Jakir057 commited on
Commit
1722f1f
1 Parent(s): 194268e

🍻 cheers

Browse files
Files changed (5) hide show
  1. README.md +3 -2
  2. all_results.json +13 -0
  3. eval_results.json +8 -0
  4. train_results.json +8 -0
  5. trainer_state.json +2662 -0
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
  metrics:
7
  - accuracy
@@ -15,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  # finetuned-indian-food
17
 
18
- This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
- - Loss: 0.0027
21
  - Accuracy: 0.9996
22
 
23
  ## Model description
 
2
  license: apache-2.0
3
  base_model: google/vit-base-patch16-224-in21k
4
  tags:
5
+ - image-classification
6
  - generated_from_trainer
7
  metrics:
8
  - accuracy
 
16
 
17
  # finetuned-indian-food
18
 
19
+ 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 indian_food_images dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 0.0026
22
  - Accuracy: 0.9996
23
 
24
  ## Model description
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "eval_accuracy": 0.9996296296296296,
4
+ "eval_loss": 0.002558160340413451,
5
+ "eval_runtime": 16.1541,
6
+ "eval_samples_per_second": 167.14,
7
+ "eval_steps_per_second": 20.923,
8
+ "total_flos": 4.742807303717683e+18,
9
+ "train_loss": 0.13919885493931725,
10
+ "train_runtime": 1939.8119,
11
+ "train_samples_per_second": 31.549,
12
+ "train_steps_per_second": 1.973
13
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "eval_accuracy": 0.9996296296296296,
4
+ "eval_loss": 0.002558160340413451,
5
+ "eval_runtime": 16.1541,
6
+ "eval_samples_per_second": 167.14,
7
+ "eval_steps_per_second": 20.923
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "total_flos": 4.742807303717683e+18,
4
+ "train_loss": 0.13919885493931725,
5
+ "train_runtime": 1939.8119,
6
+ "train_samples_per_second": 31.549,
7
+ "train_steps_per_second": 1.973
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,2662 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.002558160340413451,
3
+ "best_model_checkpoint": "finetuned-indian-food/checkpoint-3600",
4
+ "epoch": 4.0,
5
+ "eval_steps": 100,
6
+ "global_step": 3828,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 0.00019952978056426334,
14
+ "loss": 2.1561,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.02,
19
+ "learning_rate": 0.0001990073145245559,
20
+ "loss": 2.0371,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.03,
25
+ "learning_rate": 0.0001984848484848485,
26
+ "loss": 1.7093,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.04,
31
+ "learning_rate": 0.00019796238244514108,
32
+ "loss": 1.4792,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.05,
37
+ "learning_rate": 0.00019743991640543368,
38
+ "loss": 1.1808,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.06,
43
+ "learning_rate": 0.00019691745036572625,
44
+ "loss": 0.9715,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.07,
49
+ "learning_rate": 0.00019639498432601882,
50
+ "loss": 0.8425,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.08,
55
+ "learning_rate": 0.0001958725182863114,
56
+ "loss": 0.8624,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.09,
61
+ "learning_rate": 0.000195350052246604,
62
+ "loss": 0.7628,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.1,
67
+ "learning_rate": 0.00019482758620689657,
68
+ "loss": 0.7056,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.1,
73
+ "eval_accuracy": 0.8881481481481481,
74
+ "eval_loss": 0.5113354325294495,
75
+ "eval_runtime": 21.3674,
76
+ "eval_samples_per_second": 126.361,
77
+ "eval_steps_per_second": 15.818,
78
+ "step": 100
79
+ },
80
+ {
81
+ "epoch": 0.11,
82
+ "learning_rate": 0.00019430512016718914,
83
+ "loss": 0.6112,
84
+ "step": 110
85
+ },
86
+ {
87
+ "epoch": 0.13,
88
+ "learning_rate": 0.00019378265412748174,
89
+ "loss": 0.5859,
90
+ "step": 120
91
+ },
92
+ {
93
+ "epoch": 0.14,
94
+ "learning_rate": 0.0001932601880877743,
95
+ "loss": 0.5764,
96
+ "step": 130
97
+ },
98
+ {
99
+ "epoch": 0.15,
100
+ "learning_rate": 0.00019273772204806688,
101
+ "loss": 0.5528,
102
+ "step": 140
103
+ },
104
+ {
105
+ "epoch": 0.16,
106
+ "learning_rate": 0.00019221525600835945,
107
+ "loss": 0.4072,
108
+ "step": 150
109
+ },
110
+ {
111
+ "epoch": 0.17,
112
+ "learning_rate": 0.00019169278996865205,
113
+ "loss": 0.388,
114
+ "step": 160
115
+ },
116
+ {
117
+ "epoch": 0.18,
118
+ "learning_rate": 0.00019117032392894462,
119
+ "loss": 0.399,
120
+ "step": 170
121
+ },
122
+ {
123
+ "epoch": 0.19,
124
+ "learning_rate": 0.0001906478578892372,
125
+ "loss": 0.2888,
126
+ "step": 180
127
+ },
128
+ {
129
+ "epoch": 0.2,
130
+ "learning_rate": 0.0001901253918495298,
131
+ "loss": 0.3242,
132
+ "step": 190
133
+ },
134
+ {
135
+ "epoch": 0.21,
136
+ "learning_rate": 0.00018960292580982237,
137
+ "loss": 0.3027,
138
+ "step": 200
139
+ },
140
+ {
141
+ "epoch": 0.21,
142
+ "eval_accuracy": 0.9796296296296296,
143
+ "eval_loss": 0.12799212336540222,
144
+ "eval_runtime": 16.2334,
145
+ "eval_samples_per_second": 166.323,
146
+ "eval_steps_per_second": 20.821,
147
+ "step": 200
148
+ },
149
+ {
150
+ "epoch": 0.22,
151
+ "learning_rate": 0.00018908045977011494,
152
+ "loss": 0.2855,
153
+ "step": 210
154
+ },
155
+ {
156
+ "epoch": 0.23,
157
+ "learning_rate": 0.00018855799373040754,
158
+ "loss": 0.3776,
159
+ "step": 220
160
+ },
161
+ {
162
+ "epoch": 0.24,
163
+ "learning_rate": 0.0001880355276907001,
164
+ "loss": 0.2458,
165
+ "step": 230
166
+ },
167
+ {
168
+ "epoch": 0.25,
169
+ "learning_rate": 0.0001875130616509927,
170
+ "loss": 0.3555,
171
+ "step": 240
172
+ },
173
+ {
174
+ "epoch": 0.26,
175
+ "learning_rate": 0.00018699059561128528,
176
+ "loss": 0.4411,
177
+ "step": 250
178
+ },
179
+ {
180
+ "epoch": 0.27,
181
+ "learning_rate": 0.00018646812957157788,
182
+ "loss": 0.1867,
183
+ "step": 260
184
+ },
185
+ {
186
+ "epoch": 0.28,
187
+ "learning_rate": 0.00018594566353187045,
188
+ "loss": 0.5165,
189
+ "step": 270
190
+ },
191
+ {
192
+ "epoch": 0.29,
193
+ "learning_rate": 0.00018542319749216302,
194
+ "loss": 0.3386,
195
+ "step": 280
196
+ },
197
+ {
198
+ "epoch": 0.3,
199
+ "learning_rate": 0.0001849007314524556,
200
+ "loss": 0.2614,
201
+ "step": 290
202
+ },
203
+ {
204
+ "epoch": 0.31,
205
+ "learning_rate": 0.0001843782654127482,
206
+ "loss": 0.2823,
207
+ "step": 300
208
+ },
209
+ {
210
+ "epoch": 0.31,
211
+ "eval_accuracy": 0.9655555555555555,
212
+ "eval_loss": 0.157969132065773,
213
+ "eval_runtime": 16.9126,
214
+ "eval_samples_per_second": 159.645,
215
+ "eval_steps_per_second": 19.985,
216
+ "step": 300
217
+ },
218
+ {
219
+ "epoch": 0.32,
220
+ "learning_rate": 0.00018385579937304076,
221
+ "loss": 0.3546,
222
+ "step": 310
223
+ },
224
+ {
225
+ "epoch": 0.33,
226
+ "learning_rate": 0.00018333333333333334,
227
+ "loss": 0.2974,
228
+ "step": 320
229
+ },
230
+ {
231
+ "epoch": 0.34,
232
+ "learning_rate": 0.00018281086729362593,
233
+ "loss": 0.22,
234
+ "step": 330
235
+ },
236
+ {
237
+ "epoch": 0.36,
238
+ "learning_rate": 0.0001822884012539185,
239
+ "loss": 0.3668,
240
+ "step": 340
241
+ },
242
+ {
243
+ "epoch": 0.37,
244
+ "learning_rate": 0.00018176593521421108,
245
+ "loss": 0.2249,
246
+ "step": 350
247
+ },
248
+ {
249
+ "epoch": 0.38,
250
+ "learning_rate": 0.00018124346917450365,
251
+ "loss": 0.1652,
252
+ "step": 360
253
+ },
254
+ {
255
+ "epoch": 0.39,
256
+ "learning_rate": 0.00018072100313479625,
257
+ "loss": 0.1557,
258
+ "step": 370
259
+ },
260
+ {
261
+ "epoch": 0.4,
262
+ "learning_rate": 0.00018019853709508882,
263
+ "loss": 0.1811,
264
+ "step": 380
265
+ },
266
+ {
267
+ "epoch": 0.41,
268
+ "learning_rate": 0.0001796760710553814,
269
+ "loss": 0.189,
270
+ "step": 390
271
