pszemraj commited on
Commit
3739f98
1 Parent(s): cbf42df

End of training

Browse files
Files changed (5) hide show
  1. README.md +3 -1
  2. all_results.json +13 -13
  3. eval_results.json +9 -9
  4. train_results.json +4 -4
  5. trainer_state.json +253 -253
README.md CHANGED
@@ -2,6 +2,8 @@
2
  license: apache-2.0
3
  base_model: facebook/convnextv2-nano-22k-384
4
  tags:
 
 
5
  - generated_from_trainer
6
  metrics:
7
  - accuracy
@@ -19,7 +21,7 @@ should probably proofread and complete it, then remove this comment. -->
19
 
20
  # convnextv2-nano-22k-384-boulderspot-vN
21
 
22
- This model is a fine-tuned version of [facebook/convnextv2-nano-22k-384](https://huggingface.co/facebook/convnextv2-nano-22k-384) on an unknown dataset.
23
  It achieves the following results on the evaluation set:
24
  - Loss: 0.0340
25
  - Accuracy: 0.9883
 
2
  license: apache-2.0
3
  base_model: facebook/convnextv2-nano-22k-384
4
  tags:
5
+ - image-classification
6
+ - vision
7
  - generated_from_trainer
8
  metrics:
9
  - accuracy
 
21
 
22
  # convnextv2-nano-22k-384-boulderspot-vN
23
 
24
+ This model is a fine-tuned version of [facebook/convnextv2-nano-22k-384](https://huggingface.co/facebook/convnextv2-nano-22k-384) on the pszemraj/boulderspot dataset.
25
  It achieves the following results on the evaluation set:
26
  - Loss: 0.0340
27
  - Accuracy: 0.9883
all_results.json CHANGED
@@ -1,16 +1,16 @@
1
  {
2
  "epoch": 4.99,
3
- "eval_accuracy": 0.9780701754385965,
4
- "eval_f1": 0.9787695402282305,
5
- "eval_loss": 0.05532665178179741,
6
- "eval_matthews_correlation": 0.8420667383515437,
7
- "eval_precision": 0.9800898852068441,
8
- "eval_recall": 0.9780701754385965,
9
- "eval_runtime": 5.9704,
10
- "eval_samples_per_second": 114.565,
11
- "eval_steps_per_second": 7.202,
12
- "train_loss": 0.08238832812591139,
13
- "train_runtime": 705.5168,
14
- "train_samples_per_second": 92.096,
15
- "train_steps_per_second": 1.439
16
  }
 
1
  {
2
  "epoch": 4.99,
3
+ "eval_accuracy": 0.9883040935672515,
4
+ "eval_f1": 0.9883040935672515,
5
+ "eval_loss": 0.03396734222769737,
6
+ "eval_matthews_correlation": 0.8962181845768691,
7
+ "eval_precision": 0.9883040935672515,
8
+ "eval_recall": 0.9883040935672515,
9
+ "eval_runtime": 5.0986,
10
+ "eval_samples_per_second": 134.153,
11
+ "eval_steps_per_second": 8.434,
12
+ "train_loss": 0.08031210996247277,
13
+ "train_runtime": 701.9998,
14
+ "train_samples_per_second": 92.557,
15
+ "train_steps_per_second": 1.446
16
  }
eval_results.json CHANGED
@@ -1,12 +1,12 @@
1
  {
2
  "epoch": 4.99,
3
- "eval_accuracy": 0.9780701754385965,
4
- "eval_f1": 0.9787695402282305,
5
- "eval_loss": 0.05532665178179741,
6
- "eval_matthews_correlation": 0.8420667383515437,
7
- "eval_precision": 0.9800898852068441,
8
- "eval_recall": 0.9780701754385965,
9
- "eval_runtime": 5.9704,
10
- "eval_samples_per_second": 114.565,
11
- "eval_steps_per_second": 7.202
12
  }
 
1
  {
2
  "epoch": 4.99,
3
+ "eval_accuracy": 0.9883040935672515,
4
+ "eval_f1": 0.9883040935672515,
5
+ "eval_loss": 0.03396734222769737,
6
+ "eval_matthews_correlation": 0.8962181845768691,
7
+ "eval_precision": 0.9883040935672515,
8
+ "eval_recall": 0.9883040935672515,
9
+ "eval_runtime": 5.0986,
10
+ "eval_samples_per_second": 134.153,
11
+ "eval_steps_per_second": 8.434
12
  }
train_results.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "epoch": 4.99,
3
- "train_loss": 0.08238832812591139,
4
- "train_runtime": 705.5168,
5
- "train_samples_per_second": 92.096,
6
- "train_steps_per_second": 1.439
7
  }
 
1
  {
2
  "epoch": 4.99,
3
+ "train_loss": 0.08031210996247277,
4
+ "train_runtime": 701.9998,
5
+ "train_samples_per_second": 92.557,
6
+ "train_steps_per_second": 1.446
7
  }
trainer_state.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
- "best_metric": 0.9787695402282305,
3
- "best_model_checkpoint": "./outputs/convnextv2-nano-22k-384-boulderspot-vN/checkpoint-609",
4
  "epoch": 4.993849938499385,
5
  "eval_steps": 500,
6
  "global_step": 1015,
@@ -10,784 +10,784 @@
10
  "log_history": [
11
  {
12
  "epoch": 0.05,
13
- "grad_norm": 28.485111236572266,
14
  "learning_rate": 3.92156862745098e-06,
15
- "loss": 0.8997,
16
  "step": 10
17
  },
18
  {
19
  "epoch": 0.1,
20
- "grad_norm": 14.031352043151855,
21
  "learning_rate": 7.84313725490196e-06,
22
- "loss": 0.5136,
23
  "step": 20
24
  },
25
  {
26
  "epoch": 0.15,
27
- "grad_norm": 5.687072277069092,
28
  "learning_rate": 1.1764705882352942e-05,
29
- "loss": 0.2446,
30
  "step": 30
31
  },
32
  {
33
  "epoch": 0.2,
34
- "grad_norm": 4.7356672286987305,
35
  "learning_rate": 1.568627450980392e-05,
36
- "loss": 0.1611,
37
  "step": 40
38
  },
39
  {
40
  "epoch": 0.25,
41
- "grad_norm": 14.7470121383667,
42
  "learning_rate": 1.9607843137254903e-05,
43
- "loss": 0.1333,
44
  "step": 50
45
  },
46
  {
47
  "epoch": 0.3,
48
- "grad_norm": 9.882979393005371,
49
  "learning_rate": 1.9995698998770955e-05,
50
- "loss": 0.1203,
51
  "step": 60
52
  },
53
  {
54
  "epoch": 0.34,
55
- "grad_norm": 2.200605869293213,
56
  "learning_rate": 1.998083609002402e-05,
57
- "loss": 0.0866,
58
  "step": 70
59
  },
60
  {
61
  "epoch": 0.39,
62
- "grad_norm": 9.529250144958496,
63
  "learning_rate": 1.995537395500004e-05,
