Hartunka commited on
Commit
bd0a26d
·
verified ·
1 Parent(s): 55076e4

End of training

Browse files
Files changed (5) hide show
  1. README.md +16 -4
  2. all_results.json +16 -0
  3. eval_results.json +10 -0
  4. train_results.json +9 -0
  5. trainer_state.json +908 -0
README.md CHANGED
@@ -1,11 +1,23 @@
1
  ---
2
  tags:
3
  - generated_from_trainer
 
 
4
  metrics:
5
  - accuracy
6
  model-index:
7
  - name: tiny_bert_rand_10_v2
8
- results: []
 
 
 
 
 
 
 
 
 
 
9
  ---
10
 
11
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -13,10 +25,10 @@ should probably proofread and complete it, then remove this comment. -->
13
 
14
  # tiny_bert_rand_10_v2
15
 
16
- This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
17
  It achieves the following results on the evaluation set:
18
- - Loss: 10.0922
19
- - Accuracy: 0.1518
20
 
21
  ## Model description
22
 
 
1
  ---
2
  tags:
3
  - generated_from_trainer
4
+ datasets:
5
+ - Hartunka/processed_wikitext-103-raw-v1-rand-10_v2
6
  metrics:
7
  - accuracy
8
  model-index:
9
  - name: tiny_bert_rand_10_v2
10
+ results:
11
+ - task:
12
+ name: Masked Language Modeling
13
+ type: fill-mask
14
+ dataset:
15
+ name: Hartunka/processed_wikitext-103-raw-v1-rand-10_v2
16
+ type: Hartunka/processed_wikitext-103-raw-v1-rand-10_v2
17
+ metrics:
18
+ - name: Accuracy
19
+ type: accuracy
20
+ value: 0.15299203254788948
21
  ---
22
 
23
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
25
 
26
  # tiny_bert_rand_10_v2
27
 
28
+ This model is a fine-tuned version of [](https://huggingface.co/) on the Hartunka/processed_wikitext-103-raw-v1-rand-10_v2 dataset.
29
  It achieves the following results on the evaluation set:
30
+ - Loss: 8.5389
31
+ - Accuracy: 0.1530
32
 
