File size: 2,818 Bytes
5c81dec
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cd1c1a0
5c81dec
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cd1c1a0
 
5c81dec
 
 
cd1c1a0
5c81dec
 
 
 
 
cd1c1a0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5c81dec
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model: google/vit-base-patch16-224-in21k
tags:
- HHD
- 3_class
- ViT
- generated_from_trainer
model-index:
- name: ViT_face
  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_face

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 face dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6941

## 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: 128
- eval_batch_size: 128
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| No log        | 1.0   | 10   | 1.0691          |
| No log        | 2.0   | 20   | 1.0378          |
| No log        | 3.0   | 30   | 0.9958          |
| No log        | 4.0   | 40   | 0.9437          |
| No log        | 5.0   | 50   | 0.8915          |
| No log        | 6.0   | 60   | 0.8396          |
| No log        | 7.0   | 70   | 0.7950          |
| No log        | 8.0   | 80   | 0.7602          |
| No log        | 9.0   | 90   | 0.7246          |
| No log        | 10.0  | 100  | 0.7009          |
| No log        | 11.0  | 110  | 0.6882          |
| No log        | 12.0  | 120  | 0.6700          |
| No log        | 13.0  | 130  | 0.6629          |
| No log        | 14.0  | 140  | 0.6646          |
| No log        | 15.0  | 150  | 0.6558          |
| No log        | 16.0  | 160  | 0.6679          |
| No log        | 17.0  | 170  | 0.6637          |
| No log        | 18.0  | 180  | 0.6689          |
| No log        | 19.0  | 190  | 0.6690          |
| No log        | 20.0  | 200  | 0.6744          |
| No log        | 21.0  | 210  | 0.6787          |
| No log        | 22.0  | 220  | 0.6823          |
| No log        | 23.0  | 230  | 0.6832          |
| No log        | 24.0  | 240  | 0.6866          |
| No log        | 25.0  | 250  | 0.6883          |
| No log        | 26.0  | 260  | 0.6912          |
| No log        | 27.0  | 270  | 0.6923          |
| No log        | 28.0  | 280  | 0.6935          |
| No log        | 29.0  | 290  | 0.6939          |
| No log        | 30.0  | 300  | 0.6941          |


### Framework versions

- Transformers 4.42.4
- Pytorch 2.4.0+cu121
- Datasets 2.21.0
- Tokenizers 0.19.1