64
- "loss": 0.0825,
65
  "step": 80
66
  },
67
  {
68
  "epoch": 0.44,
69
- "grad_norm": 4.56328010559082,
70
  "learning_rate": 1.9919339633410737e-05,
71
- "loss": 0.0757,
72
  "step": 90
73
  },
74
  {
75
  "epoch": 0.49,
76
- "grad_norm": 8.917645454406738,
77
  "learning_rate": 1.9872771392183334e-05,
78
- "loss": 0.1336,
79
  "step": 100
80
  },
81
  {
82
  "epoch": 0.54,
83
- "grad_norm": 4.956594944000244,
84
  "learning_rate": 1.981571868482269e-05,
85
- "loss": 0.0986,
86
  "step": 110
87
  },
88
  {
89
  "epoch": 0.59,
90
- "grad_norm": 9.08370590209961,
91
  "learning_rate": 1.974824209889377e-05,
92
- "loss": 0.0839,
93
  "step": 120
94
  },
95
  {
96
  "epoch": 0.64,
97
- "grad_norm": 12.914963722229004,
98
  "learning_rate": 1.9670413291680223e-05,
99
- "loss": 0.0909,
100
  "step": 130
101
  },
102
  {
103
  "epoch": 0.69,
104
- "grad_norm": 3.6358766555786133,
105
  "learning_rate": 1.9582314914087344e-05,
106
- "loss": 0.064,
107
  "step": 140
108
  },
109
  {
110
  "epoch": 0.74,
111
- "grad_norm": 7.188825607299805,
112
  "learning_rate": 1.9484040522870333e-05,
113
- "loss": 0.0969,
114
  "step": 150
115
  },
116
  {
117
  "epoch": 0.79,
118
- "grad_norm": 16.197107315063477,
119
  "learning_rate": 1.9375694481280965e-05,
120
- "loss": 0.0871,
121
  "step": 160
122
  },
123
  {
124
  "epoch": 0.84,
125
- "grad_norm": 11.690835952758789,
126
  "learning_rate": 1.9257391848238212e-05,
127
- "loss": 0.0726,
128
  "step": 170
129
  },
130
  {
131
  "epoch": 0.89,
132
- "grad_norm": 3.613584280014038,
133
  "learning_rate": 1.9129258256140556e-05,
134
- "loss": 0.0703,
135
  "step": 180
136
  },
137
  {
138
  "epoch": 0.93,
139
- "grad_norm": 6.19437313079834,
140
  "learning_rate": 1.8991429777449674e-05,
141
- "loss": 0.0939,
142
  "step": 190
143
  },
144
  {
145
  "epoch": 0.98,
146
- "grad_norm": 11.514800071716309,
147
  "learning_rate": 1.884405278018722e-05,
148
- "loss": 0.068,
149
  "step": 200
150
  },
151
  {
152
  "epoch": 1.0,
153
- "eval_accuracy": 0.9707602339181286,
154
- "eval_f1": 0.9722762986965865,
155
- "eval_loss": 0.07348757982254028,
156
- "eval_matthews_correlation": 0.8017394581463343,
157
- "eval_precision": 0.9754056100703739,
158
- "eval_recall": 0.9707602339181286,
159
- "eval_runtime": 5.7659,
160
- "eval_samples_per_second": 118.628,
161
- "eval_steps_per_second": 7.458,
162
  "step": 203
163
  },
164
  {
165
  "epoch": 1.03,
166
- "grad_norm": 10.175304412841797,
167
  "learning_rate": 1.8687283772498205e-05,
168
- "loss": 0.0845,
169
  "step": 210
170
  },
171
  {
172
  "epoch": 1.08,
173
- "grad_norm": 2.231261968612671,
174
  "learning_rate": 1.852128923644593e-05,
175
- "loss": 0.0563,
176
  "step": 220
177
  },
178
  {
179
  "epoch": 1.13,
180
- "grad_norm": 9.75424861907959,
181
  "learning_rate": 1.8346245451215068e-05,
182
- "loss": 0.0791,
183
  "step": 230
184
  },
185
  {
186
  "epoch": 1.18,
187
- "grad_norm": 9.436419486999512,
188
  "learning_rate": 1.8162338305910636e-05,
189
- "loss": 0.0751,
190
  "step": 240
191
  },
192
  {
193
  "epoch": 1.23,
194
- "grad_norm": 5.413998126983643,
195
  "learning_rate": 1.79697631021516e-05,
196
- "loss": 0.0571,
197
  "step": 250
198
  },
199
  {
200
  "epoch": 1.28,
201
- "grad_norm": 10.832895278930664,
202
  "learning_rate": 1.776872434666882e-05,
203
- "loss": 0.0903,
204
  "step": 260
205
  },
206
  {
207
  "epoch": 1.33,
208
- "grad_norm": 14.028545379638672,
209
  "learning_rate": 1.7559435534127534e-05,
210
- "loss": 0.0924,
211
  "step": 270
212
  },
213
  {
214
  "epoch": 1.38,
215
- "grad_norm": 6.962947368621826,
216
  "learning_rate": 1.7342118920405035e-05,
217
- "loss": 0.1018,
218
  "step": 280
219
  },
220
  {
221
  "epoch": 1.43,
222
- "grad_norm": 12.277998924255371,
223
  "learning_rate": 1.7117005286564344e-05,
224
- "loss": 0.0651,
225
  "step": 290
226
  },
227
  {
228
  "epoch": 1.48,
229
- "grad_norm": 3.5764994621276855,
230
  "learning_rate": 1.688433369377444e-05,
231
- "loss": 0.0811,
232
  "step": 300
233
  },
234
  {
235
  "epoch": 1.53,
236
- "grad_norm": 6.763672351837158,
237
  "learning_rate": 1.6644351229437416e-05,
238
- "loss": 0.0785,
239
  "step": 310
240
  },
241
  {
242
  "epoch": 1.57,
243
- "grad_norm": 3.3835597038269043,
244
  "learning_rate": 1.63973127447921e-05,
245
- "loss": 0.0482,
246
  "step": 320
247
  },
248
  {
249
  "epoch": 1.62,
250
- "grad_norm": 6.173498153686523,
251
  "learning_rate": 1.6143480584272794e-05,
252
- "loss": 0.0593,
253
  "step": 330
254
  },
255
  {
256
  "epoch": 1.67,
257
- "grad_norm": 7.676692008972168,
258
  "learning_rate": 1.5883124306910563e-05,
259
- "loss": 0.0678,
260
  "step": 340
261
  },
262
  {
263
  "epoch": 1.72,
264
- "grad_norm": 2.495035171508789,
265
  "learning_rate": 1.5616520400072963e-05,
266
- "loss": 0.0572,
267
  "step": 350
268
  },
269
  {
270
  "epoch": 1.77,
271
- "grad_norm": 16.871797561645508,
272
  "learning_rate": 1.5343951985846096e-05,
273
- "loss": 0.0756,
274
  "step": 360
275
  },
276
  {
277
  "epoch": 1.82,
278
- "grad_norm": 2.3725368976593018,
279
  "learning_rate": 1.5065708520370943e-05,
280
- "loss": 0.0711,
281
  "step": 370
282
  },
283
  {
284
  "epoch": 1.87,
285
- "grad_norm": 12.699337005615234,