33
  ## Model description
34
 
all_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 25.0,
3
+ "eval_accuracy": 0.15299203254788948,
4
+ "eval_loss": 8.538854598999023,
5
+ "eval_runtime": 0.7173,
6
+ "eval_samples": 479,
7
+ "eval_samples_per_second": 667.787,
8
+ "eval_steps_per_second": 6.971,
9
+ "perplexity": 5109.488595336697,
10
+ "total_flos": 3.050247306064896e+17,
11
+ "train_loss": 7.77363054584596,
12
+ "train_runtime": 11901.4866,
13
+ "train_samples": 228639,
14
+ "train_samples_per_second": 480.274,
15
+ "train_steps_per_second": 5.004
16
+ }
eval_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 25.0,
3
+ "eval_accuracy": 0.15299203254788948,
4
+ "eval_loss": 8.538854598999023,
5
+ "eval_runtime": 0.7173,
6
+ "eval_samples": 479,
7
+ "eval_samples_per_second": 667.787,
8
+ "eval_steps_per_second": 6.971,
9
+ "perplexity": 5109.488595336697
10
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 25.0,
3
+ "total_flos": 3.050247306064896e+17,
4
+ "train_loss": 7.77363054584596,
5
+ "train_runtime": 11901.4866,
6
+ "train_samples": 228639,
7
+ "train_samples_per_second": 480.274,
8
+ "train_steps_per_second": 5.004
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,908 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 8.516840934753418,
3
+ "best_model_checkpoint": "tiny_bert_rand_10_v2/checkpoint-10000",
4
+ "epoch": 25.0,
5
+ "eval_steps": 10000,
6
+ "global_step": 59550,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.20990764063811923,
13
+ "grad_norm": 2.8956918716430664,
14
+ "learning_rate": 5e-06,
15
+ "loss": 12.0528,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.41981528127623846,
20
+ "grad_norm": 1.7897758483886719,
21
+ "learning_rate": 1e-05,
22
+ "loss": 10.4985,
23
+ "step": 1000
24
+ },
25
+ {
26
+ "epoch": 0.6297229219143576,
27
+ "grad_norm": 1.651282787322998,
28
+ "learning_rate": 1.5e-05,
29
+ "loss": 9.511,
30
+ "step": 1500
31
+ },
32
+ {
33
+ "epoch": 0.8396305625524769,
34
+ "grad_norm": 1.1039087772369385,
35
+ "learning_rate": 2e-05,
36
+ "loss": 9.313,
37
+ "step": 2000
38
+ },
39
+ {
40
+ "epoch": 1.0495382031905962,
41
+ "grad_norm": 1.4416009187698364,
42
+ "learning_rate": 2.5e-05,
43
+ "loss": 9.1977,
44
+ "step": 2500
45
+ },
46
+ {
47
+ "epoch": 1.2594458438287153,
48
+ "grad_norm": 1.8553386926651,
49
+ "learning_rate": 3e-05,
50
+ "loss": 9.0932,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 1.4693534844668346,
55
+ "grad_norm": 1.4036210775375366,
56
+ "learning_rate": 3.5e-05,
57
+ "loss": 9.0077,
58
+ "step": 3500
59
+ },
60
+ {
61
+ "epoch": 1.6792611251049538,
62
+ "grad_norm": 1.5917521715164185,
63
+ "learning_rate": 4e-05,
64
+ "loss": 8.9268,
65
+ "step": 4000
66
+ },
67
+ {
68
+ "epoch": 1.8891687657430731,
69
+ "grad_norm": 1.7779933214187622,
70
+ "learning_rate": 4.5e-05,
71
+ "loss": 8.8497,
72
+ "step": 4500
73
+ },
74
+ {
75
+ "epoch": 2.0990764063811924,
76
+ "grad_norm": 1.4821877479553223,
77
+ "learning_rate": 5e-05,
78
+ "loss": 8.7864,
79
+ "step": 5000
80
+ },
81
+ {
82
+ "epoch": 2.3089840470193117,
83
+ "grad_norm": 1.7116888761520386,
84
+ "learning_rate": 5.500000000000001e-05,
85
+ "loss": 8.7418,
86
+ "step": 5500
87
+ },
88
+ {
89
+ "epoch": 2.5188916876574305,
90
+ "grad_norm": 1.8798326253890991,
91
+ "learning_rate": 6e-05,
92
+ "loss": 8.7053,
93
+ "step": 6000
94
+ },
95
+ {
96
+ "epoch": 2.72879932829555,
97
+ "grad_norm": 2.150996685028076,
98
