File size: 6,640 Bytes
7e6c6d0
 
 
 
 
 
 
 
 
 
 
 
 
 
02ef243
6e3f7e0
 
7e6c6d0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
---
tags:
- masked-image-modeling
- generated_from_trainer
datasets:
- cifar10
model-index:
- name: vit-cifar10
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# ViT pre-trained from scratch on CIFAR10

This model is a ViT (with the same arch as Google's [vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) pre-trained from scratch on the cifar10 dataset for masked image modeling.

It achieves the following results on the evaluation set:
- Loss: 0.0891

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 1337
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 100.0

### Training results

| Training Loss | Epoch | Step   | Validation Loss |
|:-------------:|:-----:|:------:|:---------------:|
| 0.289         | 1.0   | 2657   | 0.2941          |
| 0.2858        | 2.0   | 5314   | 0.2809          |
| 0.2693        | 3.0   | 7971   | 0.2738          |
| 0.2578        | 4.0   | 10628  | 0.2546          |
| 0.2211        | 5.0   | 13285  | 0.2153          |
| 0.1799        | 6.0   | 15942  | 0.1795          |
| 0.158         | 7.0   | 18599  | 0.1623          |
| 0.1481        | 8.0   | 21256  | 0.1453          |
| 0.1391        | 9.0   | 23913  | 0.1368          |
| 0.1348        | 10.0  | 26570  | 0.1354          |
| 0.129         | 11.0  | 29227  | 0.1249          |
| 0.126         | 12.0  | 31884  | 0.1229          |
| 0.1216        | 13.0  | 34541  | 0.1184          |
| 0.1175        | 14.0  | 37198  | 0.1185          |
| 0.1137        | 15.0  | 39855  | 0.1146          |
| 0.1125        | 16.0  | 42512  | 0.1117          |
| 0.1112        | 17.0  | 45169  | 0.1100          |
| 0.1108        | 18.0  | 47826  | 0.1089          |
| 0.1061        | 19.0  | 50483  | 0.1070          |
| 0.1073        | 20.0  | 53140  | 0.1076          |
| 0.1066        | 21.0  | 55797  | 0.1061          |
| 0.1065        | 22.0  | 58454  | 0.1056          |
| 0.1045        | 23.0  | 61111  | 0.1037          |
| 0.1052        | 24.0  | 63768  | 0.1055          |
| 0.102         | 25.0  | 66425  | 0.1028          |
| 0.1025        | 26.0  | 69082  | 0.1034          |
| 0.1037        | 27.0  | 71739  | 0.1025          |
| 0.1022        | 28.0  | 74396  | 0.1014          |
| 0.1026        | 29.0  | 77053  | 0.1011          |
| 0.1022        | 30.0  | 79710  | 0.1001          |
| 0.0997        | 31.0  | 82367  | 0.1007          |
| 0.0998        | 32.0  | 85024  | 0.1016          |
| 0.1019        | 33.0  | 87681  | 0.1008          |
| 0.0999        | 34.0  | 90338  | 0.1000          |
| 0.0998        | 35.0  | 92995  | 0.0993          |
| 0.0994        | 36.0  | 95652  | 0.0992          |
| 0.0966        | 37.0  | 98309  | 0.0991          |
| 0.0997        | 38.0  | 100966 | 0.0970          |
| 0.0991        | 39.0  | 103623 | 0.0979          |
| 0.099         | 40.0  | 106280 | 0.0983          |
| 0.0974        | 41.0  | 108937 | 0.0980          |
| 0.0974        | 42.0  | 111594 | 0.0971          |
| 0.0972        | 43.0  | 114251 | 0.0970          |
| 0.0991        | 44.0  | 116908 | 0.0970          |
| 0.0979        | 45.0  | 119565 | 0.0972          |
| 0.097         | 46.0  | 122222 | 0.0970          |
| 0.0936        | 47.0  | 124879 | 0.0967          |
| 0.0948        | 48.0  | 127536 | 0.0967          |
| 0.0974        | 49.0  | 130193 | 0.0954          |
| 0.0958        | 50.0  | 132850 | 0.0954          |
| 0.0948        | 51.0  | 135507 | 0.0955          |
| 0.095         | 52.0  | 138164 | 0.0953          |
| 0.0939        | 53.0  | 140821 | 0.0945          |
| 0.0961        | 54.0  | 143478 | 0.0948          |
| 0.0964        | 55.0  | 146135 | 0.0955          |
| 0.0934        | 56.0  | 148792 | 0.0948          |
| 0.0965        | 57.0  | 151449 | 0.0943          |
| 0.0966        | 58.0  | 154106 | 0.0941          |
| 0.0926        | 59.0  | 156763 | 0.0938          |
| 0.0928        | 60.0  | 159420 | 0.0942          |
| 0.093         | 61.0  | 162077 | 0.0936          |
| 0.0939        | 62.0  | 164734 | 0.0939          |
| 0.0936        | 63.0  | 167391 | 0.0936          |
| 0.093         | 64.0  | 170048 | 0.0929          |
| 0.0929        | 65.0  | 172705 | 0.0930          |
| 0.0917        | 66.0  | 175362 | 0.0925          |
| 0.0948        | 67.0  | 178019 | 0.0932          |
| 0.0931        | 68.0  | 180676 | 0.0927          |
| 0.0911        | 69.0  | 183333 | 0.0922          |
| 0.0923        | 70.0  | 185990 | 0.0924          |
| 0.0923        | 71.0  | 188647 | 0.0923          |
| 0.0929        | 72.0  | 191304 | 0.0919          |
| 0.0916        | 73.0  | 193961 | 0.0923          |
| 0.0927        | 74.0  | 196618 | 0.0921          |
| 0.0907        | 75.0  | 199275 | 0.0922          |
| 0.0927        | 76.0  | 201932 | 0.0919          |
| 0.0925        | 77.0  | 204589 | 0.0913          |
| 0.0921        | 78.0  | 207246 | 0.0917          |
| 0.0895        | 79.0  | 209903 | 0.0912          |
| 0.0916        | 80.0  | 212560 | 0.0914          |
| 0.09          | 81.0  | 215217 | 0.0909          |
| 0.0916        | 82.0  | 217874 | 0.0908          |
| 0.0902        | 83.0  | 220531 | 0.0907          |
| 0.0911        | 84.0  | 223188 | 0.0910          |
| 0.091         | 85.0  | 225845 | 0.0903          |
| 0.0903        | 86.0  | 228502 | 0.0905          |
| 0.0907        | 87.0  | 231159 | 0.0901          |
| 0.0908        | 88.0  | 233816 | 0.0907          |
| 0.0911        | 89.0  | 236473 | 0.0902          |
| 0.0905        | 90.0  | 239130 | 0.0906          |
| 0.089         | 91.0  | 241787 | 0.0901          |
| 0.0908        | 92.0  | 244444 | 0.0896          |
| 0.0894        | 93.0  | 247101 | 0.0892          |
| 0.0899        | 94.0  | 249758 | 0.0893          |
| 0.0899        | 95.0  | 252415 | 0.0897          |
| 0.0904        | 96.0  | 255072 | 0.0898          |
| 0.0906        | 97.0  | 257729 | 0.0894          |
| 0.0892        | 98.0  | 260386 | 0.0894          |
| 0.0881        | 99.0  | 263043 | 0.0892          |
| 0.09          | 100.0 | 265700 | 0.0894          |


### Framework versions

- Transformers 4.19.0.dev0
- Pytorch 1.10.0+cu111
- Datasets 2.0.0
- Tokenizers 0.11.6