286
  "learning_rate": 1.4782085486453155e-05,
287
- "loss": 0.0769,
288
  "step": 380
289
  },
290
  {
291
  "epoch": 1.92,
292
- "grad_norm": 0.583565890789032,
293
  "learning_rate": 1.4493384079772815e-05,
294
- "loss": 0.0612,
295
  "step": 390
296
  },
297
  {
298
  "epoch": 1.97,
299
- "grad_norm": 6.7856550216674805,
300
  "learning_rate": 1.4199910889027335e-05,
301
- "loss": 0.057,
302
  "step": 400
303
  },
304
  {
305
  "epoch": 2.0,
306
- "eval_accuracy": 0.9722222222222222,
307
- "eval_f1": 0.9728689575025272,
308
- "eval_loss": 0.06863868236541748,
309
- "eval_matthews_correlation": 0.7943272582319131,
310
- "eval_precision": 0.9738281672617116,
311
- "eval_recall": 0.9722222222222222,
312
- "eval_runtime": 5.5104,
313
- "eval_samples_per_second": 124.128,
314
- "eval_steps_per_second": 7.803,
315
  "step": 406
316
  },
317
  {
318
  "epoch": 2.02,
319
- "grad_norm": 2.3636648654937744,
320
  "learning_rate": 1.390197757034721e-05,
321
- "loss": 0.0745,
322
  "step": 410
323
  },
324
  {
325
  "epoch": 2.07,
326
- "grad_norm": 21.622703552246094,
327
  "learning_rate": 1.3599900516330382e-05,
328
- "loss": 0.0724,
329
  "step": 420
330
  },
331
  {
332
  "epoch": 2.12,
333
- "grad_norm": 11.821688652038574,
334
  "learning_rate": 1.3294000520046666e-05,
335
- "loss": 0.069,
336
  "step": 430
337
  },
338
  {
339
  "epoch": 2.16,
340
- "grad_norm": 7.000888347625732,
341
  "learning_rate": 1.2984602434369058e-05,
342
- "loss": 0.0714,
343
  "step": 440
344
  },
345
  {
346
  "epoch": 2.21,
347
- "grad_norm": 4.6769256591796875,
348
  "learning_rate": 1.2672034826993716e-05,
349
- "loss": 0.0509,
350
  "step": 450
351
  },
352
  {
353
  "epoch": 2.26,
354
- "grad_norm": 1.7955386638641357,
355
  "learning_rate": 1.235662963151493e-05,
356
- "loss": 0.0512,
357
  "step": 460
358
  },
359
  {
360
  "epoch": 2.31,
361
- "grad_norm": 22.078067779541016,
362
  "learning_rate": 1.2038721794925689e-05,
363
- "loss": 0.064,
364
  "step": 470
365
  },
366
  {
367
  "epoch": 2.36,
368
- "grad_norm": 6.917346000671387,
369
  "learning_rate": 1.1718648921918112e-05,
370
- "loss": 0.0701,
371
  "step": 480
372
  },
373
  {
374
  "epoch": 2.41,
375
- "grad_norm": 8.970315933227539,
376
  "learning_rate": 1.1396750916361526e-05,
377
- "loss": 0.0915,
378
  "step": 490
379
  },
380
  {
381
  "epoch": 2.46,
382
- "grad_norm": 13.326787948608398,
383
  "learning_rate": 1.1073369620338928e-05,
384
- "loss": 0.078,
385
  "step": 500
386
  },
387
  {
388
  "epoch": 2.51,
389
- "grad_norm": 5.894493103027344,
390
  "learning_rate": 1.074884845112512e-05,
391
- "loss": 0.0636,
392
  "step": 510
393
  },
394
  {
395
  "epoch": 2.56,
396
- "grad_norm": 5.459946155548096,
397
  "learning_rate": 1.0423532036492077e-05,
398
- "loss": 0.0711,
399
  "step": 520
400
  },
401
  {
402
  "epoch": 2.61,
403
- "grad_norm": 8.83890151977539,
404
  "learning_rate": 1.0097765848728825e-05,
405
- "loss": 0.079,
406
  "step": 530
407
  },
408
  {
409
  "epoch": 2.66,
410
- "grad_norm": 2.340186357498169,
411
  "learning_rate": 9.771895837764438e-06,
412
- "loss": 0.0449,
413
  "step": 540
414
  },
415
  {
416
  "epoch": 2.71,
417
- "grad_norm": 11.93194580078125,
418
  "learning_rate": 9.446268063783853e-06,
419
- "loss": 0.0794,
420
  "step": 550
421
  },
422
  {
423
  "epoch": 2.76,
424
- "grad_norm": 6.638481616973877,
425
  "learning_rate": 9.121228329726563e-06,
426
- "loss": 0.0614,
427
  "step": 560
428
  },
429
  {
430
  "epoch": 2.8,
431
- "grad_norm": 5.963674068450928,
432
  "learning_rate": 8.797121814058502e-06,
433
- "loss": 0.0449,
434
  "step": 570
435
  },
436
  {
437
  "epoch": 2.85,
438
- "grad_norm": 7.2139506340026855,
439
  "learning_rate": 8.474292704207095e-06,
440
- "loss": 0.0752,
441
  "step": 580
442
  },
443
  {
444
  "epoch": 2.9,
445
- "grad_norm": 2.9632108211517334,
446
  "learning_rate": 8.153083831048772e-06,
447
- "loss": 0.0575,
448
  "step": 590
449
  },
450
  {
451
  "epoch": 2.95,
452
- "grad_norm": 3.8059580326080322,
453
  "learning_rate": 7.833836304837022e-06,
454
- "loss": 0.046,
455
  "step": 600
456
  },
457
  {
458
  "epoch": 3.0,
459
- "eval_accuracy": 0.9780701754385965,
460
- "eval_f1": 0.9787695402282305,
461
- "eval_loss": 0.05532665178179741,
462
- "eval_matthews_correlation": 0.8420667383515437,
463
- "eval_precision": 0.9800898852068441,
464
- "eval_recall": 0.9780701754385965,
465
- "eval_runtime": 5.6675,
466
- "eval_samples_per_second": 120.688,
467
- "eval_steps_per_second": 7.587,
468
  "step": 609
469
  },
470
  {
471
  "epoch": 3.0,
472
- "grad_norm": 6.808079719543457,
473
  "learning_rate": 7.516889152957744e-06,
474
- "loss": 0.074,
475
  "step": 610
476
  },
477
  {
478
  "epoch": 3.05,
479
- "grad_norm": 2.9367547035217285,
480
  "learning_rate": 7.202578959896491e-06,
481
- "loss": 0.0646,
482
  "step": 620
483
  },
484
  {
485
  "epoch": 3.1,
486
- "grad_norm": 15.483344078063965,
487
  "learning_rate": 6.891239509799932e-06,
488
- "loss": 0.0835,
489
  "step": 630
490
  },
491
  {
492
  "epoch": 3.15,
493
- "grad_norm": 8.343765258789062,
494
  "learning_rate": 6.583201432011217e-06,
495
- "loss": 0.0609,
496
  "step": 640
497
  },
498
  {
499
  "epoch": 3.2,
500
