th041 commited on
Commit
30d2f68
1 Parent(s): 056be51

🍻 cheers

Browse files
README.md CHANGED
@@ -2,6 +2,7 @@
2
  license: apache-2.0
3
  base_model: google/vit-base-patch16-224-in21k
4
  tags:
 
5
  - generated_from_trainer
6
  datasets:
7
  - imagefolder
@@ -22,7 +23,7 @@ model-index:
22
  metrics:
23
  - name: Accuracy
24
  type: accuracy
25
- value: 0.9460431654676259
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,8 +33,8 @@ should probably proofread and complete it, then remove this comment. -->
32
 
33
  This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the imagefolder dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 0.3006
36
- - Accuracy: 0.9460
37
 
38
  ## Model description
39
 
 
2
  license: apache-2.0
3
  base_model: google/vit-base-patch16-224-in21k
4
  tags:
5
+ - image-classification
6
  - generated_from_trainer
7
  datasets:
8
  - imagefolder
 
23
  metrics:
24
  - name: Accuracy
25
  type: accuracy
26
+ value: 0.920863309352518
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
33
 
34
  This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the imagefolder dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 0.2671
37
+ - Accuracy: 0.9209
38
 
39
  ## Model description
40
 
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 13.0,
3
+ "eval_accuracy": 0.920863309352518,
4
+ "eval_loss": 0.26709362864494324,
5
+ "eval_runtime": 5.1557,
6
+ "eval_samples_per_second": 53.921,
7
+ "eval_steps_per_second": 6.789,
8
+ "total_flos": 2.5134977377120666e+18,
9
+ "train_loss": 0.18042536613924515,
10
+ "train_runtime": 1249.5195,
11
+ "train_samples_per_second": 25.958,
12
+ "train_steps_per_second": 1.623
13
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 13.0,
3
+ "eval_accuracy": 0.920863309352518,
4
+ "eval_loss": 0.26709362864494324,
5
+ "eval_runtime": 5.1557,
6
+ "eval_samples_per_second": 53.921,
7
+ "eval_steps_per_second": 6.789
8
+ }
runs/Jun18_14-55-21_70f652825eb2/events.out.tfevents.1718723811.70f652825eb2.179.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:589fed6ca7dfb5ca8f2f6122d07d929e68a4523bef9c8973f41712a0940e26bd
3
+ size 411
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 13.0,
3
+ "total_flos": 2.5134977377120666e+18,
4
+ "train_loss": 0.18042536613924515,
5
+ "train_runtime": 1249.5195,
6
+ "train_samples_per_second": 25.958,
7
+ "train_steps_per_second": 1.623
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,1636 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.26709362864494324,
3
+ "best_model_checkpoint": "vit-weldclassifyv3/checkpoint-1000",
4
+ "epoch": 13.0,
5
+ "eval_steps": 100,
6
+ "global_step": 2028,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0641025641025641,
13
+ "grad_norm": 1.1238359212875366,
14
+ "learning_rate": 0.0001990138067061144,
15
+ "loss": 1.2792,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.1282051282051282,
20
+ "grad_norm": 1.7609381675720215,
21
+ "learning_rate": 0.0001980276134122288,
22
+ "loss": 1.1608,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.19230769230769232,
27
+ "grad_norm": 1.1302402019500732,
28
+ "learning_rate": 0.0001970414201183432,
29
+ "loss": 1.1137,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.2564102564102564,
34
+ "grad_norm": 3.0818381309509277,
35
+ "learning_rate": 0.0001960552268244576,
36
+ "loss": 1.0182,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.32051282051282054,
41
+ "grad_norm": 1.0849112272262573,
42
+ "learning_rate": 0.000195069033530572,
43
+ "loss": 1.2566,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.38461538461538464,
48
+ "grad_norm": 1.0367400646209717,
49
+ "learning_rate": 0.0001940828402366864,
50
+ "loss": 1.0554,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.44871794871794873,
55
+ "grad_norm": 2.154937744140625,
56
+ "learning_rate": 0.0001930966469428008,
57
+ "loss": 1.0788,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.5128205128205128,
62
+ "grad_norm": 1.1957019567489624,
63
+ "learning_rate": 0.0001921104536489152,
64
+ "loss": 1.2299,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.5769230769230769,
69
+ "grad_norm": 1.3072288036346436,
70
+ "learning_rate": 0.0001911242603550296,
71
+ "loss": 1.0164,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.6410256410256411,
76
+ "grad_norm": 1.340012550354004,
77
+ "learning_rate": 0.00019013806706114398,
78
+ "loss": 0.8398,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.6410256410256411,
83
+ "eval_accuracy": 0.5035971223021583,
84
+ "eval_loss": 1.031232476234436,
85
+ "eval_runtime": 2.6053,
86
+ "eval_samples_per_second": 106.706,
87
+ "eval_steps_per_second": 13.434,
88
+ "step": 100
89
+ },
90
+ {
91
+ "epoch": 0.7051282051282052,
92
+ "grad_norm": 2.5044240951538086,
93
+ "learning_rate": 0.00018915187376725837,
94
+ "loss": 0.8484,
95
+ "step": 110
96
+ },
97
+ {
98
+ "epoch": 0.7692307692307693,
99
+ "grad_norm": 2.8144619464874268,
100
+ "learning_rate": 0.00018816568047337278,
101
+ "loss": 0.8283,
102
+ "step": 120
103
+ },
104
+ {
105
+ "epoch": 0.8333333333333334,
106
+ "grad_norm": 1.8459696769714355,
107
+ "learning_rate": 0.0001871794871794872,
108
+ "loss": 0.8201,
109
+ "step": 130
110
+ },
111
+ {
112
+ "epoch": 0.8974358974358975,
113
+ "grad_norm": 3.3761656284332275,
114
+ "learning_rate": 0.0001861932938856016,
115
+ "loss": 0.7736,
116
+ "step": 140
117
+ },
118
+ {
119
+ "epoch": 0.9615384615384616,
120
+ "grad_norm": 2.285553216934204,
121
+ "learning_rate": 0.00018520710059171598,
122
+ "loss": 0.7505,
123
+ "step": 150
124
+ },
125
+ {
126
+ "epoch": 1.0256410256410255,
127
+ "grad_norm": 3.3042893409729004,
128
