File size: 2,421 Bytes
ef61f13
 
 
3413581
ef61f13
 
 
 
 
 
 
 
 
 
 
 
 
3413581
ef61f13
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
tags:
- image-classification
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: exper4_mesum5
  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. -->

# exper4_mesum5

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 sudo-s/herbier_mesuem5 dataset.
It achieves the following results on the evaluation set:
- Loss: 3.4389
- Accuracy: 0.1331

## 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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 4
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 3.3793        | 0.23  | 100  | 3.4527          | 0.1308   |
| 3.2492        | 0.47  | 200  | 3.4501          | 0.1331   |
| 3.3847        | 0.7   | 300  | 3.4500          | 0.1272   |
| 3.3739        | 0.93  | 400  | 3.4504          | 0.1320   |
| 3.4181        | 1.16  | 500  | 3.4452          | 0.1320   |
| 3.214         | 1.4   | 600  | 3.4503          | 0.1320   |
| 3.282         | 1.63  | 700  | 3.4444          | 0.1325   |
| 3.5308        | 1.86  | 800  | 3.4473          | 0.1337   |
| 3.2251        | 2.09  | 900  | 3.4415          | 0.1361   |
| 3.4385        | 2.33  | 1000 | 3.4408          | 0.1343   |
| 3.3702        | 2.56  | 1100 | 3.4406          | 0.1325   |
| 3.366         | 2.79  | 1200 | 3.4411          | 0.1355   |
| 3.2022        | 3.02  | 1300 | 3.4403          | 0.1308   |
| 3.2768        | 3.26  | 1400 | 3.4394          | 0.1320   |
| 3.3444        | 3.49  | 1500 | 3.4394          | 0.1314   |
| 3.2981        | 3.72  | 1600 | 3.4391          | 0.1331   |
| 3.3349        | 3.95  | 1700 | 3.4389          | 0.1331   |


### Framework versions

- Transformers 4.20.1
- Pytorch 1.12.0+cu113
- Datasets 2.3.2
- Tokenizers 0.12.1