+ },
272
+ {
273
+ "epoch": 0.42,
274
+ "learning_rate": 0.000179153605015674,
275
+ "loss": 0.3273,
276
+ "step": 400
277
+ },
278
+ {
279
+ "epoch": 0.42,
280
+ "eval_accuracy": 0.9837037037037037,
281
+ "eval_loss": 0.0879134014248848,
282
+ "eval_runtime": 16.3211,
283
+ "eval_samples_per_second": 165.43,
284
+ "eval_steps_per_second": 20.709,
285
+ "step": 400
286
+ },
287
+ {
288
+ "epoch": 0.43,
289
+ "learning_rate": 0.00017863113897596656,
290
+ "loss": 0.1842,
291
+ "step": 410
292
+ },
293
+ {
294
+ "epoch": 0.44,
295
+ "learning_rate": 0.00017810867293625913,
296
+ "loss": 0.2524,
297
+ "step": 420
298
+ },
299
+ {
300
+ "epoch": 0.45,
301
+ "learning_rate": 0.00017758620689655173,
302
+ "loss": 0.1823,
303
+ "step": 430
304
+ },
305
+ {
306
+ "epoch": 0.46,
307
+ "learning_rate": 0.0001770637408568443,
308
+ "loss": 0.1912,
309
+ "step": 440
310
+ },
311
+ {
312
+ "epoch": 0.47,
313
+ "learning_rate": 0.0001765412748171369,
314
+ "loss": 0.2241,
315
+ "step": 450
316
+ },
317
+ {
318
+ "epoch": 0.48,
319
+ "learning_rate": 0.00017601880877742948,
320
+ "loss": 0.2356,
321
+ "step": 460
322
+ },
323
+ {
324
+ "epoch": 0.49,
325
+ "learning_rate": 0.00017549634273772208,
326
+ "loss": 0.196,
327
+ "step": 470
328
+ },
329
+ {
330
+ "epoch": 0.5,
331
+ "learning_rate": 0.00017497387669801465,
332
+ "loss": 0.1365,
333
+ "step": 480
334
+ },
335
+ {
336
+ "epoch": 0.51,
337
+ "learning_rate": 0.00017445141065830722,
338
+ "loss": 0.2863,
339
+ "step": 490
340
+ },
341
+ {
342
+ "epoch": 0.52,
343
+ "learning_rate": 0.00017392894461859982,
344
+ "loss": 0.1808,
345
+ "step": 500
346
+ },
347
+ {
348
+ "epoch": 0.52,
349
+ "eval_accuracy": 0.9822222222222222,
350
+ "eval_loss": 0.08116234838962555,
351
+ "eval_runtime": 16.1751,
352
+ "eval_samples_per_second": 166.923,
353
+ "eval_steps_per_second": 20.896,
354
+ "step": 500
355
+ },
356
+ {
357
+ "epoch": 0.53,
358
+ "learning_rate": 0.0001734064785788924,
359
+ "loss": 0.2045,
360
+ "step": 510
361
+ },
362
+ {
363
+ "epoch": 0.54,
364
+ "learning_rate": 0.00017288401253918496,
365
+ "loss": 0.1195,
366
+ "step": 520
367
+ },
368
+ {
369
+ "epoch": 0.55,
370
+ "learning_rate": 0.00017236154649947753,
371
+ "loss": 0.1668,
372
+ "step": 530
373
+ },
374
+ {
375
+ "epoch": 0.56,
376
+ "learning_rate": 0.00017183908045977013,
377
+ "loss": 0.2419,
378
+ "step": 540
379
+ },
380
+ {
381
+ "epoch": 0.57,
382
+ "learning_rate": 0.0001713166144200627,
383
+ "loss": 0.2098,
384
+ "step": 550
385
+ },
386
+ {
387
+ "epoch": 0.59,
388
+ "learning_rate": 0.00017079414838035528,
389
+ "loss": 0.1827,
390
+ "step": 560
391
+ },
392
+ {
393
+ "epoch": 0.6,
394
+ "learning_rate": 0.00017027168234064787,
395
+ "loss": 0.184,
396
+ "step": 570
397
+ },
398
+ {
399
+ "epoch": 0.61,
400
+ "learning_rate": 0.00016974921630094045,
401
+ "loss": 0.2049,
402
+ "step": 580
403
+ },
404
+ {
405
+ "epoch": 0.62,
406
+ "learning_rate": 0.00016922675026123302,
407
+ "loss": 0.118,
408
+ "step": 590
409
+ },
410
+ {
411
+ "epoch": 0.63,
412
+ "learning_rate": 0.0001687042842215256,
413
+ "loss": 0.2101,
414
+ "step": 600
415
+ },
416
+ {
417
+ "epoch": 0.63,
418
+ "eval_accuracy": 0.9937037037037038,
419
+ "eval_loss": 0.03391774371266365,
420
+ "eval_runtime": 16.2812,
421
+ "eval_samples_per_second": 165.835,
422
+ "eval_steps_per_second": 20.76,
423
+ "step": 600
424
+ },
425
+ {
426
+ "epoch": 0.64,
427
+ "learning_rate": 0.0001681818181818182,
428
+ "loss": 0.1021,
429
+ "step": 610
430
+ },
431
+ {
432
+ "epoch": 0.65,
433
+ "learning_rate": 0.00016765935214211076,
434
+ "loss": 0.3133,
435
+ "step": 620
436
+ },
437
+ {
438
+ "epoch": 0.66,
439
+ "learning_rate": 0.00016713688610240333,
440
+ "loss": 0.3214,
441
+ "step": 630
442
+ },
443
+ {
444
+ "epoch": 0.67,
445
+ "learning_rate": 0.00016661442006269593,
446
+ "loss": 0.232,
447
+ "step": 640
448
+ },
449
+ {
450
+ "epoch": 0.68,
451
+ "learning_rate": 0.0001660919540229885,
452
+ "loss": 0.0819,
453
+ "step": 650
454
+ },
455
+ {
456
+ "epoch": 0.69,
457
+ "learning_rate": 0.0001655694879832811,
458
+ "loss": 0.1256,
459
+ "step": 660
460
+ },
461
+ {
462
+ "epoch": 0.7,
463
+ "learning_rate": 0.00016504702194357367,
464
+ "loss": 0.1577,
465
+ "step": 670
466
+ },
467
+ {
468
+ "epoch": 0.71,
469
+ "learning_rate": 0.00016452455590386627,
470
+ "loss": 0.0825,
471
+ "step": 680
472
+ },
473
+ {
474
+ "epoch": 0.72,
475
+ "learning_rate": 0.00016400208986415885,
476
+ "loss": 0.1146,
477
+ "step": 690
478
+ },
479
+ {
480
+ "epoch": 0.73,
481
+ "learning_rate": 0.00016347962382445142,
482
+ "loss": 0.1495,
483
+ "step": 700
484
+ },
485
+ {
486
+ "epoch": 0.73,
487
+ "eval_accuracy": 0.9833333333333333,
488
+ "eval_loss": 0.0568382553756237,
489
+ "eval_runtime": 16.2547,
490
+ "eval_samples_per_second": 166.106,
491
+ "eval_steps_per_second": 20.794,
492
+ "step": 700
493
+ },
494
+ {
495
+ "epoch": 0.74,
496
+ "learning_rate": 0.00016295715778474402,
497
+ "loss": 0.197,
498
+ "step": 710
499
+ },
500
+ {
501
+ "epoch": 0.75,
502
+ "learning_rate": 0.0001624346917450366,
503
+ "loss": 0.1708,
504
+ "step": 720
505
+ },
506
+ {
507
+ "epoch": 0.76,
508
+ "learning_rate": 0.00016191222570532916,
509
+ "loss": 0.1702,
510
+ "step": 730
511
+ },
512
+ {
513
+ "epoch": 0.77,
514
+ "learning_rate": 0.00016138975966562173,
515
+ "loss": 0.3196,
516
+ "step": 740
517
+ },
518
+ {
519
+ "epoch": 0.78,
520
+ "learning_rate": 0.00016086729362591433,
521
+ "loss": 0.1967,
522
+ "step": 750
523
+ },
524
+ {
525
+ "epoch": 0.79,
526
+ "learning_rate": 0.0001603448275862069,
527
+ "loss": 0.1943,
528
+ "step": 760
529
+ },
530
+ {
531
+ "epoch": 0.8,
532
+ "learning_rate": 0.00015982236154649947,
533
+ "loss": 0.1489,
534
+ "step": 770
535
+ },
536
+ {
537
+ "epoch": 0.82,
538
+ "learning_rate": 0.00015929989550679207,
539
+ "loss": 0.1284,
540
+ "step": 780
541
+ },
542
+ {
543
+ "epoch": 0.83,
544
+ "learning_rate": 0.00015877742946708464,
545
+ "loss": 0.2296,
546
+ "step": 790
547
+ },
548
+ {
549
+ "epoch": 0.84,
550
+ "learning_rate": 0.00015825496342737722,
551
+ "loss": 0.1296,
552
+ "step": 800
553
+ },
554
+ {
555
+ "epoch": 0.84,
556
+ "eval_accuracy": 0.9844444444444445,
557
+ "eval_loss": 0.06289423257112503,
558
+ "eval_runtime": 16.2541,
559
+ "eval_samples_per_second": 166.112,
560
+ "eval_steps_per_second": 20.795,
561
+ "step": 800
562
+ },
563
+ {
564
+ "epoch": 0.85,
565
+ "learning_rate": 0.00015773249738766982,
566
+ "loss": 0.0969,
567
+ "step": 810
568
+ },
569
+ {
570
+ "epoch": 0.86,
571
+ "learning_rate": 0.0001572100313479624,
572
+ "loss": 0.1199,
573
+ "step": 820
574
+ },
575
+ {
576
+ "epoch": 0.87,
577
+ "learning_rate": 0.00015668756530825496,
578
+ "loss": 0.1064,
579
+ "step": 830
580
+ },
581
+ {
582
+ "epoch": 0.88,
583
+ "learning_rate": 0.00015616509926854753,
584
+ "loss": 0.2688,
585
+ "step": 840
586
+ },
587
+ {
588
+ "epoch": 0.89,
589
+ "learning_rate": 0.00015564263322884013,
590
+ "loss": 0.2549,
591
+ "step": 850
592
+ },
593
+ {
594
+ "epoch": 0.9,
595
+ "learning_rate": 0.0001551201671891327,
596
+ "loss": 0.1788,
597
+ "step": 860
598
+ },
599
+ {
600
+ "epoch": 0.91,
601
+ "learning_rate": 0.0001545977011494253,
602
+ "loss": 0.0653,
603
+ "step": 870
604
+ },
605
+ {
606
+ "epoch": 0.92,
607
+ "learning_rate": 0.00015407523510971787,
608
+ "loss": 0.3013,
609
+ "step": 880
610
+ },
611
+ {
612
+ "epoch": 0.93,
613
+ "learning_rate": 0.00015355276907001047,
614
+ "loss": 0.167,
615
+ "step": 890
616
+ },
617
+ {
618
+ "epoch": 0.94,
619
+ "learning_rate": 0.00015303030303030304,
620
+ "loss": 0.1462,
621
+ "step": 900
622
+ },
623
+ {
624
+ "epoch": 0.94,
625
+ "eval_accuracy": 0.9733333333333334,
626