+ "learning_rate": 0.0001842209072978304,
129
+ "loss": 0.6614,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 1.0897435897435896,
134
+ "grad_norm": 3.174131393432617,
135
+ "learning_rate": 0.00018323471400394478,
136
+ "loss": 0.7393,
137
+ "step": 170
138
+ },
139
+ {
140
+ "epoch": 1.1538461538461537,
141
+ "grad_norm": 2.8764002323150635,
142
+ "learning_rate": 0.00018224852071005917,
143
+ "loss": 0.892,
144
+ "step": 180
145
+ },
146
+ {
147
+ "epoch": 1.217948717948718,
148
+ "grad_norm": 3.123760223388672,
149
+ "learning_rate": 0.00018126232741617356,
150
+ "loss": 0.7055,
151
+ "step": 190
152
+ },
153
+ {
154
+ "epoch": 1.282051282051282,
155
+ "grad_norm": 2.54199481010437,
156
+ "learning_rate": 0.00018027613412228798,
157
+ "loss": 0.5613,
158
+ "step": 200
159
+ },
160
+ {
161
+ "epoch": 1.282051282051282,
162
+ "eval_accuracy": 0.6618705035971223,
163
+ "eval_loss": 0.7067616581916809,
164
+ "eval_runtime": 2.5932,
165
+ "eval_samples_per_second": 107.202,
166
+ "eval_steps_per_second": 13.497,
167
+ "step": 200
168
+ },
169
+ {
170
+ "epoch": 1.3461538461538463,
171
+ "grad_norm": 4.848214626312256,
172
+ "learning_rate": 0.0001792899408284024,
173
+ "loss": 0.634,
174
+ "step": 210
175
+ },
176
+ {
177
+ "epoch": 1.4102564102564101,
178
+ "grad_norm": 2.621889352798462,
179
+ "learning_rate": 0.00017830374753451678,
180
+ "loss": 0.6666,
181
+ "step": 220
182
+ },
183
+ {
184
+ "epoch": 1.4743589743589745,
185
+ "grad_norm": 3.7964868545532227,
186
+ "learning_rate": 0.00017731755424063117,
187
+ "loss": 0.5344,
188
+ "step": 230
189
+ },
190
+ {
191
+ "epoch": 1.5384615384615383,
192
+ "grad_norm": 3.703036308288574,
193
+ "learning_rate": 0.00017633136094674556,
194
+ "loss": 0.5636,
195
+ "step": 240
196
+ },
197
+ {
198
+ "epoch": 1.6025641025641026,
199
+ "grad_norm": 3.081791400909424,
200
+ "learning_rate": 0.00017534516765285997,
201
+ "loss": 0.4211,
202
+ "step": 250
203
+ },
204
+ {
205
+ "epoch": 1.6666666666666665,
206
+ "grad_norm": 3.167443037033081,
207
+ "learning_rate": 0.00017435897435897436,
208
+ "loss": 0.6438,
209
+ "step": 260
210
+ },
211
+ {
212
+ "epoch": 1.7307692307692308,
213
+ "grad_norm": 6.15064811706543,
214
+ "learning_rate": 0.00017337278106508875,
215
+ "loss": 0.5872,
216
+ "step": 270
217
+ },
218
+ {
219
+ "epoch": 1.7948717948717947,
220
+ "grad_norm": 2.0773203372955322,
221
+ "learning_rate": 0.00017238658777120317,
222
+ "loss": 0.6172,
223
+ "step": 280
224
+ },
225
+ {
226
+ "epoch": 1.858974358974359,
227
+ "grad_norm": 3.2962491512298584,
228
+ "learning_rate": 0.00017140039447731758,
229
+ "loss": 0.5252,
230
+ "step": 290
231
+ },
232
+ {
233
+ "epoch": 1.9230769230769231,
234
+ "grad_norm": 2.557178020477295,
235
+ "learning_rate": 0.00017041420118343197,
236
+ "loss": 0.4296,
237
+ "step": 300
238
+ },
239
+ {
240
+ "epoch": 1.9230769230769231,
241
+ "eval_accuracy": 0.8309352517985612,
242
+ "eval_loss": 0.4007588028907776,
243
+ "eval_runtime": 2.5685,
244
+ "eval_samples_per_second": 108.233,
245
+ "eval_steps_per_second": 13.626,
246
+ "step": 300
247
+ },
248
+ {
249
+ "epoch": 1.9871794871794872,
250
+ "grad_norm": 2.061478614807129,
251
+ "learning_rate": 0.00016942800788954636,
252
+ "loss": 0.4041,
253
+ "step": 310
254
+ },
255
+ {
256
+ "epoch": 2.051282051282051,
257
+ "grad_norm": 2.703629732131958,
258
+ "learning_rate": 0.00016844181459566075,
259
+ "loss": 0.3791,
260
+ "step": 320
261
+ },
262
+ {
263
+ "epoch": 2.1153846153846154,
264
+ "grad_norm": 0.8648492693901062,
265
+ "learning_rate": 0.00016745562130177514,
266
+ "loss": 0.2943,
267
+ "step": 330
268
+ },
269
+ {
270
+ "epoch": 2.1794871794871793,
271
+ "grad_norm": 4.140072345733643,
272
+ "learning_rate": 0.00016646942800788956,
273
+ "loss": 0.2485,
274
+ "step": 340
275
+ },
276
+ {
277
+ "epoch": 2.2435897435897436,
278
+ "grad_norm": 5.02016019821167,
279
+ "learning_rate": 0.00016548323471400394,
280
+ "loss": 0.4904,
281
+ "step": 350
282
+ },
283
+ {
284
+ "epoch": 2.3076923076923075,
285
+ "grad_norm": 1.8560619354248047,
286
+ "learning_rate": 0.00016449704142011836,
287
+ "loss": 0.3835,
288
+ "step": 360
289
+ },
290
+ {
291
+ "epoch": 2.371794871794872,
292
+ "grad_norm": 10.23709487915039,
293
+ "learning_rate": 0.00016351084812623275,
294
+ "loss": 0.3015,
295
+ "step": 370
296
+ },
297
+ {
298
+ "epoch": 2.435897435897436,
299
+ "grad_norm": 5.944732666015625,
300
+ "learning_rate": 0.00016252465483234716,
301
+ "loss": 0.3199,
302
+ "step": 380
303
+ },
304
+ {
305
+ "epoch": 2.5,
306
+ "grad_norm": 2.0510830879211426,
307
+ "learning_rate": 0.00016153846153846155,
308
+ "loss": 0.2512,
309
+ "step": 390
310
+ },
311
+ {
312
+ "epoch": 2.564102564102564,
313
+ "grad_norm": 3.4166271686553955,
314
+ "learning_rate": 0.00016055226824457594,
315
+ "loss": 0.3475,
316
+ "step": 400
317
+ },
318
+ {
319
+ "epoch": 2.564102564102564,
320
+ "eval_accuracy": 0.8812949640287769,
321
+ "eval_loss": 0.33450016379356384,
322
+ "eval_runtime": 3.6169,
323
+ "eval_samples_per_second": 76.862,
324
+ "eval_steps_per_second": 9.677,
325
+ "step": 400
326
+ },
327
+ {
328
+ "epoch": 2.628205128205128,
329
+ "grad_norm": 6.59075927734375,
330
+ "learning_rate": 0.00015956607495069033,
331
+ "loss": 0.2296,
332
+ "step": 410
333
+ },
334
+ {
335
+ "epoch": 2.6923076923076925,
336
+ "grad_norm": 2.345207929611206,
337
+ "learning_rate": 0.00015857988165680475,
338
+ "loss": 0.2535,
339
+ "step": 420
340
+ },
341
+ {
342
+ "epoch": 2.7564102564102564,
343
+ "grad_norm": 3.3290200233459473,
344
+ "learning_rate": 0.00015759368836291914,
345
+ "loss": 0.2902,
346
+ "step": 430
347
+ },
348
+ {
349