+ "learning_rate": 6.500000000000001e-05,
99
+ "loss": 8.6733,
100
+ "step": 6500
101
+ },
102
+ {
103
+ "epoch": 2.938706968933669,
104
+ "grad_norm": 1.4576572179794312,
105
+ "learning_rate": 7e-05,
106
+ "loss": 8.6438,
107
+ "step": 7000
108
+ },
109
+ {
110
+ "epoch": 3.1486146095717884,
111
+ "grad_norm": 2.110964775085449,
112
+ "learning_rate": 7.500000000000001e-05,
113
+ "loss": 8.603,
114
+ "step": 7500
115
+ },
116
+ {
117
+ "epoch": 3.3585222502099077,
118
+ "grad_norm": 2.01837158203125,
119
+ "learning_rate": 8e-05,
120
+ "loss": 8.5734,
121
+ "step": 8000
122
+ },
123
+ {
124
+ "epoch": 3.568429890848027,
125
+ "grad_norm": 2.289891481399536,
126
+ "learning_rate": 8.5e-05,
127
+ "loss": 8.5586,
128
+ "step": 8500
129
+ },
130
+ {
131
+ "epoch": 3.7783375314861463,
132
+ "grad_norm": 2.272826671600342,
133
+ "learning_rate": 9e-05,
134
+ "loss": 8.5364,
135
+ "step": 9000
136
+ },
137
+ {
138
+ "epoch": 3.988245172124265,
139
+ "grad_norm": 1.7595138549804688,
140
+ "learning_rate": 9.5e-05,
141
+ "loss": 8.517,
142
+ "step": 9500
143
+ },
144
+ {
145
+ "epoch": 4.198152812762385,
146
+ "grad_norm": 2.2657644748687744,
147
+ "learning_rate": 0.0001,
148
+ "loss": 8.4324,
149
+ "step": 10000
150
+ },
151
+ {
152
+ "epoch": 4.198152812762385,
153
+ "eval_accuracy": 0.15014329836667686,
154
+ "eval_loss": 8.516840934753418,
155
+ "eval_runtime": 0.7382,
156
+ "eval_samples_per_second": 648.908,
157
+ "eval_steps_per_second": 6.774,
158
+ "step": 10000
159
+ },
160
+ {
161
+ "epoch": 4.408060453400504,
162
+ "grad_norm": 2.0055155754089355,
163
+ "learning_rate": 9.899091826437942e-05,
164
+ "loss": 8.4296,
165
+ "step": 10500
166
+ },
167
+ {
168
+ "epoch": 4.617968094038623,
169
+ "grad_norm": 2.568305015563965,
170
+ "learning_rate": 9.798183652875883e-05,
171
+ "loss": 8.4221,
172
+ "step": 11000
173
+ },
174
+ {
175
+ "epoch": 4.827875734676742,
176
+ "grad_norm": 2.9037818908691406,
177
+ "learning_rate": 9.697275479313825e-05,
178
+ "loss": 8.4254,
179
+ "step": 11500
180
+ },
181
+ {
182
+ "epoch": 5.037783375314861,
183
+ "grad_norm": 1.9787847995758057,
184
+ "learning_rate": 9.596367305751767e-05,
185
+ "loss": 8.3976,
186
+ "step": 12000
187
+ },
188
+ {
189
+ "epoch": 5.247691015952981,
190
+ "grad_norm": 2.553103446960449,
191
+ "learning_rate": 9.495459132189707e-05,
192
+ "loss": 8.3192,
193
+ "step": 12500
194
+ },
195
+ {
196
+ "epoch": 5.4575986565911,
197
+ "grad_norm": 2.580944299697876,
198
+ "learning_rate": 9.394550958627649e-05,
199
+ "loss": 8.3209,
200
+ "step": 13000
201
+ },
202
+ {
203
+ "epoch": 5.667506297229219,
204
+ "grad_norm": 2.5693955421447754,
205
+ "learning_rate": 9.293642785065591e-05,
206
+ "loss": 8.3257,
207
+ "step": 13500
208
+ },
209
+ {
210
+ "epoch": 5.877413937867338,
211
+ "grad_norm": 2.989201068878174,
212
+ "learning_rate": 9.192734611503532e-05,
213
+ "loss": 8.3246,
214
+ "step": 14000
215
+ },
216
+ {
217
+ "epoch": 6.087321578505458,
218
+ "grad_norm": 2.8055412769317627,
219
+ "learning_rate": 9.091826437941474e-05,
220
+ "loss": 8.288,
221
+ "step": 14500
222
+ },
223
+ {
224
+ "epoch": 6.297229219143577,
225
+ "grad_norm": 2.1214358806610107,
226
+ "learning_rate": 8.990918264379415e-05,
227
+ "loss": 8.2457,
228
+ "step": 15000
229
+ },
230
+ {
231
+ "epoch": 6.507136859781696,
232
+ "grad_norm": 2.7707295417785645,
233
+ "learning_rate": 8.890010090817356e-05,
234
+ "loss": 8.2513,
235
+ "step": 15500
236
+ },
237
+ {
238