+ "eval_loss": 0.08862249553203583,
627
+ "eval_runtime": 16.3153,
628
+ "eval_samples_per_second": 165.489,
629
+ "eval_steps_per_second": 20.717,
630
+ "step": 900
631
+ },
632
+ {
633
+ "epoch": 0.95,
634
+ "learning_rate": 0.00015250783699059562,
635
+ "loss": 0.1513,
636
+ "step": 910
637
+ },
638
+ {
639
+ "epoch": 0.96,
640
+ "learning_rate": 0.00015198537095088821,
641
+ "loss": 0.1546,
642
+ "step": 920
643
+ },
644
+ {
645
+ "epoch": 0.97,
646
+ "learning_rate": 0.00015146290491118079,
647
+ "loss": 0.0779,
648
+ "step": 930
649
+ },
650
+ {
651
+ "epoch": 0.98,
652
+ "learning_rate": 0.00015094043887147336,
653
+ "loss": 0.1373,
654
+ "step": 940
655
+ },
656
+ {
657
+ "epoch": 0.99,
658
+ "learning_rate": 0.00015041797283176596,
659
+ "loss": 0.0776,
660
+ "step": 950
661
+ },
662
+ {
663
+ "epoch": 1.0,
664
+ "learning_rate": 0.00014989550679205853,
665
+ "loss": 0.0995,
666
+ "step": 960
667
+ },
668
+ {
669
+ "epoch": 1.01,
670
+ "learning_rate": 0.0001493730407523511,
671
+ "loss": 0.1783,
672
+ "step": 970
673
+ },
674
+ {
675
+ "epoch": 1.02,
676
+ "learning_rate": 0.00014885057471264367,
677
+ "loss": 0.1035,
678
+ "step": 980
679
+ },
680
+ {
681
+ "epoch": 1.03,
682
+ "learning_rate": 0.00014832810867293627,
683
+ "loss": 0.1682,
684
+ "step": 990
685
+ },
686
+ {
687
+ "epoch": 1.04,
688
+ "learning_rate": 0.00014780564263322884,
689
+ "loss": 0.0519,
690
+ "step": 1000
691
+ },
692
+ {
693
+ "epoch": 1.04,
694
+ "eval_accuracy": 0.987037037037037,
695
+ "eval_loss": 0.054442230612039566,
696
+ "eval_runtime": 16.2404,
697
+ "eval_samples_per_second": 166.252,
698
+ "eval_steps_per_second": 20.812,
699
+ "step": 1000
700
+ },
701
+ {
702
+ "epoch": 1.06,
703
+ "learning_rate": 0.00014728317659352141,
704
+ "loss": 0.1676,
705
+ "step": 1010
706
+ },
707
+ {
708
+ "epoch": 1.07,
709
+ "learning_rate": 0.000146760710553814,
710
+ "loss": 0.2244,
711
+ "step": 1020
712
+ },
713
+ {
714
+ "epoch": 1.08,
715
+ "learning_rate": 0.00014623824451410659,
716
+ "loss": 0.1305,
717
+ "step": 1030
718
+ },
719
+ {
720
+ "epoch": 1.09,
721
+ "learning_rate": 0.00014571577847439916,
722
+ "loss": 0.126,
723
+ "step": 1040
724
+ },
725
+ {
726
+ "epoch": 1.1,
727
+ "learning_rate": 0.00014519331243469173,
728
+ "loss": 0.1695,
729
+ "step": 1050
730
+ },
731
+ {
732
+ "epoch": 1.11,
733
+ "learning_rate": 0.00014467084639498433,
734
+ "loss": 0.2042,
735
+ "step": 1060
736
+ },
737
+ {
738
+ "epoch": 1.12,
739
+ "learning_rate": 0.0001441483803552769,
740
+ "loss": 0.1022,
741
+ "step": 1070
742
+ },
743
+ {
744
+ "epoch": 1.13,
745
+ "learning_rate": 0.0001436259143155695,
746
+ "loss": 0.1836,
747
+ "step": 1080
748
+ },
749
+ {
750
+ "epoch": 1.14,
751
+ "learning_rate": 0.0001431034482758621,
752
+ "loss": 0.1896,
753
+ "step": 1090
754
+ },
755
+ {
756
+ "epoch": 1.15,
757
+ "learning_rate": 0.00014258098223615467,
758
+ "loss": 0.3192,
759
+ "step": 1100
760
+ },
761
+ {
762
+ "epoch": 1.15,
763
+ "eval_accuracy": 0.9725925925925926,
764
+ "eval_loss": 0.08920213580131531,
765
+ "eval_runtime": 16.309,
766
+ "eval_samples_per_second": 165.552,
767
+ "eval_steps_per_second": 20.725,
768
+ "step": 1100
769
+ },
770
+ {
771
+ "epoch": 1.16,
772
+ "learning_rate": 0.00014205851619644724,
773
+ "loss": 0.0961,
774
+ "step": 1110
775
+ },
776
+ {
777
+ "epoch": 1.17,
778
+ "learning_rate": 0.0001415360501567398,
779
+ "loss": 0.0903,
780
+ "step": 1120
781
+ },
782
+ {
783
+ "epoch": 1.18,
784
+ "learning_rate": 0.0001410135841170324,
785
+ "loss": 0.133,
786
+ "step": 1130
787
+ },
788
+ {
789
+ "epoch": 1.19,
790
+ "learning_rate": 0.00014049111807732498,
791
+ "loss": 0.1073,
792
+ "step": 1140
793
+ },
794
+ {
795
+ "epoch": 1.2,
796
+ "learning_rate": 0.00013996865203761756,
797
+ "loss": 0.1116,
798
+ "step": 1150
799
+ },
800
+ {
801
+ "epoch": 1.21,
802
+ "learning_rate": 0.00013944618599791015,
803
+ "loss": 0.1372,
804
+ "step": 1160
805
+ },
806
+ {
807
+ "epoch": 1.22,
808
+ "learning_rate": 0.00013892371995820273,
809
+ "loss": 0.1428,
810
+ "step": 1170
811
+ },
812
+ {
813
+ "epoch": 1.23,
814
+ "learning_rate": 0.0001384012539184953,
815
+ "loss": 0.1059,
816
+ "step": 1180
817
+ },
818
+ {
819
+ "epoch": 1.24,
820
+ "learning_rate": 0.0001378787878787879,
821
+ "loss": 0.0891,
822
+ "step": 1190
823
+ },
824
+ {
825
+ "epoch": 1.25,
826
+ "learning_rate": 0.00013735632183908047,
827
+ "loss": 0.158,
828
+ "step": 1200
829
+ },
830
+ {
831
+ "epoch": 1.25,
832
+ "eval_accuracy": 0.98,
833
+ "eval_loss": 0.06320571154356003,
834
+ "eval_runtime": 16.4346,
835
+ "eval_samples_per_second": 164.287,
836
+ "eval_steps_per_second": 20.566,
837
+ "step": 1200
838
+ },
839
+ {
840
+ "epoch": 1.26,
841
+ "learning_rate": 0.00013683385579937304,
842
+ "loss": 0.0856,
843
+ "step": 1210
844
+ },
845
+ {
846
+ "epoch": 1.27,
847
+ "learning_rate": 0.0001363113897596656,
848
+ "loss": 0.1741,
849
+ "step": 1220
850
+ },
851
+ {
852
+ "epoch": 1.29,
853
+ "learning_rate": 0.0001357889237199582,
854
+ "loss": 0.1105,
855
+ "step": 1230
856
+ },
857
+ {
858
+ "epoch": 1.3,
859
+ "learning_rate": 0.00013526645768025078,
860
+ "loss": 0.2382,
861
+ "step": 1240
862
+ },
863
+ {
864
+ "epoch": 1.31,
865
+ "learning_rate": 0.00013474399164054336,
866
+ "loss": 0.1309,
867
+ "step": 1250
868
+ },
869
+ {
870
+ "epoch": 1.32,
871
+ "learning_rate": 0.00013422152560083595,
872
+ "loss": 0.0947,
873
+ "step": 1260
874
+ },
875
+ {
876
+ "epoch": 1.33,
877
+ "learning_rate": 0.00013369905956112853,
878
+ "loss": 0.12,
879
+ "step": 1270
880
+ },
881
+ {
882
+ "epoch": 1.34,
883
+ "learning_rate": 0.00013317659352142112,
884
+ "loss": 0.0805,
885
+ "step": 1280
886
+ },
887
+ {
888
+ "epoch": 1.35,
889
+ "learning_rate": 0.0001326541274817137,
890
+ "loss": 0.1211,
891
+ "step": 1290
892
+ },
893
+ {
894
+ "epoch": 1.36,
895
+ "learning_rate": 0.0001321316614420063,
896
+ "loss": 0.0266,
897
+ "step": 1300
898
+ },
899
+ {
900
+ "epoch": 1.36,
901
+ "eval_accuracy": 0.9944444444444445,
902
+ "eval_loss": 0.02332974411547184,
903
+ "eval_runtime": 16.2062,
904
+ "eval_samples_per_second": 166.603,
905
+ "eval_steps_per_second": 20.856,
906
+ "step": 1300
907
+ },
908
+ {
909
+ "epoch": 1.37,
910
+ "learning_rate": 0.00013160919540229887,
911
+ "loss": 0.1716,
912
+ "step": 1310
913
+ },
914
+ {
915
+ "epoch": 1.38,
916
+ "learning_rate": 0.00013108672936259144,
917
+ "loss": 0.1854,
918
+ "step": 1320
919
+ },
920
+ {
921
+ "epoch": 1.39,
922
+ "learning_rate": 0.00013056426332288404,
923
+ "loss": 0.0916,
924
+ "step": 1330
925
+ },
926
+ {
927
+ "epoch": 1.4,
928
+ "learning_rate": 0.0001300417972831766,
929
+ "loss": 0.0378,
930
+ "step": 1340
931
+ },
932
+ {
933
+ "epoch": 1.41,
934
+ "learning_rate": 0.00012951933124346918,
935
+ "loss": 0.0995,
936
+ "step": 1350
937
+ },
938
+ {
939
+ "epoch": 1.42,
940
+ "learning_rate": 0.00012899686520376175,
941
+ "loss": 0.0803,
942
+ "step": 1360
943
+ },
944
+ {
945
+ "epoch": 1.43,
946
+ "learning_rate": 0.00012847439916405435,
947
+ "loss": 0.0585,
948
+ "step": 1370
949
+ },
950
+ {
951
+ "epoch": 1.44,
952
+ "learning_rate": 0.00012795193312434692,
953
+ "loss": 0.1277,
954
+ "step": 1380
955
+ },
956
+ {
957
+ "epoch": 1.45,
958
+ "learning_rate": 0.0001274294670846395,
959
+ "loss": 0.0938,
960
+ "step": 1390
961
+ },
962
+ {
963
+ "epoch": 1.46,
964
+ "learning_rate": 0.0001269070010449321,
965
+ "loss": 0.1832,
966
+ "step": 1400
967
+ },
968
+ {
969
+ "epoch": 1.46,
970
+ "eval_accuracy": 0.9929629629629629,
971
+ "eval_loss": 0.029181810095906258,
972
+ "eval_runtime": 16.5252,
973
+ "eval_samples_per_second": 163.387,
974
+ "eval_steps_per_second": 20.454,
975
+ "step": 1400
976
+ },
977
+ {
978