+ "epoch": 2.8205128205128203,
350
+ "grad_norm": 3.723925828933716,
351
+ "learning_rate": 0.00015660749506903355,
352
+ "loss": 0.2591,
353
+ "step": 440
354
+ },
355
+ {
356
+ "epoch": 2.8846153846153846,
357
+ "grad_norm": 4.227429389953613,
358
+ "learning_rate": 0.00015562130177514794,
359
+ "loss": 0.2915,
360
+ "step": 450
361
+ },
362
+ {
363
+ "epoch": 2.948717948717949,
364
+ "grad_norm": 3.918539047241211,
365
+ "learning_rate": 0.00015463510848126233,
366
+ "loss": 0.3859,
367
+ "step": 460
368
+ },
369
+ {
370
+ "epoch": 3.0128205128205128,
371
+ "grad_norm": 2.824794292449951,
372
+ "learning_rate": 0.00015364891518737675,
373
+ "loss": 0.3425,
374
+ "step": 470
375
+ },
376
+ {
377
+ "epoch": 3.076923076923077,
378
+ "grad_norm": 0.28933683037757874,
379
+ "learning_rate": 0.00015266272189349113,
380
+ "loss": 0.1796,
381
+ "step": 480
382
+ },
383
+ {
384
+ "epoch": 3.141025641025641,
385
+ "grad_norm": 2.101590394973755,
386
+ "learning_rate": 0.00015167652859960552,
387
+ "loss": 0.1881,
388
+ "step": 490
389
+ },
390
+ {
391
+ "epoch": 3.2051282051282053,
392
+ "grad_norm": 0.31414705514907837,
393
+ "learning_rate": 0.0001506903353057199,
394
+ "loss": 0.1183,
395
+ "step": 500
396
+ },
397
+ {
398
+ "epoch": 3.2051282051282053,
399
+ "eval_accuracy": 0.8489208633093526,
400
+ "eval_loss": 0.4293219745159149,
401
+ "eval_runtime": 3.4489,
402
+ "eval_samples_per_second": 80.605,
403
+ "eval_steps_per_second": 10.148,
404
+ "step": 500
405
+ },
406
+ {
407
+ "epoch": 3.269230769230769,
408
+ "grad_norm": 3.8199613094329834,
409
+ "learning_rate": 0.00014970414201183433,
410
+ "loss": 0.1736,
411
+ "step": 510
412
+ },
413
+ {
414
+ "epoch": 3.3333333333333335,
415
+ "grad_norm": 1.631716251373291,
416
+ "learning_rate": 0.00014871794871794872,
417
+ "loss": 0.1903,
418
+ "step": 520
419
+ },
420
+ {
421
+ "epoch": 3.3974358974358974,
422
+ "grad_norm": 3.813403844833374,
423
+ "learning_rate": 0.00014773175542406313,
424
+ "loss": 0.2243,
425
+ "step": 530
426
+ },
427
+ {
428
+ "epoch": 3.4615384615384617,
429
+ "grad_norm": 6.256704330444336,
430
+ "learning_rate": 0.00014674556213017752,
431
+ "loss": 0.1941,
432
+ "step": 540
433
+ },
434
+ {
435
+ "epoch": 3.5256410256410255,
436
+ "grad_norm": 6.570198059082031,
437
+ "learning_rate": 0.0001457593688362919,
438
+ "loss": 0.1747,
439
+ "step": 550
440
+ },
441
+ {
442
+ "epoch": 3.58974358974359,
443
+ "grad_norm": 3.7718803882598877,
444
+ "learning_rate": 0.00014477317554240633,
445
+ "loss": 0.2871,
446
+ "step": 560
447
+ },
448
+ {
449
+ "epoch": 3.6538461538461537,
450
+ "grad_norm": 3.889697790145874,
451
+ "learning_rate": 0.00014378698224852072,
452
+ "loss": 0.173,
453
+ "step": 570
454
+ },
455
+ {
456
+ "epoch": 3.717948717948718,
457
+ "grad_norm": 2.5270767211914062,
458
+ "learning_rate": 0.0001428007889546351,
459
+ "loss": 0.2472,
460
+ "step": 580
461
+ },
462
+ {
463
+ "epoch": 3.782051282051282,
464
+ "grad_norm": 0.2746649980545044,
465
+ "learning_rate": 0.0001418145956607495,
466
+ "loss": 0.1564,
467
+ "step": 590
468
+ },
469
+ {
470
+ "epoch": 3.8461538461538463,
471
+ "grad_norm": 2.9636824131011963,
472
+ "learning_rate": 0.0001408284023668639,
473
+ "loss": 0.1531,
474
+ "step": 600
475
+ },
476
+ {
477
+ "epoch": 3.8461538461538463,
478
+ "eval_accuracy": 0.9136690647482014,
479
+ "eval_loss": 0.2748344838619232,
480
+ "eval_runtime": 2.583,
481
+ "eval_samples_per_second": 107.625,
482
+ "eval_steps_per_second": 13.55,
483
+ "step": 600
484
+ },
485
+ {
486
+ "epoch": 3.91025641025641,
487
+ "grad_norm": 2.8231372833251953,
488
+ "learning_rate": 0.00013984220907297832,
489
+ "loss": 0.1651,
490
+ "step": 610
491
+ },
492
+ {
493
+ "epoch": 3.9743589743589745,
494
+ "grad_norm": 0.6206132769584656,
495
+ "learning_rate": 0.0001388560157790927,
496
+ "loss": 0.1854,
497
+ "step": 620
498
+ },
499
+ {
500
+ "epoch": 4.038461538461538,
501
+ "grad_norm": 0.720039427280426,
502
+ "learning_rate": 0.0001378698224852071,
503
+ "loss": 0.2629,
504
+ "step": 630
505
+ },
506
+ {
507
+ "epoch": 4.102564102564102,
508
+ "grad_norm": 0.28592395782470703,
509
+ "learning_rate": 0.00013688362919132152,
510
+ "loss": 0.0969,
511
+ "step": 640
512
+ },
513
+ {
514
+ "epoch": 4.166666666666667,
515
+ "grad_norm": 1.0571306943893433,
516
+ "learning_rate": 0.0001358974358974359,
517
+ "loss": 0.0816,
518
+ "step": 650
519
+ },
520
+ {
521
+ "epoch": 4.230769230769231,
522
+ "grad_norm": 5.520449638366699,
523
+ "learning_rate": 0.0001349112426035503,
524
+ "loss": 0.0997,
525
+ "step": 660
526
+ },
527
+ {
528
+ "epoch": 4.294871794871795,
529
+ "grad_norm": 10.63196849822998,
530
+ "learning_rate": 0.00013392504930966468,
531
+ "loss": 0.1369,
532
+ "step": 670
533
+ },
534
+ {
535
+ "epoch": 4.358974358974359,
536
+ "grad_norm": 2.723325729370117,
537
+ "learning_rate": 0.0001329388560157791,
538
+ "loss": 0.0631,
539
+ "step": 680
540
+ },
541
+ {
542
+ "epoch": 4.423076923076923,
543
+ "grad_norm": 0.07080604135990143,
544
+ "learning_rate": 0.00013195266272189352,
545
+ "loss": 0.1089,
546
+ "step": 690
547
+ },
548
+ {
549
+ "epoch": 4.487179487179487,
550
+ "grad_norm": 5.3309326171875,
551
+ "learning_rate": 0.0001309664694280079,
552
+ "loss": 0.1174,
553
+ "step": 700
554
+ },
555
+ {
556
+ "epoch": 4.487179487179487,
557
+ "eval_accuracy": 0.8812949640287769,
558
+ "eval_loss": 0.36486199498176575,
559
+ "eval_runtime": 3.5449,
560
+ "eval_samples_per_second": 78.422,
561
+ "eval_steps_per_second": 9.873,
562
+ "step": 700
563
+ },
564
+ {
565
+ "epoch": 4.551282051282051,
566
+ "grad_norm": 4.462215900421143,
567
+ "learning_rate": 0.0001299802761341223,