- "grad_norm": 2.9499011039733887,
501
  "learning_rate": 6.278791849955583e-06,
502
- "loss": 0.0669,
503
  "step": 650
504
  },
505
  {
506
  "epoch": 3.25,
507
- "grad_norm": 4.759853363037109,
508
  "learning_rate": 5.978334033749076e-06,
509
- "loss": 0.0434,
510
  "step": 660
511
  },
512
  {
513
  "epoch": 3.3,
514
- "grad_norm": 3.5021698474884033,
515
  "learning_rate": 5.682147056899361e-06,
516
- "loss": 0.0389,
517
  "step": 670
518
  },
519
  {
520
  "epoch": 3.35,
521
- "grad_norm": 8.20483112335205,
522
  "learning_rate": 5.390545457463134e-06,
523
- "loss": 0.0512,
524
  "step": 680
525
  },
526
  {
527
  "epoch": 3.39,
528
- "grad_norm": 15.760091781616211,
529
  "learning_rate": 5.103838904019993e-06,
530
- "loss": 0.082,
531
  "step": 690
532
  },
533
  {
534
  "epoch": 3.44,
535
- "grad_norm": 5.340324878692627,
536
  "learning_rate": 4.822331866817478e-06,
537
- "loss": 0.0412,
538
  "step": 700
539
  },
540
  {
541
  "epoch": 3.49,
542
- "grad_norm": 2.386610507965088,
543
  "learning_rate": 4.546323294436556e-06,
544
- "loss": 0.0764,
545
  "step": 710
546
  },
547
  {
548
  "epoch": 3.54,
549
- "grad_norm": 2.604590892791748,
550
  "learning_rate": 4.276106296320828e-06,
551
- "loss": 0.0322,
552
  "step": 720
553
  },
554
  {
555
  "epoch": 3.59,
556
- "grad_norm": 2.6730122566223145,
557
  "learning_rate": 4.0119678315067025e-06,
558
- "loss": 0.0532,
559
  "step": 730
560
  },
561
  {
562
  "epoch": 3.64,
563
- "grad_norm": 11.014791488647461,
564
  "learning_rate": 3.754188403885013e-06,
565
- "loss": 0.0788,
566
  "step": 740
567
  },
568
  {
569
  "epoch": 3.69,
570
- "grad_norm": 2.279393196105957,
571
  "learning_rate": 3.5030417643177416e-06,
572
- "loss": 0.0465,
573
  "step": 750
574
  },
575
  {
576
  "epoch": 3.74,
577
- "grad_norm": 7.80838680267334,
578
  "learning_rate": 3.258794619926159e-06,
579
- "loss": 0.0441,
580
  "step": 760
581
  },
582
  {
583
  "epoch": 3.79,
584
- "grad_norm": 5.728611946105957,
585
  "learning_rate": 3.021706350859147e-06,
586
- "loss": 0.0476,
587
  "step": 770
588
  },
589
  {
590
  "epoch": 3.84,
591
- "grad_norm": 2.4311773777008057,
592
  "learning_rate": 2.792028734842418e-06,
593
- "loss": 0.0421,
594
  "step": 780
595
  },
596
  {
597
  "epoch": 3.89,
598
- "grad_norm": 19.25046730041504,
599
  "learning_rate": 2.5700056798012164e-06,
600
- "loss": 0.0829,
601
  "step": 790
602
  },
603
  {
604
  "epoch": 3.94,
605
- "grad_norm": 16.992164611816406,
606
  "learning_rate": 2.3558729648404065e-06,
607
- "loss": 0.0694,
608
  "step": 800
609
  },
610
  {
611
  "epoch": 3.99,
612
- "grad_norm": 1.1272871494293213,
613
  "learning_rate": 2.1498579898570228e-06,
614
- "loss": 0.0571,
615
  "step": 810
616
  },
617
  {
618
  "epoch": 4.0,
619
- "eval_accuracy": 0.9780701754385965,
620
- "eval_f1": 0.9777344542006784,
621
- "eval_loss": 0.05015131086111069,
622
- "eval_matthews_correlation": 0.8239461397313993,
623
- "eval_precision": 0.9775337669457737,
624
- "eval_recall": 0.9780701754385965,
625
- "eval_runtime": 5.516,
626
- "eval_samples_per_second": 124.003,
627
- "eval_steps_per_second": 7.796,
628
  "step": 813
629
  },
630
  {
631
  "epoch": 4.03,
632
- "grad_norm": 20.361764907836914,
633
  "learning_rate": 1.952179534051183e-06,
634
- "loss": 0.0545,
635
  "step": 820
636
  },
637
  {
638
  "epoch": 4.08,
639
- "grad_norm": 12.892123222351074,
640
  "learning_rate": 1.763047523591831e-06,
641
- "loss": 0.0507,
642
  "step": 830
643
  },
644
  {
645
  "epoch": 4.13,
646
- "grad_norm": 4.696685791015625,
647
  "learning_rate": 1.5826628086839968e-06,
648
- "loss": 0.0635,
649
  "step": 840
650
  },
651
  {
652
  "epoch": 4.18,
653
- "grad_norm": 10.167694091796875,
654
  "learning_rate": 1.41121695027438e-06,
655
- "loss": 0.0547,
656
  "step": 850
657
  },
658
  {
659
  "epoch": 4.23,
660
- "grad_norm": 4.277148723602295,
661
  "learning_rate": 1.2488920166217034e-06,
662
- "loss": 0.059,
663
  "step": 860
664
  },
665
  {
666
  "epoch": 4.28,
667
- "grad_norm": 15.234513282775879,
668
  "learning_rate": 1.095860389947928e-06,
669
- "loss": 0.068,
670
  "step": 870
671
  },
672
  {
673
  "epoch": 4.33,
674
- "grad_norm": 5.693934440612793,
675
  "learning_rate": 9.522845833756001e-07,
676
- "loss": 0.0451,
677
  "step": 880
678
  },
679
  {
680
  "epoch": 4.38,
681
- "grad_norm": 10.728283882141113,
682
  "learning_rate": 8.183170683457986e-07,
683
- "loss": 0.0468,
684
  "step": 890
685
  },
686
  {
687
  "epoch": 4.43,
688
- "grad_norm": 8.146360397338867,
689
  "learning_rate": 6.941001126998892e-07,
690
- "loss": 0.0378,
691
  "step": 900
692
  },
693
  {
694
  "epoch": 4.48,
695
- "grad_norm": 3.4322080612182617,
696
  "learning_rate": 5.797656295970955e-07,
697
- "loss": 0.0418,
698
  "step": 910
699
  },
700
  {
701
  "epoch": 4.53,
702
- "grad_norm": 3.0174903869628906,
703
  "learning_rate": 4.754350374283001e-07,
704
- "loss": 0.0441,
705
  "step": 920
706
  },
707
  {
708
  "epoch": 4.58,
709
- "grad_norm": 16.416744232177734,
710
  "learning_rate": 3.8121913087483033e-07,
711
- "loss": 0.0495,
712
  "step": 930
713
  },
714
  {
715
  "epoch": 4.62,
716
- "grad_norm": 12.529999732971191,
717
  "learning_rate": 2.972179632491989e-07,