+ "epoch": 1.47,
979
+ "learning_rate": 0.00012638453500522467,
980
+ "loss": 0.1317,
981
+ "step": 1410
982
+ },
983
+ {
984
+ "epoch": 1.48,
985
+ "learning_rate": 0.00012586206896551724,
986
+ "loss": 0.176,
987
+ "step": 1420
988
+ },
989
+ {
990
+ "epoch": 1.49,
991
+ "learning_rate": 0.0001253396029258098,
992
+ "loss": 0.0925,
993
+ "step": 1430
994
+ },
995
+ {
996
+ "epoch": 1.5,
997
+ "learning_rate": 0.0001248171368861024,
998
+ "loss": 0.159,
999
+ "step": 1440
1000
+ },
1001
+ {
1002
+ "epoch": 1.52,
1003
+ "learning_rate": 0.00012429467084639498,
1004
+ "loss": 0.1374,
1005
+ "step": 1450
1006
+ },
1007
+ {
1008
+ "epoch": 1.53,
1009
+ "learning_rate": 0.00012377220480668755,
1010
+ "loss": 0.2189,
1011
+ "step": 1460
1012
+ },
1013
+ {
1014
+ "epoch": 1.54,
1015
+ "learning_rate": 0.00012324973876698015,
1016
+ "loss": 0.1034,
1017
+ "step": 1470
1018
+ },
1019
+ {
1020
+ "epoch": 1.55,
1021
+ "learning_rate": 0.00012272727272727272,
1022
+ "loss": 0.0965,
1023
+ "step": 1480
1024
+ },
1025
+ {
1026
+ "epoch": 1.56,
1027
+ "learning_rate": 0.00012220480668756532,
1028
+ "loss": 0.2171,
1029
+ "step": 1490
1030
+ },
1031
+ {
1032
+ "epoch": 1.57,
1033
+ "learning_rate": 0.00012168234064785788,
1034
+ "loss": 0.1212,
1035
+ "step": 1500
1036
+ },
1037
+ {
1038
+ "epoch": 1.57,
1039
+ "eval_accuracy": 0.9851851851851852,
1040
+ "eval_loss": 0.04892929643392563,
1041
+ "eval_runtime": 16.2659,
1042
+ "eval_samples_per_second": 165.991,
1043
+ "eval_steps_per_second": 20.78,
1044
+ "step": 1500
1045
+ },
1046
+ {
1047
+ "epoch": 1.58,
1048
+ "learning_rate": 0.00012115987460815048,
1049
+ "loss": 0.125,
1050
+ "step": 1510
1051
+ },
1052
+ {
1053
+ "epoch": 1.59,
1054
+ "learning_rate": 0.00012063740856844305,
1055
+ "loss": 0.0725,
1056
+ "step": 1520
1057
+ },
1058
+ {
1059
+ "epoch": 1.6,
1060
+ "learning_rate": 0.00012011494252873562,
1061
+ "loss": 0.149,
1062
+ "step": 1530
1063
+ },
1064
+ {
1065
+ "epoch": 1.61,
1066
+ "learning_rate": 0.00011959247648902822,
1067
+ "loss": 0.1072,
1068
+ "step": 1540
1069
+ },
1070
+ {
1071
+ "epoch": 1.62,
1072
+ "learning_rate": 0.00011907001044932081,
1073
+ "loss": 0.0786,
1074
+ "step": 1550
1075
+ },
1076
+ {
1077
+ "epoch": 1.63,
1078
+ "learning_rate": 0.00011854754440961338,
1079
+ "loss": 0.1061,
1080
+ "step": 1560
1081
+ },
1082
+ {
1083
+ "epoch": 1.64,
1084
+ "learning_rate": 0.00011802507836990598,
1085
+ "loss": 0.26,
1086
+ "step": 1570
1087
+ },
1088
+ {
1089
+ "epoch": 1.65,
1090
+ "learning_rate": 0.00011750261233019855,
1091
+ "loss": 0.0717,
1092
+ "step": 1580
1093
+ },
1094
+ {
1095
+ "epoch": 1.66,
1096
+ "learning_rate": 0.00011698014629049112,
1097
+ "loss": 0.1236,
1098
+ "step": 1590
1099
+ },
1100
+ {
1101
+ "epoch": 1.67,
1102
+ "learning_rate": 0.0001164576802507837,
1103
+ "loss": 0.0994,
1104
+ "step": 1600
1105
+ },
1106
+ {
1107
+ "epoch": 1.67,
1108
+ "eval_accuracy": 0.9974074074074074,
1109
+ "eval_loss": 0.014160659164190292,
1110
+ "eval_runtime": 16.4275,
1111
+ "eval_samples_per_second": 164.358,
1112
+ "eval_steps_per_second": 20.575,
1113
+ "step": 1600
1114
+ },
1115
+ {
1116
+ "epoch": 1.68,
1117
+ "learning_rate": 0.00011593521421107629,
1118
+ "loss": 0.1335,
1119
+ "step": 1610
1120
+ },
1121
+ {
1122
+ "epoch": 1.69,
1123
+ "learning_rate": 0.00011541274817136886,
1124
+ "loss": 0.1385,
1125
+ "step": 1620
1126
+ },
1127
+ {
1128
+ "epoch": 1.7,
1129
+ "learning_rate": 0.00011489028213166144,
1130
+ "loss": 0.0761,
1131
+ "step": 1630
1132
+ },
1133
+ {
1134
+ "epoch": 1.71,
1135
+ "learning_rate": 0.00011436781609195404,
1136
+ "loss": 0.1571,
1137
+ "step": 1640
1138
+ },
1139
+ {
1140
+ "epoch": 1.72,
1141
+ "learning_rate": 0.00011384535005224661,
1142
+ "loss": 0.046,
1143
+ "step": 1650
1144
+ },
1145
+ {
1146
+ "epoch": 1.73,
1147
+ "learning_rate": 0.00011332288401253919,
1148
+ "loss": 0.1345,
1149
+ "step": 1660
1150
+ },
1151
+ {
1152
+ "epoch": 1.75,
1153
+ "learning_rate": 0.00011280041797283176,
1154
+ "loss": 0.0945,
1155
+ "step": 1670
1156
+ },
1157
+ {
1158
+ "epoch": 1.76,
1159
+ "learning_rate": 0.00011227795193312436,
1160
+ "loss": 0.1222,
1161
+ "step": 1680
1162
+ },
1163
+ {
1164
+ "epoch": 1.77,
1165
+ "learning_rate": 0.00011175548589341694,
1166
+ "loss": 0.1118,
1167
+ "step": 1690
1168
+ },
1169
+ {
1170
+ "epoch": 1.78,
1171
+ "learning_rate": 0.00011123301985370951,
1172
+ "loss": 0.0219,
1173
+ "step": 1700
1174
+ },
1175
+ {
1176
+ "epoch": 1.78,
1177
+ "eval_accuracy": 0.9929629629629629,
1178
+ "eval_loss": 0.02774985507130623,
1179
+ "eval_runtime": 16.27,
1180
+ "eval_samples_per_second": 165.949,
1181
+ "eval_steps_per_second": 20.774,
1182
+ "step": 1700
1183
+ },
1184
+ {
1185
+ "epoch": 1.79,
1186
+ "learning_rate": 0.0001107105538140021,
1187
+ "loss": 0.1182,
1188
+ "step": 1710
1189
+ },
1190
+ {
1191
+ "epoch": 1.8,
1192
+ "learning_rate": 0.00011018808777429468,
1193
+ "loss": 0.0108,
1194
+ "step": 1720
1195
+ },
1196
+ {
1197
+ "epoch": 1.81,
1198
+ "learning_rate": 0.00010966562173458725,
1199
+ "loss": 0.0983,
1200
+ "step": 1730
1201
+ },
1202
+ {
1203
+ "epoch": 1.82,
1204
+ "learning_rate": 0.00010914315569487982,
1205
+ "loss": 0.0752,
1206
+ "step": 1740
1207
+ },
1208
+ {
1209
+ "epoch": 1.83,
1210
+ "learning_rate": 0.00010862068965517242,
1211
+ "loss": 0.0566,
1212
+ "step": 1750
1213
+ },
1214
+ {
1215
+ "epoch": 1.84,
1216
+ "learning_rate": 0.000108098223615465,
1217
+ "loss": 0.088,
1218
+ "step": 1760
1219
+ },
1220
+ {
1221
+ "epoch": 1.85,
1222
+ "learning_rate": 0.00010757575757575758,
1223
+ "loss": 0.1686,
1224
+ "step": 1770
1225
+ },
1226
+ {
1227
+ "epoch": 1.86,
1228
+ "learning_rate": 0.00010705329153605018,
1229
+ "loss": 0.0176,
1230
+ "step": 1780
1231
+ },
1232
+ {
1233
+ "epoch": 1.87,
1234
+ "learning_rate": 0.00010653082549634275,
1235
+ "loss": 0.0688,
1236
+ "step": 1790
1237
+ },
1238
+ {
1239
+ "epoch": 1.88,
1240
+ "learning_rate": 0.00010600835945663532,
1241
+ "loss": 0.0664,
1242
+ "step": 1800
1243
+ },
1244
+ {
1245
+ "epoch": 1.88,
1246
+ "eval_accuracy": 0.9974074074074074,
1247
+ "eval_loss": 0.015811724588274956,
1248
+ "eval_runtime": 16.2257,
1249
+ "eval_samples_per_second": 166.402,
1250
+ "eval_steps_per_second": 20.831,
1251
+ "step": 1800
1252
+ },
1253
+ {
1254
+ "epoch": 1.89,
1255
+ "learning_rate": 0.00010548589341692789,
1256
+ "loss": 0.1437,
1257
+ "step": 1810
1258
+ },
1259
+ {
1260
+ "epoch": 1.9,
1261
+ "learning_rate": 0.00010496342737722049,
1262
+ "loss": 0.0091,
1263
+ "step": 1820
1264
+ },
1265
+ {
1266
+ "epoch": 1.91,
1267
+ "learning_rate": 0.00010444096133751306,
1268
+ "loss": 0.109,
1269
+ "step": 1830
1270
+ },
1271
+ {
1272
+ "epoch": 1.92,
1273
+ "learning_rate": 0.00010391849529780563,
1274
+ "loss": 0.0641,
1275
+ "step": 1840
1276
+ },
1277
+ {
1278
+ "epoch": 1.93,
1279
+ "learning_rate": 0.00010339602925809823,
1280
+ "loss": 0.112,
1281
+ "step": 1850
1282
+ },
1283
+ {
1284
+ "epoch": 1.94,
1285
+ "learning_rate": 0.0001028735632183908,
1286
+ "loss": 0.0732,
1287
+ "step": 1860
1288
+ },
1289
+ {
1290
+ "epoch": 1.95,
1291
+ "learning_rate": 0.00010235109717868339,
1292
+ "loss": 0.1322,
1293
+ "step": 1870
1294
+ },
1295
+ {
1296
+ "epoch": 1.96,
1297
+ "learning_rate": 0.00010182863113897596,
1298
+ "loss": 0.046,
1299
+ "step": 1880
1300
+ },
1301
+ {
1302
+ "epoch": 1.97,
1303
+ "learning_rate": 0.00010130616509926856,
1304
+ "loss": 0.1062,
1305
+ "step": 1890
1306
+ },
1307
+ {
1308
+ "epoch": 1.99,
1309
+ "learning_rate": 0.00010078369905956113,
1310
+ "loss": 0.0834,
1311
+ "step": 1900
1312
+ },
1313
+ {
1314
+ "epoch": 1.99,
1315
+ "eval_accuracy": 0.9977777777777778,
1316
+ "eval_loss": 0.012378275394439697,