568
+ "loss": 0.1378,
569
+ "step": 710
570
+ },
571
+ {
572
+ "epoch": 4.615384615384615,
573
+ "grad_norm": 8.144935607910156,
574
+ "learning_rate": 0.00012899408284023668,
575
+ "loss": 0.0917,
576
+ "step": 720
577
+ },
578
+ {
579
+ "epoch": 4.67948717948718,
580
+ "grad_norm": 0.08221765607595444,
581
+ "learning_rate": 0.0001280078895463511,
582
+ "loss": 0.0562,
583
+ "step": 730
584
+ },
585
+ {
586
+ "epoch": 4.743589743589744,
587
+ "grad_norm": 8.699402809143066,
588
+ "learning_rate": 0.0001270216962524655,
589
+ "loss": 0.1787,
590
+ "step": 740
591
+ },
592
+ {
593
+ "epoch": 4.8076923076923075,
594
+ "grad_norm": 7.376129150390625,
595
+ "learning_rate": 0.00012603550295857988,
596
+ "loss": 0.064,
597
+ "step": 750
598
+ },
599
+ {
600
+ "epoch": 4.871794871794872,
601
+ "grad_norm": 0.08259747177362442,
602
+ "learning_rate": 0.0001250493096646943,
603
+ "loss": 0.0927,
604
+ "step": 760
605
+ },
606
+ {
607
+ "epoch": 4.935897435897436,
608
+ "grad_norm": 0.08165480941534042,
609
+ "learning_rate": 0.00012406311637080868,
610
+ "loss": 0.1067,
611
+ "step": 770
612
+ },
613
+ {
614
+ "epoch": 5.0,
615
+ "grad_norm": 0.19699496030807495,
616
+ "learning_rate": 0.0001230769230769231,
617
+ "loss": 0.0974,
618
+ "step": 780
619
+ },
620
+ {
621
+ "epoch": 5.064102564102564,
622
+ "grad_norm": 3.4025802612304688,
623
+ "learning_rate": 0.00012209072978303749,
624
+ "loss": 0.0977,
625
+ "step": 790
626
+ },
627
+ {
628
+ "epoch": 5.128205128205128,
629
+ "grad_norm": 2.273251533508301,
630
+ "learning_rate": 0.00012110453648915188,
631
+ "loss": 0.0498,
632
+ "step": 800
633
+ },
634
+ {
635
+ "epoch": 5.128205128205128,
636
+ "eval_accuracy": 0.8920863309352518,
637
+ "eval_loss": 0.327900767326355,
638
+ "eval_runtime": 3.5689,
639
+ "eval_samples_per_second": 77.895,
640
+ "eval_steps_per_second": 9.807,
641
+ "step": 800
642
+ },
643
+ {
644
+ "epoch": 5.1923076923076925,
645
+ "grad_norm": 0.624547004699707,
646
+ "learning_rate": 0.00012011834319526626,
647
+ "loss": 0.0821,
648
+ "step": 810
649
+ },
650
+ {
651
+ "epoch": 5.256410256410256,
652
+ "grad_norm": 3.7610857486724854,
653
+ "learning_rate": 0.00011913214990138068,
654
+ "loss": 0.0481,
655
+ "step": 820
656
+ },
657
+ {
658
+ "epoch": 5.32051282051282,
659
+ "grad_norm": 0.06995019316673279,
660
+ "learning_rate": 0.00011814595660749508,
661
+ "loss": 0.0117,
662
+ "step": 830
663
+ },
664
+ {
665
+ "epoch": 5.384615384615385,
666
+ "grad_norm": 0.08295288681983948,
667
+ "learning_rate": 0.00011715976331360947,
668
+ "loss": 0.0122,
669
+ "step": 840
670
+ },
671
+ {
672
+ "epoch": 5.448717948717949,
673
+ "grad_norm": 0.0509318970143795,
674
+ "learning_rate": 0.00011617357001972386,
675
+ "loss": 0.0447,
676
+ "step": 850
677
+ },
678
+ {
679
+ "epoch": 5.512820512820513,
680
+ "grad_norm": 0.05534667894244194,
681
+ "learning_rate": 0.00011518737672583828,
682
+ "loss": 0.0786,
683
+ "step": 860
684
+ },
685
+ {
686
+ "epoch": 5.576923076923077,
687
+ "grad_norm": 0.36779171228408813,
688
+ "learning_rate": 0.00011420118343195268,
689
+ "loss": 0.0427,
690
+ "step": 870
691
+ },
692
+ {
693
+ "epoch": 5.641025641025641,
694
+ "grad_norm": 0.0412164181470871,
695
+ "learning_rate": 0.00011321499013806707,
696
+ "loss": 0.0151,
697
+ "step": 880
698
+ },
699
+ {
700
+ "epoch": 5.705128205128205,
701
+ "grad_norm": 0.03488855063915253,
702
+ "learning_rate": 0.00011222879684418146,
703
+ "loss": 0.0331,
704
+ "step": 890
705
+ },
706
+ {
707
+ "epoch": 5.769230769230769,
708
+ "grad_norm": 5.378381252288818,
709
+ "learning_rate": 0.00011124260355029586,
710
+ "loss": 0.0817,
711
+ "step": 900
712
+ },
713
+ {
714
+ "epoch": 5.769230769230769,
715
+ "eval_accuracy": 0.935251798561151,
716
+ "eval_loss": 0.27631285786628723,
717
+ "eval_runtime": 3.5274,
718
+ "eval_samples_per_second": 78.812,
719
+ "eval_steps_per_second": 9.922,
720
+ "step": 900
721
+ },
722
+ {
723
+ "epoch": 5.833333333333333,
724
+ "grad_norm": 0.121369369328022,
725
+ "learning_rate": 0.00011025641025641027,
726
+ "loss": 0.0164,
727
+ "step": 910
728
+ },
729
+ {
730
+ "epoch": 5.897435897435898,
731
+ "grad_norm": 0.30134713649749756,
732
+ "learning_rate": 0.00010927021696252466,
733
+ "loss": 0.0399,
734
+ "step": 920
735
+ },
736
+ {
737
+ "epoch": 5.961538461538462,
738
+ "grad_norm": 9.133658409118652,
739
+ "learning_rate": 0.00010828402366863905,
740
+ "loss": 0.0893,
741
+ "step": 930
742
+ },
743
+ {
744
+ "epoch": 6.0256410256410255,
745
+ "grad_norm": 0.037670884281396866,
746
+ "learning_rate": 0.00010729783037475345,
747
+ "loss": 0.0285,
748
+ "step": 940
749
+ },
750
+ {
751
+ "epoch": 6.089743589743589,
752
+ "grad_norm": 4.902493953704834,
753
+ "learning_rate": 0.00010631163708086787,
754
+ "loss": 0.0796,
755
+ "step": 950
756
+ },
757
+ {
758
+ "epoch": 6.153846153846154,
759
+ "grad_norm": 6.080508232116699,
760
+ "learning_rate": 0.00010532544378698226,
761
+ "loss": 0.1406,
762
+ "step": 960
763
+ },
764
+ {
765
+ "epoch": 6.217948717948718,
766
+ "grad_norm": 3.923866033554077,
767
+ "learning_rate": 0.00010433925049309665,
768
+ "loss": 0.0914,
769
+ "step": 970
770
+ },
771
+ {
772
+ "epoch": 6.282051282051282,
773
+ "grad_norm": 0.9966168999671936,
774
+ "learning_rate": 0.00010335305719921105,
775
+ "loss": 0.0906,
776
+ "step": 980
777
+ },
778
+ {
779
+ "epoch": 6.346153846153846,
780
+ "grad_norm": 0.2861874997615814,
781
+ "learning_rate": 0.00010236686390532544,
782
+ "loss": 0.0335,
783
+ "step": 990
784
+ },
785
+ {
786
+ "epoch": 6.410256410256411,
787
+ "grad_norm": 0.030173303559422493,
788
+ "learning_rate": 0.00010138067061143986,
789