718
- "loss": 0.0747,
719
  "step": 940
720
  },
721
  {
722
  "epoch": 4.67,
723
- "grad_norm": 1.2379523515701294,
724
  "learning_rate": 2.23520740242712e-07,
725
- "loss": 0.0583,
726
  "step": 950
727
  },
728
  {
729
  "epoch": 4.72,
730
- "grad_norm": 5.813112735748291,
731
  "learning_rate": 1.602057251927891e-07,
732
- "loss": 0.0612,
733
  "step": 960
734
  },
735
  {
736
  "epoch": 4.77,
737
- "grad_norm": 9.207404136657715,
738
  "learning_rate": 1.0734015597060222e-07,
739
- "loss": 0.0462,
740
  "step": 970
741
  },
742
  {
743
  "epoch": 4.82,
744
- "grad_norm": 10.823348999023438,
745
  "learning_rate": 6.498017357731035e-08,
746
- "loss": 0.0653,
747
  "step": 980
748
  },
749
  {
750
  "epoch": 4.87,
751
- "grad_norm": 1.9045536518096924,
752
  "learning_rate": 3.317076252467133e-08,
753
- "loss": 0.0487,
754
  "step": 990
755
  },
756
  {
757
  "epoch": 4.92,
758
- "grad_norm": 10.43856430053711,
759
  "learning_rate": 1.1945703063402925e-08,
760
- "loss": 0.0449,
761
  "step": 1000
762
  },
763
  {
764
  "epoch": 4.97,
765
- "grad_norm": 4.215317249298096,
766
  "learning_rate": 1.327535309979533e-09,
767
- "loss": 0.0491,
768
  "step": 1010
769
  },
770
  {
771
  "epoch": 4.99,
772
- "eval_accuracy": 0.9766081871345029,
773
- "eval_f1": 0.9755931812139886,
774
- "eval_loss": 0.04963676258921623,
775
- "eval_matthews_correlation": 0.8055620626992487,
776
- "eval_precision": 0.9756262052467682,
777
- "eval_recall": 0.9766081871345029,
778
- "eval_runtime": 5.7281,
779
- "eval_samples_per_second": 119.41,
780
- "eval_steps_per_second": 7.507,
781
  "step": 1015
782
  },
783
  {
784
  "epoch": 4.99,
785
  "step": 1015,
786
  "total_flos": 2.581447650539471e+18,
787
- "train_loss": 0.08238832812591139,
788
- "train_runtime": 705.5168,
789
- "train_samples_per_second": 92.096,
790
- "train_steps_per_second": 1.439
791
  }
792
  ],
793
  "logging_steps": 10,
 
1
  {
2
+ "best_metric": 0.9883040935672515,
3
+ "best_model_checkpoint": "./outputs/convnextv2-nano-22k-384-boulderspot-vN/checkpoint-1015",
4
  "epoch": 4.993849938499385,
5
  "eval_steps": 500,
6
  "global_step": 1015,
 
10
  "log_history": [
11
  {
12
  "epoch": 0.05,
13
+ "grad_norm": 26.665199279785156,
14
  "learning_rate": 3.92156862745098e-06,
15
+ "loss": 0.6334,
16
  "step": 10
17
  },
18
  {
19
  "epoch": 0.1,
20
+ "grad_norm": 7.943151473999023,
21
  "learning_rate": 7.84313725490196e-06,
22
+ "loss": 0.365,
23
  "step": 20
24
  },
25
  {
26
  "epoch": 0.15,
27
+ "grad_norm": 16.875411987304688,
28
  "learning_rate": 1.1764705882352942e-05,
29
+ "loss": 0.2382,
30
  "step": 30
31
  },
32
  {
33
  "epoch": 0.2,
34
+ "grad_norm": 5.587479114532471,
35
  "learning_rate": 1.568627450980392e-05,
36
+ "loss": 0.1425,
37
  "step": 40
38
  },
39
  {
40
  "epoch": 0.25,
41
+ "grad_norm": 7.946935176849365,
42
  "learning_rate": 1.9607843137254903e-05,
43
+ "loss": 0.1257,
44
  "step": 50
45
  },
46
  {
47
  "epoch": 0.3,
48
+ "grad_norm": 4.481125831604004,
49
  "learning_rate": 1.9995698998770955e-05,
50
+ "loss": 0.0967,
51
  "step": 60
52
  },
53
  {
54
  "epoch": 0.34,
55
+ "grad_norm": 9.453625679016113,
56
  "learning_rate": 1.998083609002402e-05,
57
+ "loss": 0.0829,
58
  "step": 70
59
  },
60
  {
61
  "epoch": 0.39,
62
+ "grad_norm": 13.032307624816895,
63
  "learning_rate": 1.995537395500004e-05,
64
+ "loss": 0.1073,
65
  "step": 80
66
  },
67
  {
68
  "epoch": 0.44,
69
+ "grad_norm": 5.036722183227539,
70
  "learning_rate": 1.9919339633410737e-05,
71
+ "loss": 0.0788,
72
  "step": 90
73
  },
74
  {
75
  "epoch": 0.49,
76
+ "grad_norm": 10.13275146484375,
77
  "learning_rate": 1.9872771392183334e-05,
78
+ "loss": 0.1421,
79
  "step": 100
80
  },
81
  {
82
  "epoch": 0.54,
83
+ "grad_norm": 10.442366600036621,
84
  "learning_rate": 1.981571868482269e-05,
85
+ "loss": 0.1035,
86
  "step": 110
87
  },
88
  {
89
  "epoch": 0.59,
90
+ "grad_norm": 7.753129959106445,
91
  "learning_rate": 1.974824209889377e-05,
92
+ "loss": 0.1001,
93
  "step": 120
94
  },
95
  {
96
  "epoch": 0.64,
97
+ "grad_norm": 10.481592178344727,
98
  "learning_rate": 1.9670413291680223e-05,
99
+ "loss": 0.1007,
100
  "step": 130
101
  },
102
  {
103
  "epoch": 0.69,
104
+ "grad_norm": 7.881091594696045,
105
  "learning_rate": 1.9582314914087344e-05,
106
+ "loss": 0.0669,
107
  "step": 140
108
  },
109
  {
110
  "epoch": 0.74,
111
+ "grad_norm": 12.763803482055664,
112
  "learning_rate": 1.9484040522870333e-05,
113
+ "loss": 0.0723,
114
  "step": 150
115
  },
116
  {
117
  "epoch": 0.79,
118
+ "grad_norm": 6.890717029571533,
119
  "learning_rate": 1.9375694481280965e-05,
120
+ "loss": 0.0804,
121
  "step": 160
122
  },
123
  {
124
  "epoch": 0.84,
125
+ "grad_norm": 6.086670398712158,
126
  "learning_rate": 1.9257391848238212e-05,
127
+ "loss": 0.1073,
128
  "step": 170
129
  },
130
  {
131
  "epoch": 0.89,
132
+ "grad_norm": 4.475172996520996,
133
  "learning_rate": 1.9129258256140556e-05,
134
+ "loss": 0.0779,
135
  "step": 180
136
  },
137
  {
138
  "epoch": 0.93,
139
+ "grad_norm": 4.19266414642334,
140
  "learning_rate": 1.8991429777449674e-05,
141
+ "loss": 0.0934,
142
  "step": 190
143
  },
144
  {
145
  "epoch": 0.98,
146
+ "grad_norm": 14.975569725036621,
147