+ "epoch": 6.717044500419815,
239
+ "grad_norm": 1.9960660934448242,
240
+ "learning_rate": 8.789101917255298e-05,
241
+ "loss": 8.2557,
242
+ "step": 16000
243
+ },
244
+ {
245
+ "epoch": 6.926952141057934,
246
+ "grad_norm": 2.213331937789917,
247
+ "learning_rate": 8.68819374369324e-05,
248
+ "loss": 8.2587,
249
+ "step": 16500
250
+ },
251
+ {
252
+ "epoch": 7.136859781696054,
253
+ "grad_norm": 2.097681760787964,
254
+ "learning_rate": 8.58728557013118e-05,
255
+ "loss": 8.2045,
256
+ "step": 17000
257
+ },
258
+ {
259
+ "epoch": 7.346767422334173,
260
+ "grad_norm": 2.481717824935913,
261
+ "learning_rate": 8.486377396569122e-05,
262
+ "loss": 8.1824,
263
+ "step": 17500
264
+ },
265
+ {
266
+ "epoch": 7.5566750629722925,
267
+ "grad_norm": 2.7683825492858887,
268
+ "learning_rate": 8.385469223007064e-05,
269
+ "loss": 8.1951,
270
+ "step": 18000
271
+ },
272
+ {
273
+ "epoch": 7.766582703610411,
274
+ "grad_norm": 2.1083154678344727,
275
+ "learning_rate": 8.284561049445006e-05,
276
+ "loss": 8.2034,
277
+ "step": 18500
278
+ },
279
+ {
280
+ "epoch": 7.97649034424853,
281
+ "grad_norm": 1.9768694639205933,
282
+ "learning_rate": 8.183652875882947e-05,
283
+ "loss": 8.2037,
284
+ "step": 19000
285
+ },
286
+ {
287
+ "epoch": 8.18639798488665,
288
+ "grad_norm": 2.308565855026245,
289
+ "learning_rate": 8.082744702320889e-05,
290
+ "loss": 8.1362,
291
+ "step": 19500
292
+ },
293
+ {
294
+ "epoch": 8.39630562552477,
295
+ "grad_norm": 1.9111008644104004,
296
+ "learning_rate": 7.98183652875883e-05,
297
+ "loss": 8.1359,
298
+ "step": 20000
299
+ },
300
+ {
301
+ "epoch": 8.39630562552477,
302
+ "eval_accuracy": 0.1513510480092034,
303
+ "eval_loss": 8.578495025634766,
304
+ "eval_runtime": 0.6882,
305
+ "eval_samples_per_second": 696.015,
306
+ "eval_steps_per_second": 7.265,
307
+ "step": 20000
308
+ },
309
+ {
310
+ "epoch": 8.606213266162888,
311
+ "grad_norm": 2.631357192993164,
312
+ "learning_rate": 7.880928355196771e-05,
313
+ "loss": 8.1384,
314
+ "step": 20500
315
+ },
316
+ {
317
+ "epoch": 8.816120906801007,
318
+ "grad_norm": 2.378553628921509,
319
+ "learning_rate": 7.780020181634713e-05,
320
+ "loss": 8.1512,
321
+ "step": 21000
322
+ },
323
+ {
324
+ "epoch": 9.026028547439127,
325
+ "grad_norm": 2.799654722213745,
326
+ "learning_rate": 7.679112008072655e-05,
327
+ "loss": 8.1354,
328
+ "step": 21500
329
+ },
330
+ {
331
+ "epoch": 9.235936188077247,
332
+ "grad_norm": 3.0007619857788086,
333
+ "learning_rate": 7.578203834510595e-05,
334
+ "loss": 8.0683,
335
+ "step": 22000
336
+ },
337
+ {
338
+ "epoch": 9.445843828715365,
339
+ "grad_norm": 3.1445086002349854,
340
+ "learning_rate": 7.477295660948537e-05,
341
+ "loss": 8.0738,
342
+ "step": 22500
343
+ },
344
+ {
345
+ "epoch": 9.655751469353484,
346
+ "grad_norm": 2.6713919639587402,
347
+ "learning_rate": 7.376387487386479e-05,
348
+ "loss": 8.0859,
349
+ "step": 23000
350
+ },
351
+ {
352
+ "epoch": 9.865659109991604,
353
+ "grad_norm": 2.236157178878784,
354
+ "learning_rate": 7.27547931382442e-05,
355
+ "loss": 8.0858,
356
+ "step": 23500
357
+ },
358
+ {
359
+ "epoch": 10.075566750629722,
360
+ "grad_norm": 2.471778154373169,
361
+ "learning_rate": 7.174571140262362e-05,
362
+ "loss": 8.0462,
363
+ "step": 24000
364
+ },
365
+ {
366
+ "epoch": 10.285474391267842,
367
+ "grad_norm": 3.3160719871520996,
368
+ "learning_rate": 7.073662966700304e-05,
369
+ "loss": 7.9911,
370
+ "step": 24500
371
+ },
372
