File size: 4,260 Bytes
908e86b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
43bb57d
 
908e86b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
43bb57d
908e86b
 
 
 
 
43bb57d
908e86b
 
 
 
 
43bb57d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
908e86b
 
 
 
 
 
 
 
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
---
base_model: smanjil/German-MedBERT
tags:
- generated_from_trainer
metrics:
- f1
model-index:
- name: German-MedBERT
  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. -->

# German-MedBERT

This model is a fine-tuned version of [smanjil/German-MedBERT](https://huggingface.co/smanjil/German-MedBERT) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5145
- F1: 0.4561

## 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-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50

### Training results

| Training Loss | Epoch | Step | Validation Loss | F1     |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 0.693         | 1.0   | 189  | 0.6754          | 0.0698 |
| 0.6853        | 2.0   | 378  | 0.6626          | 0.0339 |
| 0.6654        | 3.0   | 567  | 0.6499          | 0.0488 |
| 0.6562        | 4.0   | 756  | 0.6399          | 0.0541 |
| 0.6554        | 5.0   | 945  | 0.6335          | 0.0556 |
| 0.6394        | 6.0   | 1134 | 0.6260          | 0.0571 |
| 0.6452        | 7.0   | 1323 | 0.6220          | 0.0571 |
| 0.6257        | 8.0   | 1512 | 0.6161          | 0.0571 |
| 0.6334        | 9.0   | 1701 | 0.6117          | 0.0571 |
| 0.6302        | 10.0  | 1890 | 0.6068          | 0.0571 |
| 0.6151        | 11.0  | 2079 | 0.6011          | 0.0571 |
| 0.6121        | 12.0  | 2268 | 0.5961          | 0.0571 |
| 0.6097        | 13.0  | 2457 | 0.5915          | 0.0571 |
| 0.5929        | 14.0  | 2646 | 0.5865          | 0.0556 |
| 0.5955        | 15.0  | 2835 | 0.5822          | 0.0556 |
| 0.5893        | 16.0  | 3024 | 0.5776          | 0.1053 |
| 0.5936        | 17.0  | 3213 | 0.5731          | 0.1    |
| 0.5769        | 18.0  | 3402 | 0.5687          | 0.1    |
| 0.5692        | 19.0  | 3591 | 0.5646          | 0.1    |
| 0.5739        | 20.0  | 3780 | 0.5604          | 0.2326 |
| 0.5705        | 21.0  | 3969 | 0.5564          | 0.2326 |
| 0.5651        | 22.0  | 4158 | 0.5525          | 0.2727 |
| 0.5654        | 23.0  | 4347 | 0.5494          | 0.2727 |
| 0.5527        | 24.0  | 4536 | 0.5456          | 0.2727 |
| 0.5542        | 25.0  | 4725 | 0.5425          | 0.2727 |
| 0.5464        | 26.0  | 4914 | 0.5395          | 0.2727 |
| 0.5383        | 27.0  | 5103 | 0.5364          | 0.3111 |
| 0.5323        | 28.0  | 5292 | 0.5348          | 0.3111 |
| 0.5343        | 29.0  | 5481 | 0.5318          | 0.3404 |
| 0.5305        | 30.0  | 5670 | 0.5299          | 0.4082 |
| 0.5252        | 31.0  | 5859 | 0.5278          | 0.4    |
| 0.516         | 32.0  | 6048 | 0.5270          | 0.3922 |
| 0.5181        | 33.0  | 6237 | 0.5243          | 0.4231 |
| 0.5202        | 34.0  | 6426 | 0.5230          | 0.4231 |
| 0.5068        | 35.0  | 6615 | 0.5224          | 0.4231 |
| 0.514         | 36.0  | 6804 | 0.5205          | 0.4528 |
| 0.5014        | 37.0  | 6993 | 0.5194          | 0.4528 |
| 0.4899        | 38.0  | 7182 | 0.5188          | 0.4444 |
| 0.5104        | 39.0  | 7371 | 0.5164          | 0.4364 |
| 0.4823        | 40.0  | 7560 | 0.5174          | 0.4444 |
| 0.515         | 41.0  | 7749 | 0.5155          | 0.4364 |
| 0.4906        | 42.0  | 7938 | 0.5154          | 0.4364 |
| 0.4853        | 43.0  | 8127 | 0.5158          | 0.4364 |
| 0.5006        | 44.0  | 8316 | 0.5153          | 0.4364 |
| 0.503         | 45.0  | 8505 | 0.5146          | 0.4561 |
| 0.4915        | 46.0  | 8694 | 0.5141          | 0.4561 |
| 0.4903        | 47.0  | 8883 | 0.5144          | 0.4561 |
| 0.4892        | 48.0  | 9072 | 0.5146          | 0.4561 |
| 0.4939        | 49.0  | 9261 | 0.5146          | 0.4561 |
| 0.5007        | 50.0  | 9450 | 0.5145          | 0.4561 |


### Framework versions

- Transformers 4.32.1
- Pytorch 2.1.2
- Datasets 2.12.0
- Tokenizers 0.13.3