  "learning_rate": 1.884405278018722e-05,
148
+ "loss": 0.1102,
149
  "step": 200
150
  },
151
  {
152
  "epoch": 1.0,
153
+ "eval_accuracy": 0.9839181286549707,
154
+ "eval_f1": 0.9840087498605248,
155
+ "eval_loss": 0.0431428886950016,
156
+ "eval_matthews_correlation": 0.8589815510372486,
157
+ "eval_precision": 0.9841155979252388,
158
+ "eval_recall": 0.9839181286549707,
159
+ "eval_runtime": 4.9799,
160
+ "eval_samples_per_second": 137.351,
161
+ "eval_steps_per_second": 8.635,
162
  "step": 203
163
  },
164
  {
165
  "epoch": 1.03,
166
+ "grad_norm": 14.823522567749023,
167
  "learning_rate": 1.8687283772498205e-05,
168
+ "loss": 0.0567,
169
  "step": 210
170
  },
171
  {
172
  "epoch": 1.08,
173
+ "grad_norm": 2.4599366188049316,
174
  "learning_rate": 1.852128923644593e-05,
175
+ "loss": 0.08,
176
  "step": 220
177
  },
178
  {
179
  "epoch": 1.13,
180
+ "grad_norm": 1.6709442138671875,
181
  "learning_rate": 1.8346245451215068e-05,
182
+ "loss": 0.056,
183
  "step": 230
184
  },
185
  {
186
  "epoch": 1.18,
187
+ "grad_norm": 6.953076362609863,
188
  "learning_rate": 1.8162338305910636e-05,
189
+ "loss": 0.0683,
190
  "step": 240
191
  },
192
  {
193
  "epoch": 1.23,
194
+ "grad_norm": 3.510495901107788,
195
  "learning_rate": 1.79697631021516e-05,
196
+ "loss": 0.0825,
197
  "step": 250
198
  },
199
  {
200
  "epoch": 1.28,
201
+ "grad_norm": 8.399980545043945,
202
  "learning_rate": 1.776872434666882e-05,
203
+ "loss": 0.0584,
204
  "step": 260
205
  },
206
  {
207
  "epoch": 1.33,
208
+ "grad_norm": 14.62066650390625,
209
  "learning_rate": 1.7559435534127534e-05,
210
+ "loss": 0.0998,
211
  "step": 270
212
  },
213
  {
214
  "epoch": 1.38,
215
+ "grad_norm": 6.287339210510254,
216
  "learning_rate": 1.7342118920405035e-05,
217
+ "loss": 0.0813,
218
  "step": 280
219
  },
220
  {
221
  "epoch": 1.43,
222
+ "grad_norm": 7.23698091506958,
223
  "learning_rate": 1.7117005286564344e-05,
224
+ "loss": 0.0973,
225
  "step": 290
226
  },
227
  {
228
  "epoch": 1.48,
229
+ "grad_norm": 12.766968727111816,
230
  "learning_rate": 1.688433369377444e-05,
231
+ "loss": 0.0928,
232
  "step": 300
233
  },
234
  {
235
  "epoch": 1.53,
236
+ "grad_norm": 16.611492156982422,
237
  "learning_rate": 1.6644351229437416e-05,
238
+ "loss": 0.08,
239
  "step": 310
240
  },
241
  {
242
  "epoch": 1.57,
243
+ "grad_norm": 1.000801920890808,
244
  "learning_rate": 1.63973127447921e-05,
245
+ "loss": 0.0582,
246
  "step": 320
247
  },
248
  {
249
  "epoch": 1.62,
250
+ "grad_norm": 10.745514869689941,
251
  "learning_rate": 1.6143480584272794e-05,
252
+ "loss": 0.0915,
253
  "step": 330
254
  },
255
  {
256
  "epoch": 1.67,
257
+ "grad_norm": 10.188589096069336,
258
  "learning_rate": 1.5883124306910563e-05,
259
+ "loss": 0.0869,
260
  "step": 340
261
  },
262
  {
263
  "epoch": 1.72,
264
+ "grad_norm": 7.215929985046387,
265
  "learning_rate": 1.5616520400072963e-05,
266
+ "loss": 0.1329,
267
  "step": 350
268
  },
269
  {
270
  "epoch": 1.77,
271
+ "grad_norm": 4.967738151550293,
272
  "learning_rate": 1.5343951985846096e-05,
273
+ "loss": 0.0781,
274
  "step": 360
275
  },
276
  {
277
  "epoch": 1.82,
278
+ "grad_norm": 2.9467406272888184,
279
  "learning_rate": 1.5065708520370943e-05,
280
+ "loss": 0.0582,
281
  "step": 370
282
  },
283
  {
284
  "epoch": 1.87,
285
+ "grad_norm": 6.54543399810791,
286
  "learning_rate": 1.4782085486453155e-05,
287
+ "loss": 0.0626,
288
  "step": 380
289
  },
290
  {
291
  "epoch": 1.92,
292
+ "grad_norm": 7.8247833251953125,
293
  "learning_rate": 1.4493384079772815e-05,
294
+ "loss": 0.081,
295
  "step": 390
296
  },
297
  {
298
  "epoch": 1.97,
299
+ "grad_norm": 8.889460563659668,
300
  "learning_rate": 1.4199910889027335e-05,
301
+ "loss": 0.0559,
302
  "step": 400
303
  },
304
  {
305
  "epoch": 2.0,
306
+ "eval_accuracy": 0.9839181286549707,
307
+ "eval_f1": 0.9845065535493497,
308
+ "eval_loss": 0.04758350923657417,
309
+ "eval_matthews_correlation": 0.8709455436036694,
310
+ "eval_precision": 0.9858048061716136,
311
+ "eval_recall": 0.9839181286549707,
312
+ "eval_runtime": 5.1397,
313
+ "eval_samples_per_second": 133.081,
314
+ "eval_steps_per_second": 8.366,
315
  "step": 406
316
  },
317
  {
318
  "epoch": 2.02,
319
+ "grad_norm": 11.741949081420898,
320
  "learning_rate": 1.390197757034721e-05,
321
+ "loss": 0.0806,
322
  "step": 410
323
  },
324
  {
325
  "epoch": 2.07,
326
+ "grad_norm": 8.677542686462402,
327
  "learning_rate": 1.3599900516330382e-05,
328
+ "loss": 0.1033,
329
  "step": 420
330
  },
331
  {
332
  "epoch": 2.12,
333
+ "grad_norm": 1.3803766965866089,
334
  "learning_rate": 1.3294000520046666e-05,
335
+ "loss": 0.0433,
336
  "step": 430
337
  },
338
  {
339
  "epoch": 2.16,
340
+ "grad_norm": 8.467944145202637,
341
  "learning_rate": 1.2984602434369058e-05,
342
+ "loss": 0.0691,
343
  "step": 440
344
  },
345
  {
346
  "epoch": 2.21,
347
+ "grad_norm": 6.276478290557861,
348
  "learning_rate": 1.2672034826993716e-05,
349
+ "loss": 0.0678,
350
  "step": 450
351
  },
352
  {
353
  "epoch": 2.26,
354
+ "grad_norm": 11.57728385925293,
355
  "learning_rate": 1.235662963151493e-05,
356
+ "loss": 0.0804,
357
  "step": 460
358
  },
359
  {
360
  "epoch": 2.31,
361