+ "loss": 0.0075,
790
+ "step": 1000
791
+ },
792
+ {
793
+ "epoch": 6.410256410256411,
794
+ "eval_accuracy": 0.920863309352518,
795
+ "eval_loss": 0.26709362864494324,
796
+ "eval_runtime": 2.5656,
797
+ "eval_samples_per_second": 108.358,
798
+ "eval_steps_per_second": 13.642,
799
+ "step": 1000
800
+ },
801
+ {
802
+ "epoch": 6.4743589743589745,
803
+ "grad_norm": 0.02864202857017517,
804
+ "learning_rate": 0.00010039447731755424,
805
+ "loss": 0.0628,
806
+ "step": 1010
807
+ },
808
+ {
809
+ "epoch": 6.538461538461538,
810
+ "grad_norm": 0.6442215442657471,
811
+ "learning_rate": 9.940828402366865e-05,
812
+ "loss": 0.0205,
813
+ "step": 1020
814
+ },
815
+ {
816
+ "epoch": 6.602564102564102,
817
+ "grad_norm": 0.047572050243616104,
818
+ "learning_rate": 9.842209072978305e-05,
819
+ "loss": 0.1029,
820
+ "step": 1030
821
+ },
822
+ {
823
+ "epoch": 6.666666666666667,
824
+ "grad_norm": 11.545117378234863,
825
+ "learning_rate": 9.743589743589744e-05,
826
+ "loss": 0.0454,
827
+ "step": 1040
828
+ },
829
+ {
830
+ "epoch": 6.730769230769231,
831
+ "grad_norm": 0.08284212648868561,
832
+ "learning_rate": 9.644970414201184e-05,
833
+ "loss": 0.0059,
834
+ "step": 1050
835
+ },
836
+ {
837
+ "epoch": 6.794871794871795,
838
+ "grad_norm": 0.19572466611862183,
839
+ "learning_rate": 9.546351084812624e-05,
840
+ "loss": 0.0907,
841
+ "step": 1060
842
+ },
843
+ {
844
+ "epoch": 6.858974358974359,
845
+ "grad_norm": 0.041858524084091187,
846
+ "learning_rate": 9.447731755424064e-05,
847
+ "loss": 0.0562,
848
+ "step": 1070
849
+ },
850
+ {
851
+ "epoch": 6.923076923076923,
852
+ "grad_norm": 0.045817919075489044,
853
+ "learning_rate": 9.349112426035503e-05,
854
+ "loss": 0.0262,
855
+ "step": 1080
856
+ },
857
+ {
858
+ "epoch": 6.987179487179487,
859
+ "grad_norm": 0.14285103976726532,
860
+ "learning_rate": 9.250493096646942e-05,
861
+ "loss": 0.0185,
862
+ "step": 1090
863
+ },
864
+ {
865
+ "epoch": 7.051282051282051,
866
+ "grad_norm": 0.027153167873620987,
867
+ "learning_rate": 9.151873767258384e-05,
868
+ "loss": 0.0265,
869
+ "step": 1100
870
+ },
871
+ {
872
+ "epoch": 7.051282051282051,
873
+ "eval_accuracy": 0.920863309352518,
874
+ "eval_loss": 0.3184675872325897,
875
+ "eval_runtime": 2.4994,
876
+ "eval_samples_per_second": 111.228,
877
+ "eval_steps_per_second": 14.004,
878
+ "step": 1100
879
+ },
880
+ {
881
+ "epoch": 7.115384615384615,
882
+ "grad_norm": 0.027239225804805756,
883
+ "learning_rate": 9.053254437869823e-05,
884
+ "loss": 0.0629,
885
+ "step": 1110
886
+ },
887
+ {
888
+ "epoch": 7.17948717948718,
889
+ "grad_norm": 0.02252044342458248,
890
+ "learning_rate": 8.954635108481263e-05,
891
+ "loss": 0.0198,
892
+ "step": 1120
893
+ },
894
+ {
895
+ "epoch": 7.243589743589744,
896
+ "grad_norm": 0.038163017481565475,
897
+ "learning_rate": 8.856015779092702e-05,
898
+ "loss": 0.0323,
899
+ "step": 1130
900
+ },
901
+ {
902
+ "epoch": 7.3076923076923075,
903
+ "grad_norm": 0.038108520209789276,
904
+ "learning_rate": 8.757396449704143e-05,
905
+ "loss": 0.0379,
906
+ "step": 1140
907
+ },
908
+ {
909
+ "epoch": 7.371794871794872,
910
+ "grad_norm": 0.019251950085163116,
911
+ "learning_rate": 8.658777120315582e-05,
912
+ "loss": 0.028,
913
+ "step": 1150
914
+ },
915
+ {
916
+ "epoch": 7.435897435897436,
917
+ "grad_norm": 0.01974482275545597,
918
+ "learning_rate": 8.560157790927023e-05,
919
+ "loss": 0.0043,
920
+ "step": 1160
921
+ },
922
+ {
923
+ "epoch": 7.5,
924
+ "grad_norm": 2.0657660961151123,
925
+ "learning_rate": 8.461538461538461e-05,
926
+ "loss": 0.0069,
927
+ "step": 1170
928
+ },
929
+ {
930
+ "epoch": 7.564102564102564,
931
+ "grad_norm": 0.025256551802158356,
932
+ "learning_rate": 8.362919132149902e-05,
933
+ "loss": 0.0045,
934
+ "step": 1180
935
+ },
936
+ {
937
+ "epoch": 7.628205128205128,
938
+ "grad_norm": 0.018220046535134315,
939
+ "learning_rate": 8.264299802761342e-05,
940
+ "loss": 0.0038,
941
+ "step": 1190
942
+ },
943
+ {
944
+ "epoch": 7.6923076923076925,
945
+ "grad_norm": 15.359424591064453,
946
+ "learning_rate": 8.165680473372781e-05,
947
+ "loss": 0.0457,
948
+ "step": 1200
949
+ },
950
+ {
951
+ "epoch": 7.6923076923076925,
952
+ "eval_accuracy": 0.9100719424460432,
953
+ "eval_loss": 0.37759149074554443,
954
+ "eval_runtime": 3.7105,
955
+ "eval_samples_per_second": 74.922,
956
+ "eval_steps_per_second": 9.433,
957
+ "step": 1200
958
+ },
959
+ {
960
+ "epoch": 7.756410256410256,
961
+ "grad_norm": 0.017574544996023178,
962
+ "learning_rate": 8.067061143984221e-05,
963
+ "loss": 0.0038,
964
+ "step": 1210
965
+ },
966
+ {
967
+ "epoch": 7.82051282051282,
968
+ "grad_norm": 0.01818062551319599,
969
+ "learning_rate": 7.968441814595661e-05,
970
+ "loss": 0.004,
971
+ "step": 1220
972
+ },
973
+ {
974
+ "epoch": 7.884615384615385,
975
+ "grad_norm": 0.024309920147061348,
976
+ "learning_rate": 7.869822485207101e-05,
977
+ "loss": 0.0038,
978
+ "step": 1230
979
+ },
980
+ {
981
+ "epoch": 7.948717948717949,
982
+ "grad_norm": 0.07784882932901382,
983
+ "learning_rate": 7.77120315581854e-05,
984
+ "loss": 0.0059,
985
+ "step": 1240
986
+ },
987
+ {
988
+ "epoch": 8.012820512820513,
989
+ "grad_norm": 0.035918813198804855,
990
+ "learning_rate": 7.67258382642998e-05,
991
+ "loss": 0.0047,
992
+ "step": 1250
993
+ },
994
+ {
995
+ "epoch": 8.076923076923077,
996
+ "grad_norm": 0.35960128903388977,
997
+ "learning_rate": 7.573964497041421e-05,
998
+ "loss": 0.0068,
999
+ "step": 1260
1000
+ },
1001
+ {
1002
+ "epoch": 8.14102564102564,
1003
+ "grad_norm": 0.015659425407648087,
1004
+ "learning_rate": 7.475345167652861e-05,
1005
+ "loss": 0.0039,
1006
+ "step": 1270
1007
+ },
1008
+ {
1009