1317
+ "eval_runtime": 16.3569,
1318
+ "eval_samples_per_second": 165.068,
1319
+ "eval_steps_per_second": 20.664,
1320
+ "step": 1900
1321
+ },
1322
+ {
1323
+ "epoch": 2.0,
1324
+ "learning_rate": 0.0001002612330198537,
1325
+ "loss": 0.0779,
1326
+ "step": 1910
1327
+ },
1328
+ {
1329
+ "epoch": 2.01,
1330
+ "learning_rate": 9.973876698014629e-05,
1331
+ "loss": 0.061,
1332
+ "step": 1920
1333
+ },
1334
+ {
1335
+ "epoch": 2.02,
1336
+ "learning_rate": 9.921630094043888e-05,
1337
+ "loss": 0.0608,
1338
+ "step": 1930
1339
+ },
1340
+ {
1341
+ "epoch": 2.03,
1342
+ "learning_rate": 9.869383490073145e-05,
1343
+ "loss": 0.0975,
1344
+ "step": 1940
1345
+ },
1346
+ {
1347
+ "epoch": 2.04,
1348
+ "learning_rate": 9.817136886102403e-05,
1349
+ "loss": 0.0532,
1350
+ "step": 1950
1351
+ },
1352
+ {
1353
+ "epoch": 2.05,
1354
+ "learning_rate": 9.764890282131662e-05,
1355
+ "loss": 0.1071,
1356
+ "step": 1960
1357
+ },
1358
+ {
1359
+ "epoch": 2.06,
1360
+ "learning_rate": 9.71264367816092e-05,
1361
+ "loss": 0.0721,
1362
+ "step": 1970
1363
+ },
1364
+ {
1365
+ "epoch": 2.07,
1366
+ "learning_rate": 9.660397074190179e-05,
1367
+ "loss": 0.1187,
1368
+ "step": 1980
1369
+ },
1370
+ {
1371
+ "epoch": 2.08,
1372
+ "learning_rate": 9.608150470219436e-05,
1373
+ "loss": 0.0543,
1374
+ "step": 1990
1375
+ },
1376
+ {
1377
+ "epoch": 2.09,
1378
+ "learning_rate": 9.555903866248695e-05,
1379
+ "loss": 0.1093,
1380
+ "step": 2000
1381
+ },
1382
+ {
1383
+ "epoch": 2.09,
1384
+ "eval_accuracy": 0.9974074074074074,
1385
+ "eval_loss": 0.013967631384730339,
1386
+ "eval_runtime": 16.4607,
1387
+ "eval_samples_per_second": 164.027,
1388
+ "eval_steps_per_second": 20.534,
1389
+ "step": 2000
1390
+ },
1391
+ {
1392
+ "epoch": 2.1,
1393
+ "learning_rate": 9.503657262277952e-05,
1394
+ "loss": 0.0528,
1395
+ "step": 2010
1396
+ },
1397
+ {
1398
+ "epoch": 2.11,
1399
+ "learning_rate": 9.45141065830721e-05,
1400
+ "loss": 0.0391,
1401
+ "step": 2020
1402
+ },
1403
+ {
1404
+ "epoch": 2.12,
1405
+ "learning_rate": 9.399164054336469e-05,
1406
+ "loss": 0.0918,
1407
+ "step": 2030
1408
+ },
1409
+ {
1410
+ "epoch": 2.13,
1411
+ "learning_rate": 9.346917450365726e-05,
1412
+ "loss": 0.0567,
1413
+ "step": 2040
1414
+ },
1415
+ {
1416
+ "epoch": 2.14,
1417
+ "learning_rate": 9.294670846394985e-05,
1418
+ "loss": 0.0128,
1419
+ "step": 2050
1420
+ },
1421
+ {
1422
+ "epoch": 2.15,
1423
+ "learning_rate": 9.242424242424242e-05,
1424
+ "loss": 0.0776,
1425
+ "step": 2060
1426
+ },
1427
+ {
1428
+ "epoch": 2.16,
1429
+ "learning_rate": 9.1901776384535e-05,
1430
+ "loss": 0.0597,
1431
+ "step": 2070
1432
+ },
1433
+ {
1434
+ "epoch": 2.17,
1435
+ "learning_rate": 9.137931034482759e-05,
1436
+ "loss": 0.0866,
1437
+ "step": 2080
1438
+ },
1439
+ {
1440
+ "epoch": 2.18,
1441
+ "learning_rate": 9.085684430512017e-05,
1442
+ "loss": 0.1428,
1443
+ "step": 2090
1444
+ },
1445
+ {
1446
+ "epoch": 2.19,
1447
+ "learning_rate": 9.033437826541276e-05,
1448
+ "loss": 0.1726,
1449
+ "step": 2100
1450
+ },
1451
+ {
1452
+ "epoch": 2.19,
1453
+ "eval_accuracy": 0.9962962962962963,
1454
+ "eval_loss": 0.01471767295151949,
1455
+ "eval_runtime": 16.5946,
1456
+ "eval_samples_per_second": 162.703,
1457
+ "eval_steps_per_second": 20.368,
1458
+ "step": 2100
1459
+ },
1460
+ {
1461
+ "epoch": 2.2,
1462
+ "learning_rate": 8.981191222570533e-05,
1463
+ "loss": 0.0762,
1464
+ "step": 2110
1465
+ },
1466
+ {
1467
+ "epoch": 2.22,
1468
+ "learning_rate": 8.928944618599792e-05,
1469
+ "loss": 0.0663,
1470
+ "step": 2120
1471
+ },
1472
+ {
1473
+ "epoch": 2.23,
1474
+ "learning_rate": 8.876698014629049e-05,
1475
+ "loss": 0.0743,
1476
+ "step": 2130
1477
+ },
1478
+ {
1479
+ "epoch": 2.24,
1480
+ "learning_rate": 8.824451410658307e-05,
1481
+ "loss": 0.0549,
1482
+ "step": 2140
1483
+ },
1484
+ {
1485
+ "epoch": 2.25,
1486
+ "learning_rate": 8.772204806687566e-05,
1487
+ "loss": 0.0409,
1488
+ "step": 2150
1489
+ },
1490
+ {
1491
+ "epoch": 2.26,
1492
+ "learning_rate": 8.719958202716823e-05,
1493
+ "loss": 0.0222,
1494
+ "step": 2160
1495
+ },
1496
+ {
1497
+ "epoch": 2.27,
1498
+ "learning_rate": 8.667711598746082e-05,
1499
+ "loss": 0.0481,
1500
+ "step": 2170
1501
+ },
1502
+ {
1503
+ "epoch": 2.28,
1504
+ "learning_rate": 8.61546499477534e-05,
1505
+ "loss": 0.0117,
1506
+ "step": 2180
1507
+ },
1508
+ {
1509
+ "epoch": 2.29,
1510
+ "learning_rate": 8.563218390804599e-05,
1511
+ "loss": 0.0415,
1512
+ "step": 2190
1513
+ },
1514
+ {
1515
+ "epoch": 2.3,
1516
+ "learning_rate": 8.510971786833856e-05,
1517
+ "loss": 0.0476,
1518
+ "step": 2200
1519
+ },
1520
+ {
1521
+ "epoch": 2.3,
1522
+ "eval_accuracy": 0.9992592592592593,
1523
+ "eval_loss": 0.005821824073791504,
1524
+ "eval_runtime": 16.2815,
1525
+ "eval_samples_per_second": 165.832,
1526
+ "eval_steps_per_second": 20.76,
1527
+ "step": 2200
1528
+ },
1529
+ {
1530
+ "epoch": 2.31,
1531
+ "learning_rate": 8.458725182863114e-05,
1532
+ "loss": 0.005,
1533
+ "step": 2210
1534
+ },
1535
+ {
1536
+ "epoch": 2.32,
1537
+ "learning_rate": 8.406478578892373e-05,
1538
+ "loss": 0.0456,
1539
+ "step": 2220
1540
+ },
1541
+ {
1542
+ "epoch": 2.33,
1543
+ "learning_rate": 8.35423197492163e-05,
1544
+ "loss": 0.1262,
1545
+ "step": 2230
1546
+ },
1547
+ {
1548
+ "epoch": 2.34,
1549
+ "learning_rate": 8.301985370950889e-05,
1550
+ "loss": 0.0189,
1551
+ "step": 2240
1552
+ },
1553
+ {
1554
+ "epoch": 2.35,
1555
+ "learning_rate": 8.249738766980146e-05,
1556
+ "loss": 0.1614,
1557
+ "step": 2250
1558
+ },
1559
+ {
1560
+ "epoch": 2.36,
1561
+ "learning_rate": 8.197492163009404e-05,
1562
+ "loss": 0.1015,
1563
+ "step": 2260
1564
+ },
1565
+ {
1566
+ "epoch": 2.37,
1567
+ "learning_rate": 8.145245559038663e-05,
1568
+ "loss": 0.0352,
1569
+ "step": 2270
1570
+ },
1571
+ {
1572
+ "epoch": 2.38,
1573
+ "learning_rate": 8.09299895506792e-05,
1574
+ "loss": 0.0751,
1575
+ "step": 2280
1576
+ },
1577
+ {
1578
+ "epoch": 2.39,
1579
+ "learning_rate": 8.04075235109718e-05,
1580
+ "loss": 0.0369,
1581
+ "step": 2290
1582
+ },
1583
+ {
1584
+ "epoch": 2.4,
1585
+ "learning_rate": 7.988505747126437e-05,
1586
+ "loss": 0.0257,
1587
+ "step": 2300
1588
+ },
1589
+ {
1590
+ "epoch": 2.4,
1591
+ "eval_accuracy": 0.9911111111111112,
1592
+ "eval_loss": 0.04240882024168968,
1593
+ "eval_runtime": 16.3138,
1594
+ "eval_samples_per_second": 165.504,
1595
+ "eval_steps_per_second": 20.719,
1596
+ "step": 2300
1597
+ },
1598
+ {
1599
+ "epoch": 2.41,
1600
+ "learning_rate": 7.936259143155696e-05,
1601
+ "loss": 0.1677,
1602
+ "step": 2310
1603
+ },
1604
+ {
1605
+ "epoch": 2.42,
1606
+ "learning_rate": 7.884012539184953e-05,
1607
+ "loss": 0.0875,
1608
+ "step": 2320
1609
+ },
1610
+ {
1611
+ "epoch": 2.43,
1612
+ "learning_rate": 7.831765935214211e-05,
1613
+ "loss": 0.0303,
1614
+ "step": 2330
1615
+ },
1616
+ {
1617
+ "epoch": 2.45,
1618
+ "learning_rate": 7.77951933124347e-05,
1619
+ "loss": 0.0992,
1620
+ "step": 2340
1621
+ },
1622
+ {
1623
+ "epoch": 2.46,
1624
+ "learning_rate": 7.727272727272727e-05,
1625
+ "loss": 0.0388,
1626
+ "step": 2350
1627
+ },
1628
+ {
1629
+ "epoch": 2.47,
1630
+ "learning_rate": 7.675026123301986e-05,
1631
+ "loss": 0.1306,
1632
+ "step": 2360
1633
+ },
1634
+ {
1635
+ "epoch": 2.48,
1636
+ "learning_rate": 7.622779519331243e-05,
1637
+ "loss": 0.0915,
1638
+ "step": 2370
1639
+ },
1640
+ {
1641
+ "epoch": 2.49,
1642
+ "learning_rate": 7.570532915360501e-05,
1643
+ "loss": 0.039,
1644
+ "step": 2380
1645
+ },
1646
+ {
1647
+ "epoch": 2.5,
1648
+ "learning_rate": 7.51828631138976e-05,
1649
+ "loss": 0.0623,
1650
+ "step": 2390
1651
+ },
1652
+ {
1653
+ "epoch": 2.51,
1654
+ "learning_rate": 7.466039707419019e-05,
1655
+ "loss": 0.0215,