+ "grad_norm": 0.9122611880302429,
362
  "learning_rate": 1.2038721794925689e-05,
363
+ "loss": 0.0476,
364
  "step": 470
365
  },
366
  {
367
  "epoch": 2.36,
368
+ "grad_norm": 1.983340859413147,
369
  "learning_rate": 1.1718648921918112e-05,
370
+ "loss": 0.0545,
371
  "step": 480
372
  },
373
  {
374
  "epoch": 2.41,
375
+ "grad_norm": 17.866886138916016,
376
  "learning_rate": 1.1396750916361526e-05,
377
+ "loss": 0.0512,
378
  "step": 490
379
  },
380
  {
381
  "epoch": 2.46,
382
+ "grad_norm": 9.374113082885742,
383
  "learning_rate": 1.1073369620338928e-05,
384
+ "loss": 0.0604,
385
  "step": 500
386
  },
387
  {
388
  "epoch": 2.51,
389
+ "grad_norm": 4.585148811340332,
390
  "learning_rate": 1.074884845112512e-05,
391
+ "loss": 0.0629,
392
  "step": 510
393
  },
394
  {
395
  "epoch": 2.56,
396
+ "grad_norm": 9.917503356933594,
397
  "learning_rate": 1.0423532036492077e-05,
398
+ "loss": 0.054,
399
  "step": 520
400
  },
401
  {
402
  "epoch": 2.61,
403
+ "grad_norm": 1.8329569101333618,
404
  "learning_rate": 1.0097765848728825e-05,
405
+ "loss": 0.071,
406
  "step": 530
407
  },
408
  {
409
  "epoch": 2.66,
410
+ "grad_norm": 10.834183692932129,
411
  "learning_rate": 9.771895837764438e-06,
412
+ "loss": 0.0642,
413
  "step": 540
414
  },
415
  {
416
  "epoch": 2.71,
417
+ "grad_norm": 5.503483295440674,
418
  "learning_rate": 9.446268063783853e-06,
419
+ "loss": 0.0654,
420
  "step": 550
421
  },
422
  {
423
  "epoch": 2.76,
424
+ "grad_norm": 6.054934024810791,
425
  "learning_rate": 9.121228329726563e-06,
426
+ "loss": 0.061,
427
  "step": 560
428
  },
429
  {
430
  "epoch": 2.8,
431
+ "grad_norm": 8.71143627166748,
432
  "learning_rate": 8.797121814058502e-06,
433
+ "loss": 0.0759,
434
  "step": 570
435
  },
436
  {
437
  "epoch": 2.85,
438
+ "grad_norm": 4.225427150726318,
439
  "learning_rate": 8.474292704207095e-06,
440
+ "loss": 0.0662,
441
  "step": 580
442
  },
443
  {
444
  "epoch": 2.9,
445
+ "grad_norm": 6.926836967468262,
446
  "learning_rate": 8.153083831048772e-06,
447
+ "loss": 0.07,
448
  "step": 590
449
  },
450
  {
451
  "epoch": 2.95,
452
+ "grad_norm": 1.132318377494812,
453
  "learning_rate": 7.833836304837022e-06,
454
+ "loss": 0.0402,
455
  "step": 600
456
  },
457
  {
458
  "epoch": 3.0,
459
+ "eval_accuracy": 0.9809941520467836,
460
+ "eval_f1": 0.9816895632855951,
461
+ "eval_loss": 0.04637432098388672,
462
+ "eval_matthews_correlation": 0.8468348607550005,
463
+ "eval_precision": 0.9830779451248668,
464
+ "eval_recall": 0.9809941520467836,
465
+ "eval_runtime": 5.1222,
466
+ "eval_samples_per_second": 133.536,
467
+ "eval_steps_per_second": 8.395,
468
  "step": 609
469
  },
470
  {
471
  "epoch": 3.0,
472
+ "grad_norm": 2.548271656036377,
473
  "learning_rate": 7.516889152957744e-06,
474
+ "loss": 0.0706,
475
  "step": 610
476
  },
477
  {
478
  "epoch": 3.05,
479
+ "grad_norm": 17.59124755859375,
480
  "learning_rate": 7.202578959896491e-06,
481
+ "loss": 0.0777,
482
  "step": 620
483
  },
484
  {
485
  "epoch": 3.1,
486
+ "grad_norm": 10.620137214660645,
487
  "learning_rate": 6.891239509799932e-06,
488
+ "loss": 0.0443,
489
  "step": 630
490
  },
491
  {
492
  "epoch": 3.15,
493
+ "grad_norm": 20.15471839904785,
494
  "learning_rate": 6.583201432011217e-06,
495
+ "loss": 0.0746,
496
  "step": 640
497
  },
498
  {
499
  "epoch": 3.2,
500
+ "grad_norm": 7.445899486541748,
501
  "learning_rate": 6.278791849955583e-06,
502
+ "loss": 0.0468,
503
  "step": 650
504
  },
505
  {
506
  "epoch": 3.25,
507
+ "grad_norm": 10.309353828430176,
508
  "learning_rate": 5.978334033749076e-06,
509
+ "loss": 0.0698,
510
  "step": 660
511
  },
512
  {
513
  "epoch": 3.3,
514
+ "grad_norm": 4.898565769195557,
515
  "learning_rate": 5.682147056899361e-06,
516
+ "loss": 0.058,
517
  "step": 670
518
  },
519
  {
520
  "epoch": 3.35,
521
+ "grad_norm": 8.0520658493042,
522
  "learning_rate": 5.390545457463134e-06,
523
+ "loss": 0.0468,
524
  "step": 680
525
  },
526
  {
527
  "epoch": 3.39,
528
+ "grad_norm": 3.444305419921875,
529
  "learning_rate": 5.103838904019993e-06,
530
+ "loss": 0.0381,
531
  "step": 690
532
  },
533
  {
534
  "epoch": 3.44,
535
+ "grad_norm": 11.149181365966797,
536
  "learning_rate": 4.822331866817478e-06,
537
+ "loss": 0.0693,
538
  "step": 700
539
  },
540
  {
541
  "epoch": 3.49,
542
+ "grad_norm": 12.743651390075684,
543
  "learning_rate": 4.546323294436556e-06,
544
+ "loss": 0.0739,
545
  "step": 710
546
  },
547
  {
548
  "epoch": 3.54,
549
+ "grad_norm": 9.802193641662598,
550
  "learning_rate": 4.276106296320828e-06,
551
+ "loss": 0.0655,
552
  "step": 720
553
  },
554
  {
555
  "epoch": 3.59,
556
+ "grad_norm": 6.496978759765625,
557
  "learning_rate": 4.0119678315067025e-06,
558
+ "loss": 0.066,
559
  "step": 730
560
  },
561
  {
562
  "epoch": 3.64,
563
+ "grad_norm": 12.13724422454834,
564
  "learning_rate": 3.754188403885013e-06,
565
+ "loss": 0.0483,
566
  "step": 740
567
  },
568
  {
569
  "epoch": 3.69,
570
+ "grad_norm": 11.302549362182617,
571
  "learning_rate": 3.5030417643177416e-06,
572
+ "loss": 0.0727,
573
  "step": 750
574
  },
575
  {
576
  "epoch": 3.74,
577
+ "grad_norm": 3.291935443878174,
578
  "learning_rate": 3.258794619926159e-06,
579
+ "loss": 0.0685,
580
