.9666263090601819,
1575
+ "learning_rate": 1.8659982033381928e-07,
1576
+ "loss": 0.3615,
1577
+ "step": 1115
1578
+ },
1579
+ {
1580
+ "epoch": 0.9250464588065249,
1581
+ "grad_norm": 2.0277290304841142,
1582
+ "learning_rate": 1.6758223026681507e-07,
1583
+ "loss": 0.3613,
1584
+ "step": 1120
1585
+ },
1586
+ {
1587
+ "epoch": 0.9291761304976255,
1588
+ "grad_norm": 2.106138721077295,
1589
+ "learning_rate": 1.4957004496048256e-07,
1590
+ "loss": 0.3621,
1591
+ "step": 1125
1592
+ },
1593
+ {
1594
+ "epoch": 0.9333058021887259,
1595
+ "grad_norm": 1.9942405326271708,
1596
+ "learning_rate": 1.3256701192542853e-07,
1597
+ "loss": 0.3659,
1598
+ "step": 1130
1599
+ },
1600
+ {
1601
+ "epoch": 0.9374354738798265,
1602
+ "grad_norm": 1.9945899501208246,
1603
+ "learning_rate": 1.1657666871390471e-07,
1604
+ "loss": 0.3671,
1605
+ "step": 1135
1606
+ },
1607
+ {
1608
+ "epoch": 0.9415651455709271,
1609
+ "grad_norm": 1.8785796566381439,
1610
+ "learning_rate": 1.0160234218380305e-07,
1611
+ "loss": 0.3593,
1612
+ "step": 1140
1613
+ },
1614
+ {
1615
+ "epoch": 0.9456948172620276,
1616
+ "grad_norm": 2.025291462525694,
1617
+ "learning_rate": 8.76471478064872e-08,
1618
+ "loss": 0.3723,
1619
+ "step": 1145
1620
+ },
1621
+ {
1622
+ "epoch": 0.9498244889531282,
1623
+ "grad_norm": 1.9718773659356506,
1624
+ "learning_rate": 7.471398901860772e-08,
1625
+ "loss": 0.3665,
1626
+ "step": 1150
1627
+ },
1628
+ {
1629
+ "epoch": 0.9539541606442288,
1630
+ "grad_norm": 1.9769770127109265,
1631
+ "learning_rate": 6.280555661802857e-08,
1632
+ "loss": 0.3646,
1633
+ "step": 1155
1634
+ },
1635
+ {
1636
+ "epoch": 0.9580838323353293,
1637
+ "grad_norm": 2.0267147785284716,
1638
+ "learning_rate": 5.192432820399718e-08,
1639
+ "loss": 0.3657,
1640
+ "step": 1160
1641
+ },
1642
+ {
1643
+ "epoch": 0.9622135040264299,
1644
+ "grad_norm": 2.07581128095638,
1645
+ "learning_rate": 4.207256766166845e-08,
1646
+ "loss": 0.3565,
1647
+ "step": 1165
1648
+ },
1649
+ {
1650
+ "epoch": 0.9663431757175305,
1651
+ "grad_norm": 2.0255359776120065,
1652
+ "learning_rate": 3.3252324691093185e-08,
1653
+ "loss": 0.3616,
1654
+ "step": 1170
1655
+ },
1656
+ {
1657
+ "epoch": 0.970472847408631,
1658
+ "grad_norm": 2.0868262303852396,
1659
+ "learning_rate": 2.546543438077087e-08,
1660
+ "loss": 0.3683,
1661
+ "step": 1175
1662
+ },
1663
+ {
1664
+ "epoch": 0.9746025190997316,
1665
+ "grad_norm": 2.0003015972698504,
1666
+ "learning_rate": 1.8713516825851207e-08,
1667
+ "loss": 0.371,
1668
+ "step": 1180
1669
+ },
1670
+ {
1671
+ "epoch": 0.9787321907908322,
1672
+ "grad_norm": 1.9827933568645475,
1673
+ "learning_rate": 1.2997976791065403e-08,
1674
+ "loss": 0.3549,
1675
+ "step": 1185
1676
+ },
1677
+ {
1678
+ "epoch": 0.9828618624819326,
1679
+ "grad_norm": 2.020348792034948,
1680
+ "learning_rate": 8.32000341846162e-09,
1681
+ "loss": 0.3645,
1682
+ "step": 1190
1683
+ },
1684
+ {
1685
+ "epoch": 0.9869915341730332,
1686
+ "grad_norm": 1.9998961679855551,
1687
+ "learning_rate": 4.6805699799967744e-09,
1688
+ "loss": 0.3624,
1689
+ "step": 1195
1690
+ },
1691
+ {
1692
+ "epoch": 0.9911212058641338,
1693
+ "grad_norm": 1.952421429252492,
1694
+ "learning_rate": 2.0804336750429588e-09,
1695
+ "loss": 0.3733,
1696
+ "step": 1200
1697
+ },
1698
+ {
1699
+ "epoch": 0.9952508775552343,
1700
+ "grad_norm": 2.101729255399416,
1701
+ "learning_rate": 5.201354728517905e-10,
1702
+ "loss": 0.3664,
1703
+ "step": 1205
1704
+ },
1705
+ {
1706
+ "epoch": 0.9993805492463349,
1707
+ "grad_norm": 2.1247280097277037,
1708
+ "learning_rate": 0.0,
1709
+ "loss": 0.3624,
1710
+ "step": 1210
1711
+ },
1712
+ {
1713
+ "epoch": 0.9993805492463349,
1714
+ "eval_loss": 0.3181818723678589,
1715
+ "eval_runtime": 1.1842,
1716
+ "eval_samples_per_second": 2.533,
1717
+ "eval_steps_per_second": 0.844,
1718
+ "step": 1210
1719
+ },
1720
+ {
1721
+ "epoch": 0.9993805492463349,
1722
+ "step": 1210,
1723
+ "total_flos": 253297038458880.0,
1724
+ "train_loss": 0.5432274522860188,
1725
+ "train_runtime": 27235.9072,
1726
+ "train_samples_per_second": 1.422,
1727
+ "train_steps_per_second": 0.044
1728
  }
1729
  ],
1730
  "logging_steps": 5,
1731
+ "max_steps": 1210,
1732
  "num_input_tokens_seen": 0,
1733
  "num_train_epochs": 1,
1734
  "save_steps": 100,
 
1744
  "attributes": {}
1745
  }
1746
  },
1747
+ "total_flos": 253297038458880.0,
1748
+ "train_batch_size": 2,
1749
  "trial_name": null,
1750
  "trial_params": null
1751
  }