đ fogsmog_hfclass
This model is a fine - tuned image classification model based on the ViT architecture, which achieves high accuracy on the imagefolder dataset.
đ Quick Start
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.
It achieves the following results on the evaluation set:
- Loss: 0.3700
- Accuracy: 0.91
đ Documentation
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
đ§ Technical Details
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e - 05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 20
Training results
Training Loss |
Epoch |
Step |
Validation Loss |
Accuracy |
0.6766 |
1.0 |
25 |
0.6299 |
0.795 |
0.3444 |
2.0 |
50 |
0.3701 |
0.8625 |
0.2456 |
3.0 |
75 |
0.2988 |
0.885 |
0.1402 |
4.0 |
100 |
0.3076 |
0.905 |
0.1275 |
5.0 |
125 |
0.4505 |
0.8525 |
0.0909 |
6.0 |
150 |
0.3739 |
0.8825 |
0.0792 |
7.0 |
175 |
0.3642 |
0.885 |
0.0482 |
8.0 |
200 |
0.3812 |
0.885 |
0.0451 |
9.0 |
225 |
0.3290 |
0.9 |
0.0526 |
10.0 |
250 |
0.4004 |
0.8825 |
0.0575 |
11.0 |
275 |
0.2842 |
0.925 |
0.0457 |
12.0 |
300 |
0.3952 |
0.895 |
0.0505 |
13.0 |
325 |
0.4411 |
0.885 |
0.0324 |
14.0 |
350 |
0.4185 |
0.8925 |
0.0354 |
15.0 |
375 |
0.3347 |
0.9025 |
0.0443 |
16.0 |
400 |
0.2949 |
0.915 |
0.0305 |
17.0 |
425 |
0.3603 |
0.905 |
0.0234 |
18.0 |
450 |
0.3858 |
0.8875 |
0.0219 |
19.0 |
475 |
0.3541 |
0.91 |
0.0284 |
20.0 |
500 |
0.3700 |
0.91 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1+cu117
- Datasets 2.9.0
- Tokenizers 0.13.2
đ License
This model is licensed under the Apache - 2.0 license.
đĻ Additional Information
Property |
Details |
Model Type |
Fine - tuned on google/vit - base - patch16 - 224 - in21k for image classification |
Training Data |
imagefolder |
Metrics |
accuracy |
Results |
On the test split of imagefolder dataset, accuracy is 0.91 |