  "step": 760
581
  },
582
  {
583
  "epoch": 3.79,
584
+ "grad_norm": 7.776196479797363,
585
  "learning_rate": 3.021706350859147e-06,
586
+ "loss": 0.0482,
587
  "step": 770
588
  },
589
  {
590
  "epoch": 3.84,
591
+ "grad_norm": 7.8189568519592285,
592
  "learning_rate": 2.792028734842418e-06,
593
+ "loss": 0.0504,
594
  "step": 780
595
  },
596
  {
597
  "epoch": 3.89,
598
+ "grad_norm": 1.9077569246292114,
599
  "learning_rate": 2.5700056798012164e-06,
600
+ "loss": 0.0578,
601
  "step": 790
602
  },
603
  {
604
  "epoch": 3.94,
605
+ "grad_norm": 5.7739667892456055,
606
  "learning_rate": 2.3558729648404065e-06,
607
+ "loss": 0.0712,
608
  "step": 800
609
  },
610
  {
611
  "epoch": 3.99,
612
+ "grad_norm": 6.1386027336120605,
613
  "learning_rate": 2.1498579898570228e-06,
614
+ "loss": 0.0334,
615
  "step": 810
616
  },
617
  {
618
  "epoch": 4.0,
619
+ "eval_accuracy": 0.9868421052631579,
620
+ "eval_f1": 0.986916249885884,
621
+ "eval_loss": 0.03484980762004852,
622
+ "eval_matthews_correlation": 0.884636226651146,
623
+ "eval_precision": 0.9870070425284395,
624
+ "eval_recall": 0.9868421052631579,
625
+ "eval_runtime": 5.1774,
626
+ "eval_samples_per_second": 132.112,
627
+ "eval_steps_per_second": 8.305,
628
  "step": 813
629
  },
630
  {
631
  "epoch": 4.03,
632
+ "grad_norm": 2.534424304962158,
633
  "learning_rate": 1.952179534051183e-06,
634
+ "loss": 0.0613,
635
  "step": 820
636
  },
637
  {
638
  "epoch": 4.08,
639
+ "grad_norm": 3.645782947540283,
640
  "learning_rate": 1.763047523591831e-06,
641
+ "loss": 0.0817,
642
  "step": 830
643
  },
644
  {
645
  "epoch": 4.13,
646
+ "grad_norm": 10.930766105651855,
647
  "learning_rate": 1.5826628086839968e-06,
648
+ "loss": 0.081,
649
  "step": 840
650
  },
651
  {
652
  "epoch": 4.18,
653
+ "grad_norm": 6.021533489227295,
654
  "learning_rate": 1.41121695027438e-06,
655
+ "loss": 0.0522,
656
  "step": 850
657
  },
658
  {
659
  "epoch": 4.23,
660
+ "grad_norm": 6.584732532501221,
661
  "learning_rate": 1.2488920166217034e-06,
662
+ "loss": 0.054,
663
  "step": 860
664
  },
665
  {
666
  "epoch": 4.28,
667
+ "grad_norm": 5.629753112792969,
668
  "learning_rate": 1.095860389947928e-06,
669
+ "loss": 0.0475,
670
  "step": 870
671
  },
672
  {
673
  "epoch": 4.33,
674
+ "grad_norm": 4.272614479064941,
675
  "learning_rate": 9.522845833756001e-07,
676
+ "loss": 0.0458,
677
  "step": 880
678
  },
679
  {
680
  "epoch": 4.38,
681
+ "grad_norm": 9.34067153930664,
682
  "learning_rate": 8.183170683457986e-07,
683
+ "loss": 0.0746,
684
  "step": 890
685
  },
686
  {
687
  "epoch": 4.43,
688
+ "grad_norm": 2.2390189170837402,
689
  "learning_rate": 6.941001126998892e-07,
690
+ "loss": 0.0509,
691
  "step": 900
692
  },
693
  {
694
  "epoch": 4.48,
695
+ "grad_norm": 1.508949637413025,
696
  "learning_rate": 5.797656295970955e-07,
697
+ "loss": 0.0329,
698
  "step": 910
699
  },
700
  {
701
  "epoch": 4.53,
702
+ "grad_norm": 7.0009355545043945,
703
  "learning_rate": 4.754350374283001e-07,
704
+ "loss": 0.0709,
705
  "step": 920
706
  },
707
  {
708
  "epoch": 4.58,
709
+ "grad_norm": 6.07660436630249,
710
  "learning_rate": 3.8121913087483033e-07,
711
+ "loss": 0.0393,
712
  "step": 930
713
  },
714
  {
715
  "epoch": 4.62,
716
+ "grad_norm": 6.966567039489746,
717
  "learning_rate": 2.972179632491989e-07,
718
+ "loss": 0.0464,
719
  "step": 940
720
  },
721
  {
722
  "epoch": 4.67,
723
+ "grad_norm": 3.8328962326049805,
724
  "learning_rate": 2.23520740242712e-07,
725
+ "loss": 0.0457,
726
  "step": 950
727
  },
728
  {
729
  "epoch": 4.72,
730
+ "grad_norm": 3.535372018814087,
731
  "learning_rate": 1.602057251927891e-07,
732
+ "loss": 0.0599,
733
  "step": 960
734
  },
735
  {
736
  "epoch": 4.77,
737
+ "grad_norm": 11.542516708374023,
738
  "learning_rate": 1.0734015597060222e-07,
739
+ "loss": 0.0557,
740
  "step": 970
741
  },
742
  {
743
  "epoch": 4.82,
744
+ "grad_norm": 7.840158939361572,
745
  "learning_rate": 6.498017357731035e-08,
746
+ "loss": 0.0564,
747
  "step": 980
748
  },
749
  {
750
  "epoch": 4.87,
751
+ "grad_norm": 3.7357852458953857,
752
  "learning_rate": 3.317076252467133e-08,
753
+ "loss": 0.033,
754
  "step": 990
755
  },
756
  {
757
  "epoch": 4.92,
758
+ "grad_norm": 4.429032802581787,
759
  "learning_rate": 1.1945703063402925e-08,
760
+ "loss": 0.0496,
761
  "step": 1000
762
  },
763
  {
764
  "epoch": 4.97,
765
+ "grad_norm": 4.065526485443115,
766
  "learning_rate": 1.327535309979533e-09,
767
+ "loss": 0.0445,
768
  "step": 1010
769
  },
770
  {
771
  "epoch": 4.99,
772
+ "eval_accuracy": 0.9883040935672515,
773
+ "eval_f1": 0.9883040935672515,
774
+ "eval_loss": 0.03396734222769737,
775
+ "eval_matthews_correlation": 0.8962181845768691,
776
+ "eval_precision": 0.9883040935672515,
777
+ "eval_recall": 0.9883040935672515,
778
+ "eval_runtime": 5.2898,
779
+ "eval_samples_per_second": 129.305,
780
+ "eval_steps_per_second": 8.129,
781
  "step": 1015
782
  },
783
  {
784
  "epoch": 4.99,
785
  "step": 1015,
786
  "total_flos": 2.581447650539471e+18,
787
+ "train_loss": 0.08031210996247277,
788
+ "train_runtime": 701.9998,
789
+ "train_samples_per_second": 92.557,
790
+ "train_steps_per_second": 1.446
791
  }
792
  ],
793
  "logging_steps": 10,