1656
+ "step": 2400
1657
+ },
1658
+ {
1659
+ "epoch": 2.51,
1660
+ "eval_accuracy": 0.9988888888888889,
1661
+ "eval_loss": 0.007595046889036894,
1662
+ "eval_runtime": 16.5367,
1663
+ "eval_samples_per_second": 163.274,
1664
+ "eval_steps_per_second": 20.439,
1665
+ "step": 2400
1666
+ },
1667
+ {
1668
+ "epoch": 2.52,
1669
+ "learning_rate": 7.413793103448277e-05,
1670
+ "loss": 0.0512,
1671
+ "step": 2410
1672
+ },
1673
+ {
1674
+ "epoch": 2.53,
1675
+ "learning_rate": 7.361546499477534e-05,
1676
+ "loss": 0.0809,
1677
+ "step": 2420
1678
+ },
1679
+ {
1680
+ "epoch": 2.54,
1681
+ "learning_rate": 7.309299895506793e-05,
1682
+ "loss": 0.0449,
1683
+ "step": 2430
1684
+ },
1685
+ {
1686
+ "epoch": 2.55,
1687
+ "learning_rate": 7.25705329153605e-05,
1688
+ "loss": 0.0824,
1689
+ "step": 2440
1690
+ },
1691
+ {
1692
+ "epoch": 2.56,
1693
+ "learning_rate": 7.204806687565309e-05,
1694
+ "loss": 0.0512,
1695
+ "step": 2450
1696
+ },
1697
+ {
1698
+ "epoch": 2.57,
1699
+ "learning_rate": 7.152560083594567e-05,
1700
+ "loss": 0.108,
1701
+ "step": 2460
1702
+ },
1703
+ {
1704
+ "epoch": 2.58,
1705
+ "learning_rate": 7.100313479623824e-05,
1706
+ "loss": 0.1016,
1707
+ "step": 2470
1708
+ },
1709
+ {
1710
+ "epoch": 2.59,
1711
+ "learning_rate": 7.048066875653083e-05,
1712
+ "loss": 0.0212,
1713
+ "step": 2480
1714
+ },
1715
+ {
1716
+ "epoch": 2.6,
1717
+ "learning_rate": 6.995820271682341e-05,
1718
+ "loss": 0.1275,
1719
+ "step": 2490
1720
+ },
1721
+ {
1722
+ "epoch": 2.61,
1723
+ "learning_rate": 6.9435736677116e-05,
1724
+ "loss": 0.0748,
1725
+ "step": 2500
1726
+ },
1727
+ {
1728
+ "epoch": 2.61,
1729
+ "eval_accuracy": 0.9974074074074074,
1730
+ "eval_loss": 0.00986641738563776,
1731
+ "eval_runtime": 16.1049,
1732
+ "eval_samples_per_second": 167.651,
1733
+ "eval_steps_per_second": 20.987,
1734
+ "step": 2500
1735
+ },
1736
+ {
1737
+ "epoch": 2.62,
1738
+ "learning_rate": 6.891327063740857e-05,
1739
+ "loss": 0.0733,
1740
+ "step": 2510
1741
+ },
1742
+ {
1743
+ "epoch": 2.63,
1744
+ "learning_rate": 6.839080459770116e-05,
1745
+ "loss": 0.0985,
1746
+ "step": 2520
1747
+ },
1748
+ {
1749
+ "epoch": 2.64,
1750
+ "learning_rate": 6.786833855799374e-05,
1751
+ "loss": 0.059,
1752
+ "step": 2530
1753
+ },
1754
+ {
1755
+ "epoch": 2.65,
1756
+ "learning_rate": 6.734587251828631e-05,
1757
+ "loss": 0.0151,
1758
+ "step": 2540
1759
+ },
1760
+ {
1761
+ "epoch": 2.66,
1762
+ "learning_rate": 6.68234064785789e-05,
1763
+ "loss": 0.0597,
1764
+ "step": 2550
1765
+ },
1766
+ {
1767
+ "epoch": 2.68,
1768
+ "learning_rate": 6.630094043887147e-05,
1769
+ "loss": 0.1076,
1770
+ "step": 2560
1771
+ },
1772
+ {
1773
+ "epoch": 2.69,
1774
+ "learning_rate": 6.577847439916406e-05,
1775
+ "loss": 0.007,
1776
+ "step": 2570
1777
+ },
1778
+ {
1779
+ "epoch": 2.7,
1780
+ "learning_rate": 6.525600835945663e-05,
1781
+ "loss": 0.0186,
1782
+ "step": 2580
1783
+ },
1784
+ {
1785
+ "epoch": 2.71,
1786
+ "learning_rate": 6.473354231974921e-05,
1787
+ "loss": 0.0184,
1788
+ "step": 2590
1789
+ },
1790
+ {
1791
+ "epoch": 2.72,
1792
+ "learning_rate": 6.42110762800418e-05,
1793
+ "loss": 0.0059,
1794
+ "step": 2600
1795
+ },
1796
+ {
1797
+ "epoch": 2.72,
1798
+ "eval_accuracy": 0.9992592592592593,
1799
+ "eval_loss": 0.005262466147542,
1800
+ "eval_runtime": 16.3428,
1801
+ "eval_samples_per_second": 165.211,
1802
+ "eval_steps_per_second": 20.682,
1803
+ "step": 2600
1804
+ },
1805
+ {
1806
+ "epoch": 2.73,
1807
+ "learning_rate": 6.368861024033438e-05,
1808
+ "loss": 0.0509,
1809
+ "step": 2610
1810
+ },
1811
+ {
1812
+ "epoch": 2.74,
1813
+ "learning_rate": 6.316614420062697e-05,
1814
+ "loss": 0.0477,
1815
+ "step": 2620
1816
+ },
1817
+ {
1818
+ "epoch": 2.75,
1819
+ "learning_rate": 6.264367816091954e-05,
1820
+ "loss": 0.0112,
1821
+ "step": 2630
1822
+ },
1823
+ {
1824
+ "epoch": 2.76,
1825
+ "learning_rate": 6.212121212121213e-05,
1826
+ "loss": 0.1193,
1827
+ "step": 2640
1828
+ },
1829
+ {
1830
+ "epoch": 2.77,
1831
+ "learning_rate": 6.159874608150471e-05,
1832
+ "loss": 0.1131,
1833
+ "step": 2650
1834
+ },
1835
+ {
1836
+ "epoch": 2.78,
1837
+ "learning_rate": 6.107628004179728e-05,
1838
+ "loss": 0.0899,
1839
+ "step": 2660
1840
+ },
1841
+ {
1842
+ "epoch": 2.79,
1843
+ "learning_rate": 6.055381400208987e-05,
1844
+ "loss": 0.0171,
1845
+ "step": 2670
1846
+ },
1847
+ {
1848
+ "epoch": 2.8,
1849
+ "learning_rate": 6.003134796238245e-05,
1850
+ "loss": 0.0834,
1851
+ "step": 2680
1852
+ },
1853
+ {
1854
+ "epoch": 2.81,
1855
+ "learning_rate": 5.950888192267503e-05,
1856
+ "loss": 0.078,
1857
+ "step": 2690
1858
+ },
1859
+ {
1860
+ "epoch": 2.82,
1861
+ "learning_rate": 5.8986415882967604e-05,
1862
+ "loss": 0.0527,
1863
+ "step": 2700
1864
+ },
1865
+ {
1866
+ "epoch": 2.82,
1867
+ "eval_accuracy": 0.9962962962962963,
1868
+ "eval_loss": 0.014925166964530945,
1869
+ "eval_runtime": 16.5271,
1870
+ "eval_samples_per_second": 163.369,
1871
+ "eval_steps_per_second": 20.451,
1872
+ "step": 2700
1873
+ },
1874
+ {
1875
+ "epoch": 2.83,
1876
+ "learning_rate": 5.846394984326019e-05,
1877
+ "loss": 0.0311,
1878
+ "step": 2710
1879
+ },
1880
+ {
1881
+ "epoch": 2.84,
1882
+ "learning_rate": 5.7941483803552775e-05,
1883
+ "loss": 0.0198,
1884
+ "step": 2720
1885
+ },
1886
+ {
1887
+ "epoch": 2.85,
1888
+ "learning_rate": 5.7419017763845354e-05,
1889
+ "loss": 0.0303,
1890
+ "step": 2730
1891
+ },
1892
+ {
1893
+ "epoch": 2.86,
1894
+ "learning_rate": 5.689655172413794e-05,
1895
+ "loss": 0.0612,
1896
+ "step": 2740
1897
+ },
1898
+ {
1899
+ "epoch": 2.87,
1900
+ "learning_rate": 5.637408568443051e-05,
1901
+ "loss": 0.0949,
1902
+ "step": 2750
1903
+ },
1904
+ {
1905
+ "epoch": 2.88,
1906
+ "learning_rate": 5.5851619644723096e-05,
1907
+ "loss": 0.0499,
1908
+ "step": 2760
1909
+ },
1910
+ {
1911
+ "epoch": 2.89,
1912
+ "learning_rate": 5.532915360501567e-05,
1913
+ "loss": 0.0596,
1914
+ "step": 2770
1915
+ },
1916
+ {
1917
+ "epoch": 2.9,
1918
+ "learning_rate": 5.4806687565308254e-05,
1919
+ "loss": 0.0293,
1920
+ "step": 2780
1921
+ },
1922
+ {
1923
+ "epoch": 2.92,
1924
+ "learning_rate": 5.4284221525600846e-05,
1925
+ "loss": 0.053,
1926
+ "step": 2790
1927
+ },
1928
+ {
1929
+ "epoch": 2.93,
1930
+ "learning_rate": 5.376175548589342e-05,
1931
+ "loss": 0.0203,
1932
+ "step": 2800
1933
+ },
1934
+ {
1935
+ "epoch": 2.93,
1936
+ "eval_accuracy": 0.9992592592592593,
1937
+ "eval_loss": 0.004079786594957113,
1938
+ "eval_runtime": 16.3325,
1939
+ "eval_samples_per_second": 165.315,
1940
+ "eval_steps_per_second": 20.695,
1941
+ "step": 2800
1942
+ },
1943
+ {
1944
+ "epoch": 2.94,
1945
+ "learning_rate": 5.3239289446186e-05,
1946
+ "loss": 0.0386,
1947
+ "step": 2810
1948
+ },
1949
+ {
1950
+ "epoch": 2.95,
1951
+ "learning_rate": 5.2716823406478575e-05,
1952
+ "loss": 0.0208,
1953
+ "step": 2820
1954
+ },
1955
+ {
1956
+ "epoch": 2.96,
1957
+ "learning_rate": 5.219435736677116e-05,
1958
+ "loss": 0.1105,
1959
+ "step": 2830
1960
+ },
1961
+ {
1962
+ "epoch": 2.97,
1963
+ "learning_rate": 5.1671891327063745e-05,
1964
+ "loss": 0.0827,
1965
+ "step": 2840
1966
+ },
1967
+ {
1968
+ "epoch": 2.98,
1969
+ "learning_rate": 5.1149425287356324e-05,
1970
+ "loss": 0.0249,
1971
+ "step": 2850
1972
+ },
1973
+ {
1974
+ "epoch": 2.99,
1975
+ "learning_rate": 5.062695924764891e-05,
1976
+ "loss": 0.0509,
1977
+ "step": 2860
1978
+ },
1979
+ {
1980
+ "epoch": 3.0,
1981
+ "learning_rate": 5.010449320794148e-05,
1982
+ "loss": 0.0324,
1983
+ "step": 2870
1984
+ },
1985
+ {
1986
+ "epoch": 3.01,
1987
+ "learning_rate": 4.958202716823407e-05,
1988
+ "loss": 0.0158,
1989
+ "step": 2880
1990
+ },
1991
+ {
1992
+ "epoch": 3.02,
1993
+ "learning_rate": 4.905956112852665e-05,
