đ vit-large-patch32-384-finetuned-melanoma
This model is a fine - tuned version of google/vit-large-patch32-384 on the imagefolder dataset. It can be used for image classification tasks and achieves an accuracy of 0.8273 on the evaluation set.
đ Quick Start
This model is a fine - tuned version of google/vit-large-patch32-384 on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0767
- Accuracy: 0.8273
đ Documentation
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: 1e - 05
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 40
Training results
Training Loss |
Epoch |
Step |
Validation Loss |
Accuracy |
1.0081 |
1.0 |
550 |
0.7650 |
0.68 |
0.7527 |
2.0 |
1100 |
0.6693 |
0.7364 |
0.6234 |
3.0 |
1650 |
0.6127 |
0.7709 |
2.6284 |
4.0 |
2200 |
0.6788 |
0.7655 |
0.1406 |
5.0 |
2750 |
0.6657 |
0.7836 |
0.317 |
6.0 |
3300 |
0.6936 |
0.78 |
2.5358 |
7.0 |
3850 |
0.7104 |
0.7909 |
1.5802 |
8.0 |
4400 |
0.6928 |
0.8 |
0.088 |
9.0 |
4950 |
0.8060 |
0.7982 |
0.0183 |
10.0 |
5500 |
0.7811 |
0.8091 |
0.0074 |
11.0 |
6050 |
0.7185 |
0.7945 |
0.0448 |
12.0 |
6600 |
0.8780 |
0.7909 |
0.4288 |
13.0 |
7150 |
0.8229 |
0.82 |
0.017 |
14.0 |
7700 |
0.7516 |
0.8182 |
0.0057 |
15.0 |
8250 |
0.7974 |
0.7964 |
1.7571 |
16.0 |
8800 |
0.7866 |
0.8218 |
1.3159 |
17.0 |
9350 |
0.8491 |
0.8073 |
1.649 |
18.0 |
9900 |
0.8432 |
0.7891 |
0.0014 |
19.0 |
10450 |
0.8870 |
0.82 |
0.002 |
20.0 |
11000 |
0.9460 |
0.8236 |
0.3717 |
21.0 |
11550 |
0.8866 |
0.8327 |
0.0025 |
22.0 |
12100 |
1.0287 |
0.8073 |
0.0094 |
23.0 |
12650 |
0.9696 |
0.8091 |
0.002 |
24.0 |
13200 |
0.9659 |
0.8018 |
0.1001 |
25.0 |
13750 |
0.9712 |
0.8327 |
0.2953 |
26.0 |
14300 |
1.0512 |
0.8236 |
0.0141 |
27.0 |
14850 |
1.0503 |
0.82 |
0.612 |
28.0 |
15400 |
1.2020 |
0.8109 |
0.0792 |
29.0 |
15950 |
1.0498 |
0.8364 |
0.0117 |
30.0 |
16500 |
1.0079 |
0.8327 |
0.0568 |
31.0 |
17050 |
1.0199 |
0.8255 |
0.0001 |
32.0 |
17600 |
1.0319 |
0.8291 |
0.075 |
33.0 |
18150 |
1.0427 |
0.8382 |
0.001 |
34.0 |
18700 |
1.1289 |
0.8382 |
0.0001 |
35.0 |
19250 |
1.0589 |
0.8364 |
0.0006 |
36.0 |
19800 |
1.0349 |
0.8236 |
0.0023 |
37.0 |
20350 |
1.1192 |
0.8273 |
0.0002 |
38.0 |
20900 |
1.0863 |
0.8273 |
0.2031 |
39.0 |
21450 |
1.0604 |
0.8255 |
0.0006 |
40.0 |
22000 |
1.0767 |
0.8273 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.12.1+cu113
- Datasets 2.7.0
- Tokenizers 0.13.2
đ License
This model is licensed under the Apache - 2.0 license.
đ Model Index
- name: vit-large-patch32-384-finetuned-melanoma
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: train
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.8272727272727273
Property |
Details |
Model Type |
vit-large-patch32-384-finetuned-melanoma |
Training Data |
imagefolder |
Metrics |
Accuracy: 0.8272727272727273 |