+ {
373
+ "epoch": 10.495382031905962,
374
+ "grad_norm": 2.7243497371673584,
375
+ "learning_rate": 6.972754793138244e-05,
376
+ "loss": 7.9938,
377
+ "step": 25000
378
+ },
379
+ {
380
+ "epoch": 10.705289672544081,
381
+ "grad_norm": 3.617450475692749,
382
+ "learning_rate": 6.871846619576186e-05,
383
+ "loss": 8.0075,
384
+ "step": 25500
385
+ },
386
+ {
387
+ "epoch": 10.9151973131822,
388
+ "grad_norm": 3.0119311809539795,
389
+ "learning_rate": 6.770938446014128e-05,
390
+ "loss": 8.0181,
391
+ "step": 26000
392
+ },
393
+ {
394
+ "epoch": 11.125104953820319,
395
+ "grad_norm": 3.3908631801605225,
396
+ "learning_rate": 6.670030272452068e-05,
397
+ "loss": 7.93,
398
+ "step": 26500
399
+ },
400
+ {
401
+ "epoch": 11.335012594458439,
402
+ "grad_norm": 3.8109664916992188,
403
+ "learning_rate": 6.56912209889001e-05,
404
+ "loss": 7.8902,
405
+ "step": 27000
406
+ },
407
+ {
408
+ "epoch": 11.544920235096557,
409
+ "grad_norm": 3.2663817405700684,
410
+ "learning_rate": 6.468213925327952e-05,
411
+ "loss": 7.9052,
412
+ "step": 27500
413
+ },
414
+ {
415
+ "epoch": 11.754827875734676,
416
+ "grad_norm": 3.765650749206543,
417
+ "learning_rate": 6.367305751765893e-05,
418
+ "loss": 7.9112,
419
+ "step": 28000
420
+ },
421
+ {
422
+ "epoch": 11.964735516372796,
423
+ "grad_norm": 3.401289463043213,
424
+ "learning_rate": 6.266397578203835e-05,
425
+ "loss": 7.9237,
426
+ "step": 28500
427
+ },
428
+ {
429
+ "epoch": 12.174643157010916,
430
+ "grad_norm": 3.8314595222473145,
431
+ "learning_rate": 6.165489404641777e-05,
432
+ "loss": 7.7856,
433
+ "step": 29000
434
+ },
435
+ {
436
+ "epoch": 12.384550797649034,
437
+ "grad_norm": 4.359664440155029,
438
+ "learning_rate": 6.064581231079718e-05,
439
+ "loss": 7.7689,
440
+ "step": 29500
441
+ },
442
+ {
443
+ "epoch": 12.594458438287154,
444
+ "grad_norm": 4.1663923263549805,
445
+ "learning_rate": 5.963673057517659e-05,
446
+ "loss": 7.791,
447
+ "step": 30000
448
+ },
449
+ {
450
+ "epoch": 12.594458438287154,
451
+ "eval_accuracy": 0.15407001147270336,
452
+ "eval_loss": 9.017232894897461,
453
+ "eval_runtime": 0.6928,
454
+ "eval_samples_per_second": 691.393,
455
+ "eval_steps_per_second": 7.217,
456
+ "step": 30000
457
+ },
458
+ {
459
+ "epoch": 12.804366078925273,
460
+ "grad_norm": 3.787510871887207,
461
+ "learning_rate": 5.8627648839556e-05,
462
+ "loss": 7.8107,
463
+ "step": 30500
464
+ },
465
+ {
466
+ "epoch": 13.014273719563391,
467
+ "grad_norm": 4.561687469482422,
468
+ "learning_rate": 5.7618567103935416e-05,
469
+ "loss": 7.8127,
470
+ "step": 31000
471
+ },
472
+ {
473
+ "epoch": 13.224181360201511,
474
+ "grad_norm": 4.102744102478027,
475
+ "learning_rate": 5.660948536831484e-05,
476
+ "loss": 7.6261,
477
+ "step": 31500
478
+ },
479
+ {
480
+ "epoch": 13.43408900083963,
481
+ "grad_norm": 3.8658947944641113,
482
+ "learning_rate": 5.5600403632694254e-05,
483
+ "loss": 7.6531,
484
+ "step": 32000
485
+ },
486
+ {
487
+ "epoch": 13.64399664147775,
488
+ "grad_norm": 4.556894779205322,
489
+ "learning_rate": 5.4591321897073666e-05,
490
+ "loss": 7.683,
491
+ "step": 32500
492
+ },
493
+ {
494
+ "epoch": 13.853904282115868,
495
+ "grad_norm": 4.922505855560303,
496
+ "learning_rate": 5.358224016145308e-05,
497
+ "loss": 7.6933,
498
+ "step": 33000
499
+ },
500
+ {
501
+ "epoch": 14.063811922753988,
502
+ "grad_norm": 4.5583696365356445,
503
+ "learning_rate": 5.257315842583249e-05,
504
+ "loss": 7.6311,
505