1994
+ "loss": 0.0029,
1995
+ "step": 2890
1996
+ },
1997
+ {
1998
+ "epoch": 3.03,
1999
+ "learning_rate": 4.853709508881923e-05,
2000
+ "loss": 0.0791,
2001
+ "step": 2900
2002
+ },
2003
+ {
2004
+ "epoch": 3.03,
2005
+ "eval_accuracy": 0.9988888888888889,
2006
+ "eval_loss": 0.0032932416070252657,
2007
+ "eval_runtime": 16.2412,
2008
+ "eval_samples_per_second": 166.244,
2009
+ "eval_steps_per_second": 20.811,
2010
+ "step": 2900
2011
+ },
2012
+ {
2013
+ "epoch": 3.04,
2014
+ "learning_rate": 4.801462904911181e-05,
2015
+ "loss": 0.0097,
2016
+ "step": 2910
2017
+ },
2018
+ {
2019
+ "epoch": 3.05,
2020
+ "learning_rate": 4.749216300940439e-05,
2021
+ "loss": 0.0335,
2022
+ "step": 2920
2023
+ },
2024
+ {
2025
+ "epoch": 3.06,
2026
+ "learning_rate": 4.696969696969697e-05,
2027
+ "loss": 0.0392,
2028
+ "step": 2930
2029
+ },
2030
+ {
2031
+ "epoch": 3.07,
2032
+ "learning_rate": 4.644723092998955e-05,
2033
+ "loss": 0.0222,
2034
+ "step": 2940
2035
+ },
2036
+ {
2037
+ "epoch": 3.08,
2038
+ "learning_rate": 4.592476489028214e-05,
2039
+ "loss": 0.059,
2040
+ "step": 2950
2041
+ },
2042
+ {
2043
+ "epoch": 3.09,
2044
+ "learning_rate": 4.5402298850574716e-05,
2045
+ "loss": 0.0029,
2046
+ "step": 2960
2047
+ },
2048
+ {
2049
+ "epoch": 3.1,
2050
+ "learning_rate": 4.4879832810867294e-05,
2051
+ "loss": 0.0562,
2052
+ "step": 2970
2053
+ },
2054
+ {
2055
+ "epoch": 3.11,
2056
+ "learning_rate": 4.435736677115987e-05,
2057
+ "loss": 0.0233,
2058
+ "step": 2980
2059
+ },
2060
+ {
2061
+ "epoch": 3.12,
2062
+ "learning_rate": 4.383490073145246e-05,
2063
+ "loss": 0.0073,
2064
+ "step": 2990
2065
+ },
2066
+ {
2067
+ "epoch": 3.13,
2068
+ "learning_rate": 4.3312434691745044e-05,
2069
+ "loss": 0.0389,
2070
+ "step": 3000
2071
+ },
2072
+ {
2073
+ "epoch": 3.13,
2074
+ "eval_accuracy": 0.9988888888888889,
2075
+ "eval_loss": 0.0033078393898904324,
2076
+ "eval_runtime": 16.4559,
2077
+ "eval_samples_per_second": 164.075,
2078
+ "eval_steps_per_second": 20.54,
2079
+ "step": 3000
2080
+ },
2081
+ {
2082
+ "epoch": 3.15,
2083
+ "learning_rate": 4.278996865203762e-05,
2084
+ "loss": 0.0057,
2085
+ "step": 3010
2086
+ },
2087
+ {
2088
+ "epoch": 3.16,
2089
+ "learning_rate": 4.22675026123302e-05,
2090
+ "loss": 0.0355,
2091
+ "step": 3020
2092
+ },
2093
+ {
2094
+ "epoch": 3.17,
2095
+ "learning_rate": 4.174503657262278e-05,
2096
+ "loss": 0.0169,
2097
+ "step": 3030
2098
+ },
2099
+ {
2100
+ "epoch": 3.18,
2101
+ "learning_rate": 4.122257053291536e-05,
2102
+ "loss": 0.0377,
2103
+ "step": 3040
2104
+ },
2105
+ {
2106
+ "epoch": 3.19,
2107
+ "learning_rate": 4.0700104493207943e-05,
2108
+ "loss": 0.038,
2109
+ "step": 3050
2110
+ },
2111
+ {
2112
+ "epoch": 3.2,
2113
+ "learning_rate": 4.017763845350053e-05,
2114
+ "loss": 0.0647,
2115
+ "step": 3060
2116
+ },
2117
+ {
2118
+ "epoch": 3.21,
2119
+ "learning_rate": 3.965517241379311e-05,
2120
+ "loss": 0.038,
2121
+ "step": 3070
2122
+ },
2123
+ {
2124
+ "epoch": 3.22,
2125
+ "learning_rate": 3.9132706374085686e-05,
2126
+ "loss": 0.0565,
2127
+ "step": 3080
2128
+ },
2129
+ {
2130
+ "epoch": 3.23,
2131
+ "learning_rate": 3.8610240334378265e-05,
2132
+ "loss": 0.0445,
2133
+ "step": 3090
2134
+ },
2135
+ {
2136
+ "epoch": 3.24,
2137
+ "learning_rate": 3.808777429467085e-05,
2138
+ "loss": 0.0459,
2139
+ "step": 3100
2140
+ },
2141
+ {
2142
+ "epoch": 3.24,
2143
+ "eval_accuracy": 0.9988888888888889,
2144
+ "eval_loss": 0.004428654909133911,
2145
+ "eval_runtime": 16.26,
2146
+ "eval_samples_per_second": 166.052,
2147
+ "eval_steps_per_second": 20.787,
2148
+ "step": 3100
2149
+ },
2150
+ {
2151
+ "epoch": 3.25,
2152
+ "learning_rate": 3.756530825496343e-05,
2153
+ "loss": 0.0411,
2154
+ "step": 3110
2155
+ },
2156
+ {
2157
+ "epoch": 3.26,
2158
+ "learning_rate": 3.7042842215256014e-05,
2159
+ "loss": 0.0815,
2160
+ "step": 3120
2161
+ },
2162
+ {
2163
+ "epoch": 3.27,
2164
+ "learning_rate": 3.652037617554859e-05,
2165
+ "loss": 0.0029,
2166
+ "step": 3130
2167
+ },
2168
+ {
2169
+ "epoch": 3.28,
2170
+ "learning_rate": 3.599791013584117e-05,
2171
+ "loss": 0.0208,
2172
+ "step": 3140
2173
+ },
2174
+ {
2175
+ "epoch": 3.29,
2176
+ "learning_rate": 3.547544409613375e-05,
2177
+ "loss": 0.039,
2178
+ "step": 3150
2179
+ },
2180
+ {
2181
+ "epoch": 3.3,
2182
+ "learning_rate": 3.4952978056426335e-05,
2183
+ "loss": 0.0322,
2184
+ "step": 3160
2185
+ },
2186
+ {
2187
+ "epoch": 3.31,
2188
+ "learning_rate": 3.4430512016718914e-05,
2189
+ "loss": 0.0486,
2190
+ "step": 3170
2191
+ },
2192
+ {
2193
+ "epoch": 3.32,
2194
+ "learning_rate": 3.390804597701149e-05,
2195
+ "loss": 0.0396,
2196
+ "step": 3180
2197
+ },
2198
+ {
2199
+ "epoch": 3.33,
2200
+ "learning_rate": 3.338557993730408e-05,
2201
+ "loss": 0.0591,
2202
+ "step": 3190
2203
+ },
2204
+ {
2205
+ "epoch": 3.34,
2206
+ "learning_rate": 3.2863113897596656e-05,
2207
+ "loss": 0.0276,
2208
+ "step": 3200
2209
+ },
2210
+ {
2211
+ "epoch": 3.34,
2212
+ "eval_accuracy": 0.9996296296296296,
2213
+ "eval_loss": 0.0030968771316111088,
2214
+ "eval_runtime": 16.225,
2215
+ "eval_samples_per_second": 166.409,
2216
+ "eval_steps_per_second": 20.832,
2217
+ "step": 3200
2218
+ },
2219
+ {
2220
+ "epoch": 3.35,
2221
+ "learning_rate": 3.234064785788924e-05,
2222
+ "loss": 0.0438,
2223
+ "step": 3210
2224
+ },
2225
+ {
2226
+ "epoch": 3.36,
2227
+ "learning_rate": 3.181818181818182e-05,
2228
+ "loss": 0.0394,
2229
+ "step": 3220
2230
+ },
2231
+ {
2232
+ "epoch": 3.38,
2233
+ "learning_rate": 3.12957157784744e-05,
2234
+ "loss": 0.0418,
2235
+ "step": 3230
2236
+ },
2237
+ {
2238
+ "epoch": 3.39,
2239
+ "learning_rate": 3.077324973876698e-05,
2240
+ "loss": 0.0278,
2241
+ "step": 3240
2242
+ },
2243
+ {
2244
+ "epoch": 3.4,
2245
+ "learning_rate": 3.0250783699059566e-05,
2246
+ "loss": 0.0578,
2247
+ "step": 3250
2248
+ },
2249
+ {
2250
+ "epoch": 3.41,
2251
+ "learning_rate": 2.9728317659352145e-05,
2252
+ "loss": 0.0365,
2253
+ "step": 3260
2254
+ },
2255
+ {
2256
+ "epoch": 3.42,
2257
+ "learning_rate": 2.9205851619644723e-05,
2258
+ "loss": 0.0528,
2259
+ "step": 3270
2260
+ },
2261
+ {
2262
+ "epoch": 3.43,
2263
+ "learning_rate": 2.8683385579937305e-05,
2264
+ "loss": 0.0597,
2265
+ "step": 3280
2266
+ },
2267
+ {
2268
+ "epoch": 3.44,
2269
+ "learning_rate": 2.8160919540229884e-05,
2270
+ "loss": 0.0318,
2271
+ "step": 3290
2272
+ },
2273
+ {
2274
+ "epoch": 3.45,
2275
+ "learning_rate": 2.7638453500522466e-05,
2276
+ "loss": 0.0139,
2277
+ "step": 3300
2278
+ },
2279
+ {
2280
+ "epoch": 3.45,
2281
+ "eval_accuracy": 0.9996296296296296,
2282
+ "eval_loss": 0.002758789574727416,
2283
+ "eval_runtime": 16.3192,
2284
+ "eval_samples_per_second": 165.449,
2285
+ "eval_steps_per_second": 20.712,
2286
+ "step": 3300
2287
+ },
2288
+ {
2289
+ "epoch": 3.46,
2290
+ "learning_rate": 2.711598746081505e-05,
2291
+ "loss": 0.0542,
2292
+ "step": 3310
2293
+ },
2294
+ {
2295
+ "epoch": 3.47,
2296
+ "learning_rate": 2.659352142110763e-05,
2297
+ "loss": 0.0132,
2298
+ "step": 3320
2299
+ },
2300
+ {
2301
+ "epoch": 3.48,
2302
+ "learning_rate": 2.6071055381400212e-05,
2303
+ "loss": 0.0025,
2304
+ "step": 3330
2305
+ },
2306
+ {
2307
+ "epoch": 3.49,
2308
+ "learning_rate": 2.554858934169279e-05,
2309
+ "loss": 0.0418,
2310
+ "step": 3340
2311
+ },
2312
+ {
2313
+ "epoch": 3.5,
2314
+ "learning_rate": 2.5026123301985373e-05,
2315
+ "loss": 0.0376,
2316
+ "step": 3350
2317
+ },
2318
+ {
2319
+ "epoch": 3.51,
2320
+ "learning_rate": 2.4503657262277955e-05,
2321
+ "loss": 0.0566,
2322
+ "step": 3360
2323
+ },
2324
+ {
2325
+ "epoch": 3.52,
2326
+ "learning_rate": 2.3981191222570533e-05,
2327
+ "loss": 0.1575,
2328
+ "step": 3370
2329
+ },
2330
+ {
2331
+ "epoch": 3.53,