+ "epoch": 8.205128205128204,
1010
+ "grad_norm": 0.01647733524441719,
1011
+ "learning_rate": 7.3767258382643e-05,
1012
+ "loss": 0.0035,
1013
+ "step": 1280
1014
+ },
1015
+ {
1016
+ "epoch": 8.26923076923077,
1017
+ "grad_norm": 0.015693532302975655,
1018
+ "learning_rate": 7.27810650887574e-05,
1019
+ "loss": 0.0203,
1020
+ "step": 1290
1021
+ },
1022
+ {
1023
+ "epoch": 8.333333333333334,
1024
+ "grad_norm": 0.01847326196730137,
1025
+ "learning_rate": 7.17948717948718e-05,
1026
+ "loss": 0.0032,
1027
+ "step": 1300
1028
+ },
1029
+ {
1030
+ "epoch": 8.333333333333334,
1031
+ "eval_accuracy": 0.9388489208633094,
1032
+ "eval_loss": 0.2834855318069458,
1033
+ "eval_runtime": 3.4846,
1034
+ "eval_samples_per_second": 79.78,
1035
+ "eval_steps_per_second": 10.044,
1036
+ "step": 1300
1037
+ },
1038
+ {
1039
+ "epoch": 8.397435897435898,
1040
+ "grad_norm": 0.01613900437951088,
1041
+ "learning_rate": 7.08086785009862e-05,
1042
+ "loss": 0.0055,
1043
+ "step": 1310
1044
+ },
1045
+ {
1046
+ "epoch": 8.461538461538462,
1047
+ "grad_norm": 0.015385860577225685,
1048
+ "learning_rate": 6.98224852071006e-05,
1049
+ "loss": 0.0118,
1050
+ "step": 1320
1051
+ },
1052
+ {
1053
+ "epoch": 8.525641025641026,
1054
+ "grad_norm": 0.014069377444684505,
1055
+ "learning_rate": 6.883629191321498e-05,
1056
+ "loss": 0.0029,
1057
+ "step": 1330
1058
+ },
1059
+ {
1060
+ "epoch": 8.58974358974359,
1061
+ "grad_norm": 0.015751365572214127,
1062
+ "learning_rate": 6.78500986193294e-05,
1063
+ "loss": 0.0116,
1064
+ "step": 1340
1065
+ },
1066
+ {
1067
+ "epoch": 8.653846153846153,
1068
+ "grad_norm": 0.0143636055290699,
1069
+ "learning_rate": 6.686390532544379e-05,
1070
+ "loss": 0.0554,
1071
+ "step": 1350
1072
+ },
1073
+ {
1074
+ "epoch": 8.717948717948717,
1075
+ "grad_norm": 0.015580832026898861,
1076
+ "learning_rate": 6.587771203155819e-05,
1077
+ "loss": 0.0031,
1078
+ "step": 1360
1079
+ },
1080
+ {
1081
+ "epoch": 8.782051282051283,
1082
+ "grad_norm": 0.014274963177740574,
1083
+ "learning_rate": 6.489151873767258e-05,
1084
+ "loss": 0.0029,
1085
+ "step": 1370
1086
+ },
1087
+ {
1088
+ "epoch": 8.846153846153847,
1089
+ "grad_norm": 0.02313074842095375,
1090
+ "learning_rate": 6.390532544378698e-05,
1091
+ "loss": 0.0339,
1092
+ "step": 1380
1093
+ },
1094
+ {
1095
+ "epoch": 8.91025641025641,
1096
+ "grad_norm": 1.4907852411270142,
1097
+ "learning_rate": 6.291913214990139e-05,
1098
+ "loss": 0.0046,
1099
+ "step": 1390
1100
+ },
1101
+ {
1102
+ "epoch": 8.974358974358974,
1103
+ "grad_norm": 0.015585100278258324,
1104
+ "learning_rate": 6.193293885601579e-05,
1105
+ "loss": 0.0027,
1106
+ "step": 1400
1107
+ },
1108
+ {
1109
+ "epoch": 8.974358974358974,
1110
+ "eval_accuracy": 0.8884892086330936,
1111
+ "eval_loss": 0.5365095734596252,
1112
+ "eval_runtime": 2.5385,
1113
+ "eval_samples_per_second": 109.513,
1114
+ "eval_steps_per_second": 13.788,
1115
+ "step": 1400
1116
+ },
1117
+ {
1118
+ "epoch": 9.038461538461538,
1119
+ "grad_norm": 0.014531627297401428,
1120
+ "learning_rate": 6.094674556213018e-05,
1121
+ "loss": 0.0253,
1122
+ "step": 1410
1123
+ },
1124
+ {
1125
+ "epoch": 9.102564102564102,
1126
+ "grad_norm": 0.012606761418282986,
1127
+ "learning_rate": 5.996055226824457e-05,
1128
+ "loss": 0.0031,
1129
+ "step": 1420
1130
+ },
1131
+ {
1132
+ "epoch": 9.166666666666666,
1133
+ "grad_norm": 0.012980329804122448,
1134
+ "learning_rate": 5.897435897435898e-05,
1135
+ "loss": 0.003,
1136
+ "step": 1430
1137
+ },
1138
+ {
1139
+ "epoch": 9.23076923076923,
1140
+ "grad_norm": 0.012560625560581684,
1141
+ "learning_rate": 5.798816568047337e-05,
1142
+ "loss": 0.0029,
1143
+ "step": 1440
1144
+ },
1145
+ {
1146
+ "epoch": 9.294871794871796,
1147
+ "grad_norm": 0.01267662551254034,
1148
+ "learning_rate": 5.700197238658778e-05,
1149
+ "loss": 0.0027,
1150
+ "step": 1450
1151
+ },
1152
+ {
1153
+ "epoch": 9.35897435897436,
1154
+ "grad_norm": 0.018838873133063316,
1155
+ "learning_rate": 5.601577909270217e-05,
1156
+ "loss": 0.0026,
1157
+ "step": 1460
1158
+ },
1159
+ {
1160
+ "epoch": 9.423076923076923,
1161
+ "grad_norm": 0.012279819697141647,
1162
+ "learning_rate": 5.502958579881658e-05,
1163
+ "loss": 0.0025,
1164
+ "step": 1470
1165
+ },
1166
+ {
1167
+ "epoch": 9.487179487179487,
1168
+ "grad_norm": 0.011527939699590206,
1169
+ "learning_rate": 5.4043392504930966e-05,
1170
+ "loss": 0.0023,
1171
+ "step": 1480
1172
+ },
1173
+ {
1174
+ "epoch": 9.551282051282051,
1175
+ "grad_norm": 0.01282353326678276,
1176
+ "learning_rate": 5.3057199211045375e-05,
1177
+ "loss": 0.0024,
1178
+ "step": 1490
1179
+ },
1180
+ {
1181
+ "epoch": 9.615384615384615,
1182
+ "grad_norm": 0.012979278340935707,
1183
+ "learning_rate": 5.2071005917159764e-05,
1184
+ "loss": 0.0024,
1185
+ "step": 1500
1186
+ },
1187
+ {
1188
+ "epoch": 9.615384615384615,
1189
+ "eval_accuracy": 0.9460431654676259,
1190
+ "eval_loss": 0.28174683451652527,
1191
+ "eval_runtime": 2.5662,
1192
+ "eval_samples_per_second": 108.33,
1193
+ "eval_steps_per_second": 13.639,
1194
+ "step": 1500
1195
+ },
1196
+ {
1197
+ "epoch": 9.679487179487179,
1198
+ "grad_norm": 0.012905549257993698,
1199
+ "learning_rate": 5.1084812623274167e-05,
1200
+ "loss": 0.0023,
1201
+ "step": 1510
1202
+ },
1203
+ {
1204
+ "epoch": 9.743589743589745,
1205
+ "grad_norm": 0.011163388378918171,
1206
+ "learning_rate": 5.009861932938856e-05,
1207
+ "loss": 0.0023,
1208
+ "step": 1520
1209
+ },
1210
+ {
1211
+ "epoch": 9.807692307692308,
1212
+ "grad_norm": 0.012818182818591595,
1213
+ "learning_rate": 4.9112426035502965e-05,
1214
+ "loss": 0.0023,
1215
+ "step": 1530
1216
+ },
1217
+ {
1218
+ "epoch": 9.871794871794872,