+ "step": 33500
506
+ },
507
+ {
508
+ "epoch": 14.273719563392108,
509
+ "grad_norm": 5.245599269866943,
510
+ "learning_rate": 5.15640766902119e-05,
511
+ "loss": 7.5044,
512
+ "step": 34000
513
+ },
514
+ {
515
+ "epoch": 14.483627204030226,
516
+ "grad_norm": 4.251413822174072,
517
+ "learning_rate": 5.055499495459133e-05,
518
+ "loss": 7.5302,
519
+ "step": 34500
520
+ },
521
+ {
522
+ "epoch": 14.693534844668346,
523
+ "grad_norm": 4.954182147979736,
524
+ "learning_rate": 4.954591321897074e-05,
525
+ "loss": 7.5437,
526
+ "step": 35000
527
+ },
528
+ {
529
+ "epoch": 14.903442485306465,
530
+ "grad_norm": 5.95086145401001,
531
+ "learning_rate": 4.853683148335015e-05,
532
+ "loss": 7.5642,
533
+ "step": 35500
534
+ },
535
+ {
536
+ "epoch": 15.113350125944585,
537
+ "grad_norm": 5.5313310623168945,
538
+ "learning_rate": 4.7527749747729566e-05,
539
+ "loss": 7.4553,
540
+ "step": 36000
541
+ },
542
+ {
543
+ "epoch": 15.323257766582703,
544
+ "grad_norm": 6.156241416931152,
545
+ "learning_rate": 4.6518668012108985e-05,
546
+ "loss": 7.3874,
547
+ "step": 36500
548
+ },
549
+ {
550
+ "epoch": 15.533165407220823,
551
+ "grad_norm": 5.391365051269531,
552
+ "learning_rate": 4.55095862764884e-05,
553
+ "loss": 7.4066,
554
+ "step": 37000
555
+ },
556
+ {
557
+ "epoch": 15.743073047858942,
558
+ "grad_norm": 5.771765232086182,
559
+ "learning_rate": 4.450050454086781e-05,
560
+ "loss": 7.4263,
561
+ "step": 37500
562
+ },
563
+ {
564
+ "epoch": 15.95298068849706,
565
+ "grad_norm": 4.8253631591796875,
566
+ "learning_rate": 4.349142280524723e-05,
567
+ "loss": 7.4372,
568
+ "step": 38000
569
+ },
570
+ {
571
+ "epoch": 16.162888329135182,
572
+ "grad_norm": 5.755190849304199,
573
+ "learning_rate": 4.248234106962664e-05,
574
+ "loss": 7.2846,
575
+ "step": 38500
576
+ },
577
+ {
578
+ "epoch": 16.3727959697733,
579
+ "grad_norm": 5.651447772979736,
580
+ "learning_rate": 4.147325933400605e-05,
581
+ "loss": 7.2686,
582
+ "step": 39000
583
+ },
584
+ {
585
+ "epoch": 16.582703610411418,
586
+ "grad_norm": 5.568587303161621,
587
+ "learning_rate": 4.046417759838547e-05,
588
+ "loss": 7.2883,
589
+ "step": 39500
590
+ },
591
+ {
592
+ "epoch": 16.79261125104954,
593
+ "grad_norm": 6.008751392364502,
594
+ "learning_rate": 3.9455095862764884e-05,
595
+ "loss": 7.3,
596
+ "step": 40000
597
+ },
598
+ {
599
+ "epoch": 16.79261125104954,
600
+ "eval_accuracy": 0.15122799147695412,
601
+ "eval_loss": 9.511892318725586,
602
+ "eval_runtime": 0.6891,
603
+ "eval_samples_per_second": 695.131,
604
+ "eval_steps_per_second": 7.256,
605
+ "step": 40000
606
+ },
607
+ {
608
+ "epoch": 17.002518891687657,
609
+ "grad_norm": 6.794548511505127,
610
+ "learning_rate": 3.84460141271443e-05,
611
+ "loss": 7.3289,
612
+ "step": 40500
613
+ },
614
+ {
615
+ "epoch": 17.212426532325775,
616
+ "grad_norm": 6.964717388153076,
617
+ "learning_rate": 3.7436932391523715e-05,
618
+ "loss": 7.1414,
619
+ "step": 41000
620
+ },
621
+ {
622
+ "epoch": 17.422334172963897,
623
+ "grad_norm": 6.341984748840332,
624
+ "learning_rate": 3.642785065590313e-05,
625
+ "loss": 7.1565,
626
+ "step": 41500
627
+ },
628
+ {
629
+ "epoch": 17.632241813602015,
630
+ "grad_norm": 7.717092514038086,
631
+ "learning_rate": 3.541876892028255e-05,
632
+ "loss": 7.1761,
633
+ "step": 42000
634
+ },
635
+ {
636
+ "epoch": 17.842149454240133,
637
+ "grad_norm": 6.288142204284668,
638
+ "learning_rate": 3.440968718466196e-05,
639
+ "loss": 7.1904,