2332
+ "learning_rate": 2.3458725182863115e-05,
2333
+ "loss": 0.0056,
2334
+ "step": 3380
2335
+ },
2336
+ {
2337
+ "epoch": 3.54,
2338
+ "learning_rate": 2.2936259143155697e-05,
2339
+ "loss": 0.0066,
2340
+ "step": 3390
2341
+ },
2342
+ {
2343
+ "epoch": 3.55,
2344
+ "learning_rate": 2.2413793103448276e-05,
2345
+ "loss": 0.0076,
2346
+ "step": 3400
2347
+ },
2348
+ {
2349
+ "epoch": 3.55,
2350
+ "eval_accuracy": 0.9985185185185185,
2351
+ "eval_loss": 0.005528000649064779,
2352
+ "eval_runtime": 16.457,
2353
+ "eval_samples_per_second": 164.064,
2354
+ "eval_steps_per_second": 20.538,
2355
+ "step": 3400
2356
+ },
2357
+ {
2358
+ "epoch": 3.56,
2359
+ "learning_rate": 2.1891327063740858e-05,
2360
+ "loss": 0.0064,
2361
+ "step": 3410
2362
+ },
2363
+ {
2364
+ "epoch": 3.57,
2365
+ "learning_rate": 2.1368861024033436e-05,
2366
+ "loss": 0.0376,
2367
+ "step": 3420
2368
+ },
2369
+ {
2370
+ "epoch": 3.58,
2371
+ "learning_rate": 2.0846394984326022e-05,
2372
+ "loss": 0.0471,
2373
+ "step": 3430
2374
+ },
2375
+ {
2376
+ "epoch": 3.59,
2377
+ "learning_rate": 2.03239289446186e-05,
2378
+ "loss": 0.0029,
2379
+ "step": 3440
2380
+ },
2381
+ {
2382
+ "epoch": 3.61,
2383
+ "learning_rate": 1.980146290491118e-05,
2384
+ "loss": 0.0406,
2385
+ "step": 3450
2386
+ },
2387
+ {
2388
+ "epoch": 3.62,
2389
+ "learning_rate": 1.9278996865203764e-05,
2390
+ "loss": 0.0174,
2391
+ "step": 3460
2392
+ },
2393
+ {
2394
+ "epoch": 3.63,
2395
+ "learning_rate": 1.8756530825496343e-05,
2396
+ "loss": 0.027,
2397
+ "step": 3470
2398
+ },
2399
+ {
2400
+ "epoch": 3.64,
2401
+ "learning_rate": 1.8234064785788925e-05,
2402
+ "loss": 0.0181,
2403
+ "step": 3480
2404
+ },
2405
+ {
2406
+ "epoch": 3.65,
2407
+ "learning_rate": 1.7711598746081507e-05,
2408
+ "loss": 0.0319,
2409
+ "step": 3490
2410
+ },
2411
+ {
2412
+ "epoch": 3.66,
2413
+ "learning_rate": 1.7189132706374085e-05,
2414
+ "loss": 0.0097,
2415
+ "step": 3500
2416
+ },
2417
+ {
2418
+ "epoch": 3.66,
2419
+ "eval_accuracy": 0.9996296296296296,
2420
+ "eval_loss": 0.0027309097349643707,
2421
+ "eval_runtime": 16.2262,
2422
+ "eval_samples_per_second": 166.398,
2423
+ "eval_steps_per_second": 20.831,
2424
+ "step": 3500
2425
+ },
2426
+ {
2427
+ "epoch": 3.67,
2428
+ "learning_rate": 1.6666666666666667e-05,
2429
+ "loss": 0.0221,
2430
+ "step": 3510
2431
+ },
2432
+ {
2433
+ "epoch": 3.68,
2434
+ "learning_rate": 1.614420062695925e-05,
2435
+ "loss": 0.0124,
2436
+ "step": 3520
2437
+ },
2438
+ {
2439
+ "epoch": 3.69,
2440
+ "learning_rate": 1.5621734587251828e-05,
2441
+ "loss": 0.0257,
2442
+ "step": 3530
2443
+ },
2444
+ {
2445
+ "epoch": 3.7,
2446
+ "learning_rate": 1.509926854754441e-05,
2447
+ "loss": 0.0794,
2448
+ "step": 3540
2449
+ },
2450
+ {
2451
+ "epoch": 3.71,
2452
+ "learning_rate": 1.4576802507836992e-05,
2453
+ "loss": 0.0382,
2454
+ "step": 3550
2455
+ },
2456
+ {
2457
+ "epoch": 3.72,
2458
+ "learning_rate": 1.4054336468129572e-05,
2459
+ "loss": 0.0441,
2460
+ "step": 3560
2461
+ },
2462
+ {
2463
+ "epoch": 3.73,
2464
+ "learning_rate": 1.3531870428422153e-05,
2465
+ "loss": 0.0057,
2466
+ "step": 3570
2467
+ },
2468
+ {
2469
+ "epoch": 3.74,
2470
+ "learning_rate": 1.3009404388714735e-05,
2471
+ "loss": 0.085,
2472
+ "step": 3580
2473
+ },
2474
+ {
2475
+ "epoch": 3.75,
2476
+ "learning_rate": 1.2486938349007315e-05,
2477
+ "loss": 0.0658,
2478
+ "step": 3590
2479
+ },
2480
+ {
2481
+ "epoch": 3.76,
2482
+ "learning_rate": 1.1964472309299895e-05,
2483
+ "loss": 0.0193,
2484
+ "step": 3600
2485
+ },
2486
+ {
2487
+ "epoch": 3.76,
2488
+ "eval_accuracy": 0.9996296296296296,
2489
+ "eval_loss": 0.002558160340413451,
2490
+ "eval_runtime": 16.2201,
2491
+ "eval_samples_per_second": 166.46,
2492
+ "eval_steps_per_second": 20.838,
2493
+ "step": 3600
2494
+ },
2495
+ {
2496
+ "epoch": 3.77,
2497
+ "learning_rate": 1.1442006269592477e-05,
2498
+ "loss": 0.0276,
2499
+ "step": 3610
2500
+ },
2501
+ {
2502
+ "epoch": 3.78,
2503
+ "learning_rate": 1.091954022988506e-05,
2504
+ "loss": 0.0265,
2505
+ "step": 3620
2506
+ },
2507
+ {
2508
+ "epoch": 3.79,
2509
+ "learning_rate": 1.039707419017764e-05,
2510
+ "loss": 0.0297,
2511
+ "step": 3630
2512
+ },
2513
+ {
2514
+ "epoch": 3.8,
2515
+ "learning_rate": 9.87460815047022e-06,
2516
+ "loss": 0.0288,
2517
+ "step": 3640
2518
+ },
2519
+ {
2520
+ "epoch": 3.81,
2521
+ "learning_rate": 9.352142110762802e-06,
2522
+ "loss": 0.0017,
2523
+ "step": 3650
2524
+ },
2525
+ {
2526
+ "epoch": 3.82,
2527
+ "learning_rate": 8.829676071055382e-06,
2528
+ "loss": 0.0258,
2529
+ "step": 3660
2530
+ },
2531
+ {
2532
+ "epoch": 3.83,
2533
+ "learning_rate": 8.307210031347962e-06,
2534
+ "loss": 0.0044,
2535
+ "step": 3670
2536
+ },
2537
+ {
2538
+ "epoch": 3.85,
2539
+ "learning_rate": 7.784743991640543e-06,
2540
+ "loss": 0.0484,
2541
+ "step": 3680
2542
+ },
2543
+ {
2544
+ "epoch": 3.86,
2545
+ "learning_rate": 7.262277951933125e-06,
2546
+ "loss": 0.0367,
2547
+ "step": 3690
2548
+ },
2549
+ {
2550
+ "epoch": 3.87,
2551
+ "learning_rate": 6.739811912225706e-06,
2552
+ "loss": 0.0471,
2553
+ "step": 3700
2554
+ },
2555
+ {
2556
+ "epoch": 3.87,
2557
+ "eval_accuracy": 0.9996296296296296,
2558
+ "eval_loss": 0.002674533985555172,
2559
+ "eval_runtime": 16.1924,
2560
+ "eval_samples_per_second": 166.745,
2561
+ "eval_steps_per_second": 20.874,
2562
+ "step": 3700
2563
+ },
2564
+ {
2565
+ "epoch": 3.88,
2566
+ "learning_rate": 6.217345872518287e-06,
2567
+ "loss": 0.0245,
2568
+ "step": 3710
2569
+ },
2570
+ {
2571
+ "epoch": 3.89,
2572
+ "learning_rate": 5.694879832810868e-06,
2573
+ "loss": 0.0392,
2574
+ "step": 3720
2575
+ },
2576
+ {
2577
+ "epoch": 3.9,
2578
+ "learning_rate": 5.172413793103448e-06,
2579
+ "loss": 0.0109,
2580
+ "step": 3730
2581
+ },
2582
+ {
2583
+ "epoch": 3.91,
2584
+ "learning_rate": 4.6499477533960295e-06,
2585
+ "loss": 0.0022,
2586
+ "step": 3740
2587
+ },
2588
+ {
2589
+ "epoch": 3.92,
2590
+ "learning_rate": 4.127481713688611e-06,
2591
+ "loss": 0.0765,
2592
+ "step": 3750
2593
+ },
2594
+ {
2595
+ "epoch": 3.93,
2596
+ "learning_rate": 3.6050156739811913e-06,
2597
+ "loss": 0.0029,
2598
+ "step": 3760
2599
+ },
2600
+ {
2601
+ "epoch": 3.94,
2602
+ "learning_rate": 3.0825496342737725e-06,
2603
+ "loss": 0.0236,
2604
+ "step": 3770
2605
+ },
2606
+ {
2607
+ "epoch": 3.95,
2608
+ "learning_rate": 2.560083594566353e-06,
2609
+ "loss": 0.0308,
2610
+ "step": 3780
2611
+ },
2612
+ {
2613
+ "epoch": 3.96,
2614
+ "learning_rate": 2.0376175548589343e-06,
2615
+ "loss": 0.0288,
2616
+ "step": 3790
2617
+ },
2618
+ {
2619
+ "epoch": 3.97,
2620
+ "learning_rate": 1.5151515151515152e-06,
2621
+ "loss": 0.0282,
2622
+ "step": 3800
2623
+ },
2624
+ {
2625
+ "epoch": 3.97,
2626
+ "eval_accuracy": 0.9996296296296296,
2627
+ "eval_loss": 0.0026795254088938236,
2628
+ "eval_runtime": 16.4275,
2629
+ "eval_samples_per_second": 164.358,
2630
+ "eval_steps_per_second": 20.575,
2631
+ "step": 3800
2632
+ },
2633
+ {
2634
+ "epoch": 3.98,
2635
+ "learning_rate": 9.926854754440962e-07,
2636
+ "loss": 0.0188,
2637
+ "step": 3810
2638
+ },
2639
+ {
2640
+ "epoch": 3.99,
2641
+ "learning_rate": 4.7021943573667715e-07,
2642
+ "loss": 0.0323,
2643
+ "step": 3820
2644
+ },
2645
+ {
2646
+ "epoch": 4.0,
2647
+ "step": 3828,
2648
+ "total_flos": 4.742807303717683e+18,
2649
+ "train_loss": 0.13919885493931725,
2650
+ "train_runtime": 1939.8119,
2651
+ "train_samples_per_second": 31.549,
2652
+ "train_steps_per_second": 1.973
2653
+ }
2654
+ ],
2655
+ "logging_steps": 10,
2656
+ "max_steps": 3828,
2657
+ "num_train_epochs": 4,
2658
+ "save_steps": 100,
2659
+ "total_flos": 4.742807303717683e+18,
2660
+ "trial_name": null,
2661
+ "trial_params": null
2662
+ }