1219
+ "grad_norm": 0.012148367241024971,
1220
+ "learning_rate": 4.812623274161736e-05,
1221
+ "loss": 0.0023,
1222
+ "step": 1540
1223
+ },
1224
+ {
1225
+ "epoch": 9.935897435897436,
1226
+ "grad_norm": 0.012448051013052464,
1227
+ "learning_rate": 4.714003944773176e-05,
1228
+ "loss": 0.0021,
1229
+ "step": 1550
1230
+ },
1231
+ {
1232
+ "epoch": 10.0,
1233
+ "grad_norm": 0.01428731344640255,
1234
+ "learning_rate": 4.615384615384616e-05,
1235
+ "loss": 0.0024,
1236
+ "step": 1560
1237
+ },
1238
+ {
1239
+ "epoch": 10.064102564102564,
1240
+ "grad_norm": 0.012622198089957237,
1241
+ "learning_rate": 4.5167652859960554e-05,
1242
+ "loss": 0.0021,
1243
+ "step": 1570
1244
+ },
1245
+ {
1246
+ "epoch": 10.128205128205128,
1247
+ "grad_norm": 0.011053148657083511,
1248
+ "learning_rate": 4.418145956607495e-05,
1249
+ "loss": 0.0022,
1250
+ "step": 1580
1251
+ },
1252
+ {
1253
+ "epoch": 10.192307692307692,
1254
+ "grad_norm": 0.010602343827486038,
1255
+ "learning_rate": 4.319526627218935e-05,
1256
+ "loss": 0.0022,
1257
+ "step": 1590
1258
+ },
1259
+ {
1260
+ "epoch": 10.256410256410255,
1261
+ "grad_norm": 0.012038341723382473,
1262
+ "learning_rate": 4.220907297830375e-05,
1263
+ "loss": 0.0021,
1264
+ "step": 1600
1265
+ },
1266
+ {
1267
+ "epoch": 10.256410256410255,
1268
+ "eval_accuracy": 0.9460431654676259,
1269
+ "eval_loss": 0.28900742530822754,
1270
+ "eval_runtime": 3.7223,
1271
+ "eval_samples_per_second": 74.686,
1272
+ "eval_steps_per_second": 9.403,
1273
+ "step": 1600
1274
+ },
1275
+ {
1276
+ "epoch": 10.320512820512821,
1277
+ "grad_norm": 0.010555533692240715,
1278
+ "learning_rate": 4.122287968441815e-05,
1279
+ "loss": 0.0022,
1280
+ "step": 1610
1281
+ },
1282
+ {
1283
+ "epoch": 10.384615384615385,
1284
+ "grad_norm": 0.013419007882475853,
1285
+ "learning_rate": 4.0236686390532545e-05,
1286
+ "loss": 0.0023,
1287
+ "step": 1620
1288
+ },
1289
+ {
1290
+ "epoch": 10.448717948717949,
1291
+ "grad_norm": 0.009542490355670452,
1292
+ "learning_rate": 3.925049309664695e-05,
1293
+ "loss": 0.0021,
1294
+ "step": 1630
1295
+ },
1296
+ {
1297
+ "epoch": 10.512820512820513,
1298
+ "grad_norm": 0.01066176313906908,
1299
+ "learning_rate": 3.826429980276134e-05,
1300
+ "loss": 0.0021,
1301
+ "step": 1640
1302
+ },
1303
+ {
1304
+ "epoch": 10.576923076923077,
1305
+ "grad_norm": 0.009535272605717182,
1306
+ "learning_rate": 3.7278106508875746e-05,
1307
+ "loss": 0.002,
1308
+ "step": 1650
1309
+ },
1310
+ {
1311
+ "epoch": 10.64102564102564,
1312
+ "grad_norm": 0.010441762395203114,
1313
+ "learning_rate": 3.629191321499014e-05,
1314
+ "loss": 0.0021,
1315
+ "step": 1660
1316
+ },
1317
+ {
1318
+ "epoch": 10.705128205128204,
1319
+ "grad_norm": 0.009478352032601833,
1320
+ "learning_rate": 3.5305719921104544e-05,
1321
+ "loss": 0.002,
1322
+ "step": 1670
1323
+ },
1324
+ {
1325
+ "epoch": 10.76923076923077,
1326
+ "grad_norm": 0.009695933200418949,
1327
+ "learning_rate": 3.431952662721893e-05,
1328
+ "loss": 0.002,
1329
+ "step": 1680
1330
+ },
1331
+ {
1332
+ "epoch": 10.833333333333334,
1333
+ "grad_norm": 0.009181487374007702,
1334
+ "learning_rate": 3.3333333333333335e-05,
1335
+ "loss": 0.0019,
1336
+ "step": 1690
1337
+ },
1338
+ {
1339
+ "epoch": 10.897435897435898,
1340
+ "grad_norm": 0.010168403387069702,
1341
+ "learning_rate": 3.234714003944773e-05,
1342
+ "loss": 0.002,
1343
+ "step": 1700
1344
+ },
1345
+ {
1346
+ "epoch": 10.897435897435898,
1347
+ "eval_accuracy": 0.9460431654676259,
1348
+ "eval_loss": 0.29342445731163025,
1349
+ "eval_runtime": 2.6399,
1350
+ "eval_samples_per_second": 105.306,
1351
+ "eval_steps_per_second": 13.258,
1352
+ "step": 1700
1353
+ },
1354
+ {
1355
+ "epoch": 10.961538461538462,
1356
+ "grad_norm": 0.012188075110316277,
1357
+ "learning_rate": 3.136094674556213e-05,
1358
+ "loss": 0.0019,
1359
+ "step": 1710
1360
+ },
1361
+ {
1362
+ "epoch": 11.025641025641026,
1363
+ "grad_norm": 0.01147390529513359,
1364
+ "learning_rate": 3.037475345167653e-05,
1365
+ "loss": 0.002,
1366
+ "step": 1720
1367
+ },
1368
+ {
1369
+ "epoch": 11.08974358974359,
1370
+ "grad_norm": 0.00882646068930626,
1371
+ "learning_rate": 2.9388560157790928e-05,
1372
+ "loss": 0.0019,
1373
+ "step": 1730
1374
+ },
1375
+ {
1376
+ "epoch": 11.153846153846153,
1377
+ "grad_norm": 0.00944901816546917,
1378
+ "learning_rate": 2.8402366863905327e-05,
1379
+ "loss": 0.002,
1380
+ "step": 1740
1381
+ },
1382
+ {
1383
+ "epoch": 11.217948717948717,
1384
+ "grad_norm": 0.009782733395695686,
1385
+ "learning_rate": 2.7416173570019726e-05,
1386
+ "loss": 0.0019,
1387
+ "step": 1750
1388
+ },
1389
+ {
1390
+ "epoch": 11.282051282051283,
1391
+ "grad_norm": 0.008780627511441708,
1392
+ "learning_rate": 2.6429980276134125e-05,
1393
+ "loss": 0.0019,
1394
+ "step": 1760
1395
+ },
1396
+ {
1397
+ "epoch": 11.346153846153847,
1398
+ "grad_norm": 0.011177603155374527,
1399
+ "learning_rate": 2.5443786982248524e-05,
1400
+ "loss": 0.0019,
1401
+ "step": 1770
1402
+ },
1403
+ {
1404
+ "epoch": 11.41025641025641,
1405
+ "grad_norm": 0.009699178859591484,
1406
+ "learning_rate": 2.445759368836292e-05,
1407
+ "loss": 0.0018,
1408
+ "step": 1780
1409
+ },
1410
+ {
1411
+ "epoch": 11.474358974358974,
1412
+ "grad_norm": 0.012270525097846985,
1413
+ "learning_rate": 2.3471400394477318e-05,
1414
+ "loss": 0.0019,
1415
+ "step": 1790
1416
+ },
1417
+ {
1418
+ "epoch": 11.538461538461538,
1419
+ "grad_norm": 0.00872350949794054,
1420
+ "learning_rate": 2.2485207100591717e-05,
1421
+ "loss": 0.0019,
1422
+ "step": 1800
1423
+ },
1424
+ {
1425
+ "epoch": 11.538461538461538,
1426
+ "eval_accuracy": 0.9460431654676259,
1427
+ "eval_loss": 0.2975642681121826,