640
+ "step": 42500
641
+ },
642
+ {
643
+ "epoch": 18.052057094878254,
644
+ "grad_norm": 7.551562309265137,
645
+ "learning_rate": 3.340060544904137e-05,
646
+ "loss": 7.1564,
647
+ "step": 43000
648
+ },
649
+ {
650
+ "epoch": 18.261964735516372,
651
+ "grad_norm": 6.559009075164795,
652
+ "learning_rate": 3.239152371342079e-05,
653
+ "loss": 7.0359,
654
+ "step": 43500
655
+ },
656
+ {
657
+ "epoch": 18.471872376154494,
658
+ "grad_norm": 8.913935661315918,
659
+ "learning_rate": 3.13824419778002e-05,
660
+ "loss": 7.0632,
661
+ "step": 44000
662
+ },
663
+ {
664
+ "epoch": 18.68178001679261,
665
+ "grad_norm": 7.239193916320801,
666
+ "learning_rate": 3.0373360242179615e-05,
667
+ "loss": 7.0772,
668
+ "step": 44500
669
+ },
670
+ {
671
+ "epoch": 18.89168765743073,
672
+ "grad_norm": 7.293575286865234,
673
+ "learning_rate": 2.9364278506559034e-05,
674
+ "loss": 7.1013,
675
+ "step": 45000
676
+ },
677
+ {
678
+ "epoch": 19.10159529806885,
679
+ "grad_norm": 7.253826141357422,
680
+ "learning_rate": 2.8355196770938446e-05,
681
+ "loss": 7.0193,
682
+ "step": 45500
683
+ },
684
+ {
685
+ "epoch": 19.31150293870697,
686
+ "grad_norm": 7.525722503662109,
687
+ "learning_rate": 2.734611503531786e-05,
688
+ "loss": 6.9527,
689
+ "step": 46000
690
+ },
691
+ {
692
+ "epoch": 19.521410579345087,
693
+ "grad_norm": 8.178430557250977,
694
+ "learning_rate": 2.6337033299697278e-05,
695
+ "loss": 6.9639,
696
+ "step": 46500
697
+ },
698
+ {
699
+ "epoch": 19.73131821998321,
700
+ "grad_norm": 8.326142311096191,
701
+ "learning_rate": 2.532795156407669e-05,
702
+ "loss": 6.9762,
703
+ "step": 47000
704
+ },
705
+ {
706
+ "epoch": 19.941225860621326,
707
+ "grad_norm": 6.413033485412598,
708
+ "learning_rate": 2.4318869828456106e-05,
709
+ "loss": 6.99,
710
+ "step": 47500
711
+ },
712
+ {
713
+ "epoch": 20.151133501259444,
714
+ "grad_norm": 8.123743057250977,
715
+ "learning_rate": 2.330978809283552e-05,
716
+ "loss": 6.9007,
717
+ "step": 48000
718
+ },
719
+ {
720
+ "epoch": 20.361041141897566,
721
+ "grad_norm": 7.148409366607666,
722
+ "learning_rate": 2.2300706357214934e-05,
723
+ "loss": 6.8722,
724
+ "step": 48500
725
+ },
726
+ {
727
+ "epoch": 20.570948782535684,
728
+ "grad_norm": 6.879838466644287,
729
+ "learning_rate": 2.129162462159435e-05,
730
+ "loss": 6.8796,
731
+ "step": 49000
732
+ },
733
+ {
734
+ "epoch": 20.7808564231738,
735
+ "grad_norm": 7.241456985473633,
736
+ "learning_rate": 2.0282542885973765e-05,
737
+ "loss": 6.8861,
738
+ "step": 49500
739
+ },
740
+ {
741
+ "epoch": 20.990764063811923,
742
+ "grad_norm": 7.6523895263671875,
743
+ "learning_rate": 1.927346115035318e-05,
744
+ "loss": 6.9042,
745
+ "step": 50000
746
+ },
747
+ {
748
+ "epoch": 20.990764063811923,
749
+ "eval_accuracy": 0.1517929738974402,
750
+ "eval_loss": 10.092184066772461,
751
+ "eval_runtime": 0.6859,
752
+ "eval_samples_per_second": 698.316,
753
+ "eval_steps_per_second": 7.289,
754
+ "step": 50000
755
+ },
756
+ {
757
+ "epoch": 21.20067170445004,
758
+ "grad_norm": 7.539754867553711,
759
+ "learning_rate": 1.8264379414732593e-05,
760
+ "loss": 6.7883,
761
+ "step": 50500
762
+ },
763
+ {
764
+ "epoch": 21.410579345088163,
765
+ "grad_norm": 7.285797595977783,
766
+ "learning_rate": 1.725529767911201e-05,
767
+ "loss": 6.7968,
768
+ "step": 51000
769
+ },
770
+ {
771
+ "epoch": 21.62048698572628,
772
+ "grad_norm": 8.29957389831543,
773
+ "learning_rate": 1.6246215943491424e-05,