1428
+ "eval_runtime": 2.6404,
1429
+ "eval_samples_per_second": 105.285,
1430
+ "eval_steps_per_second": 13.255,
1431
+ "step": 1800
1432
+ },
1433
+ {
1434
+ "epoch": 11.602564102564102,
1435
+ "grad_norm": 0.009663211181759834,
1436
+ "learning_rate": 2.1499013806706113e-05,
1437
+ "loss": 0.0019,
1438
+ "step": 1810
1439
+ },
1440
+ {
1441
+ "epoch": 11.666666666666666,
1442
+ "grad_norm": 0.00840259250253439,
1443
+ "learning_rate": 2.0512820512820512e-05,
1444
+ "loss": 0.0018,
1445
+ "step": 1820
1446
+ },
1447
+ {
1448
+ "epoch": 11.73076923076923,
1449
+ "grad_norm": 0.010098517872393131,
1450
+ "learning_rate": 1.952662721893491e-05,
1451
+ "loss": 0.0019,
1452
+ "step": 1830
1453
+ },
1454
+ {
1455
+ "epoch": 11.794871794871796,
1456
+ "grad_norm": 0.013970685191452503,
1457
+ "learning_rate": 1.854043392504931e-05,
1458
+ "loss": 0.0019,
1459
+ "step": 1840
1460
+ },
1461
+ {
1462
+ "epoch": 11.85897435897436,
1463
+ "grad_norm": 0.009059489704668522,
1464
+ "learning_rate": 1.755424063116371e-05,
1465
+ "loss": 0.0018,
1466
+ "step": 1850
1467
+ },
1468
+ {
1469
+ "epoch": 11.923076923076923,
1470
+ "grad_norm": 0.009793447330594063,
1471
+ "learning_rate": 1.6568047337278108e-05,
1472
+ "loss": 0.0019,
1473
+ "step": 1860
1474
+ },
1475
+ {
1476
+ "epoch": 11.987179487179487,
1477
+ "grad_norm": 0.008815682493150234,
1478
+ "learning_rate": 1.5581854043392503e-05,
1479
+ "loss": 0.0019,
1480
+ "step": 1870
1481
+ },
1482
+ {
1483
+ "epoch": 12.051282051282051,
1484
+ "grad_norm": 0.011554540134966373,
1485
+ "learning_rate": 1.4595660749506904e-05,
1486
+ "loss": 0.0018,
1487
+ "step": 1880
1488
+ },
1489
+ {
1490
+ "epoch": 12.115384615384615,
1491
+ "grad_norm": 0.010211811400949955,
1492
+ "learning_rate": 1.3609467455621303e-05,
1493
+ "loss": 0.0019,
1494
+ "step": 1890
1495
+ },
1496
+ {
1497
+ "epoch": 12.179487179487179,
1498
+ "grad_norm": 0.008440189994871616,
1499
+ "learning_rate": 1.2623274161735702e-05,
1500
+ "loss": 0.0018,
1501
+ "step": 1900
1502
+ },
1503
+ {
1504
+ "epoch": 12.179487179487179,
1505
+ "eval_accuracy": 0.9460431654676259,
1506
+ "eval_loss": 0.2996102273464203,
1507
+ "eval_runtime": 3.4755,
1508
+ "eval_samples_per_second": 79.989,
1509
+ "eval_steps_per_second": 10.071,
1510
+ "step": 1900
1511
+ },
1512
+ {
1513
+ "epoch": 12.243589743589743,
1514
+ "grad_norm": 0.010010773316025734,
1515
+ "learning_rate": 1.16370808678501e-05,
1516
+ "loss": 0.0018,
1517
+ "step": 1910
1518
+ },
1519
+ {
1520
+ "epoch": 12.307692307692308,
1521
+ "grad_norm": 0.008343766443431377,
1522
+ "learning_rate": 1.0650887573964498e-05,
1523
+ "loss": 0.0017,
1524
+ "step": 1920
1525
+ },
1526
+ {
1527
+ "epoch": 12.371794871794872,
1528
+ "grad_norm": 0.009077221155166626,
1529
+ "learning_rate": 9.664694280078896e-06,
1530
+ "loss": 0.0018,
1531
+ "step": 1930
1532
+ },
1533
+ {
1534
+ "epoch": 12.435897435897436,
1535
+ "grad_norm": 0.008946426212787628,
1536
+ "learning_rate": 8.678500986193295e-06,
1537
+ "loss": 0.0018,
1538
+ "step": 1940
1539
+ },
1540
+ {
1541
+ "epoch": 12.5,
1542
+ "grad_norm": 0.008481796830892563,
1543
+ "learning_rate": 7.692307692307694e-06,
1544
+ "loss": 0.0018,
1545
+ "step": 1950
1546
+ },
1547
+ {
1548
+ "epoch": 12.564102564102564,
1549
+ "grad_norm": 0.012229247018694878,
1550
+ "learning_rate": 6.706114398422091e-06,
1551
+ "loss": 0.0018,
1552
+ "step": 1960
1553
+ },
1554
+ {
1555
+ "epoch": 12.628205128205128,
1556
+ "grad_norm": 0.009792056865990162,
1557
+ "learning_rate": 5.71992110453649e-06,
1558
+ "loss": 0.0019,
1559
+ "step": 1970
1560
+ },
1561
+ {
1562
+ "epoch": 12.692307692307692,
1563
+ "grad_norm": 0.008133339695632458,
1564
+ "learning_rate": 4.733727810650888e-06,
1565
+ "loss": 0.0017,
1566
+ "step": 1980
1567
+ },
1568
+ {
1569
+ "epoch": 12.756410256410255,
1570
+ "grad_norm": 0.00821082666516304,
1571
+ "learning_rate": 3.7475345167652858e-06,
1572
+ "loss": 0.0018,
1573
+ "step": 1990
1574
+ },
1575
+ {
1576
+ "epoch": 12.820512820512821,
1577
+ "grad_norm": 0.008551714941859245,
1578
+ "learning_rate": 2.7613412228796843e-06,
1579
+ "loss": 0.0018,
1580
+ "step": 2000
1581
+ },
1582
+ {
1583
+ "epoch": 12.820512820512821,
1584
+ "eval_accuracy": 0.9460431654676259,
1585
+ "eval_loss": 0.3006349802017212,
1586
+ "eval_runtime": 2.5194,
1587
+ "eval_samples_per_second": 110.343,
1588
+ "eval_steps_per_second": 13.892,
1589
+ "step": 2000
1590
+ },
1591
+ {
1592
+ "epoch": 12.884615384615385,
1593
+ "grad_norm": 0.008309854194521904,
1594
+ "learning_rate": 1.775147928994083e-06,
1595
+ "loss": 0.0019,
1596
+ "step": 2010
1597
+ },
1598
+ {
1599
+ "epoch": 12.948717948717949,
1600
+ "grad_norm": 0.008332509547472,
1601
+ "learning_rate": 7.889546351084814e-07,
1602
+ "loss": 0.0017,
1603
+ "step": 2020
1604
+ },
1605
+ {
1606
+ "epoch": 13.0,
1607
+ "step": 2028,
1608
+ "total_flos": 2.5134977377120666e+18,
1609
+ "train_loss": 0.18042536613924515,
1610
+ "train_runtime": 1249.5195,
1611
+ "train_samples_per_second": 25.958,
1612
+ "train_steps_per_second": 1.623
1613
+ }
1614
+ ],
1615
+ "logging_steps": 10,
1616
+ "max_steps": 2028,
1617
+ "num_input_tokens_seen": 0,
1618
+ "num_train_epochs": 13,
1619
+ "save_steps": 100,
1620
+ "stateful_callbacks": {
1621
+ "TrainerControl": {
1622
+ "args": {
1623
+ "should_epoch_stop": false,
1624
+ "should_evaluate": false,
1625
+ "should_log": false,
1626
+ "should_save": true,
1627
+ "should_training_stop": false
1628
+ },
1629
+ "attributes": {}
1630
+ }
1631
+ },
1632
+ "total_flos": 2.5134977377120666e+18,
1633
+ "train_batch_size": 16,
1634
+ "trial_name": null,
1635
+ "trial_params": null
1636
+ }