774
+ "loss": 6.8139,
775
+ "step": 51500
776
+ },
777
+ {
778
+ "epoch": 21.8303946263644,
779
+ "grad_norm": 8.12922477722168,
780
+ "learning_rate": 1.5237134207870837e-05,
781
+ "loss": 6.8139,
782
+ "step": 52000
783
+ },
784
+ {
785
+ "epoch": 22.04030226700252,
786
+ "grad_norm": 7.652002334594727,
787
+ "learning_rate": 1.4228052472250252e-05,
788
+ "loss": 6.7965,
789
+ "step": 52500
790
+ },
791
+ {
792
+ "epoch": 22.250209907640638,
793
+ "grad_norm": 9.461636543273926,
794
+ "learning_rate": 1.3218970736629668e-05,
795
+ "loss": 6.7156,
796
+ "step": 53000
797
+ },
798
+ {
799
+ "epoch": 22.460117548278756,
800
+ "grad_norm": 6.493343830108643,
801
+ "learning_rate": 1.2209889001009082e-05,
802
+ "loss": 6.7325,
803
+ "step": 53500
804
+ },
805
+ {
806
+ "epoch": 22.670025188916878,
807
+ "grad_norm": 7.998405456542969,
808
+ "learning_rate": 1.1200807265388496e-05,
809
+ "loss": 6.7438,
810
+ "step": 54000
811
+ },
812
+ {
813
+ "epoch": 22.879932829554996,
814
+ "grad_norm": 8.43569278717041,
815
+ "learning_rate": 1.0191725529767912e-05,
816
+ "loss": 6.7429,
817
+ "step": 54500
818
+ },
819
+ {
820
+ "epoch": 23.089840470193113,
821
+ "grad_norm": 8.705036163330078,
822
+ "learning_rate": 9.182643794147326e-06,
823
+ "loss": 6.7146,
824
+ "step": 55000
825
+ },
826
+ {
827
+ "epoch": 23.299748110831235,
828
+ "grad_norm": 8.198236465454102,
829
+ "learning_rate": 8.173562058526741e-06,
830
+ "loss": 6.6754,
831
+ "step": 55500
832
+ },
833
+ {
834
+ "epoch": 23.509655751469353,
835
+ "grad_norm": 8.432761192321777,
836
+ "learning_rate": 7.164480322906156e-06,
837
+ "loss": 6.6801,
838
+ "step": 56000
839
+ },
840
+ {
841
+ "epoch": 23.719563392107474,
842
+ "grad_norm": 8.555904388427734,
843
+ "learning_rate": 6.155398587285571e-06,
844
+ "loss": 6.6826,
845
+ "step": 56500
846
+ },
847
+ {
848
+ "epoch": 23.929471032745592,
849
+ "grad_norm": 8.87362003326416,
850
+ "learning_rate": 5.146316851664986e-06,
851
+ "loss": 6.6981,
852
+ "step": 57000
853
+ },
854
+ {
855
+ "epoch": 24.13937867338371,
856
+ "grad_norm": 6.883061408996582,
857
+ "learning_rate": 4.1372351160444e-06,
858
+ "loss": 6.6573,
859
+ "step": 57500
860
+ },
861
+ {
862
+ "epoch": 24.349286314021832,
863
+ "grad_norm": 7.8268585205078125,
864
+ "learning_rate": 3.1281533804238145e-06,
865
+ "loss": 6.655,
866
+ "step": 58000
867
+ },
868
+ {
869
+ "epoch": 24.55919395465995,
870
+ "grad_norm": 8.802783012390137,
871
+ "learning_rate": 2.1190716448032293e-06,
872
+ "loss": 6.6503,
873
+ "step": 58500
874
+ },
875
+ {
876
+ "epoch": 24.769101595298068,
877
+ "grad_norm": 9.875340461730957,
878
+ "learning_rate": 1.109989909182644e-06,
879
+ "loss": 6.6436,
880
+ "step": 59000
881
+ },
882
+ {
883
+ "epoch": 24.97900923593619,
884
+ "grad_norm": 6.891987323760986,
885
+ "learning_rate": 1.0090817356205853e-07,
886
+ "loss": 6.6469,
887
+ "step": 59500
888
+ },
889
+ {
890
+ "epoch": 25.0,
891
+ "step": 59550,
892
+ "total_flos": 3.050247306064896e+17,
893
+ "train_loss": 7.77363054584596,
894
+ "train_runtime": 11901.4866,
895
+ "train_samples_per_second": 480.274,
896
+ "train_steps_per_second": 5.004
897
+ }
898
+ ],
899
+ "logging_steps": 500,
900
+ "max_steps": 59550,
901
+ "num_input_tokens_seen": 0,
902
+ "num_train_epochs": 25,
903
+ "save_steps": 10000,
904
+ "total_flos": 3.050247306064896e+17,
905
+ "train_batch_size": 96,
906
+ "trial_name": null,
907
+ "trial_params": null
908
+ }