File size: 2,427 Bytes
4ee231e
 
 
 
de71cec
 
4ee231e
 
 
 
de71cec
 
 
 
 
 
 
 
 
 
 
 
4ee231e
 
 
 
 
 
 
de71cec
4ee231e
de71cec
 
4ee231e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
base_model: demdecuong/vihealthbert-base-word
tags:
- generated_from_trainer
datasets:
- tmnam20/pretrained-vn-med-nli
metrics:
- accuracy
model-index:
- name: vihealthbert-w_unsup-SynPD
  results:
  - task:
      name: Masked Language Modeling
      type: fill-mask
    dataset:
      name: tmnam20/pretrained-vn-med-nli all
      type: tmnam20/pretrained-vn-med-nli
      args: all
    metrics:
    - name: Accuracy
      type: accuracy
      value: 0.686153705209395
---

<!-- 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. -->

# vihealthbert-w_unsup-SynPD

This model is a fine-tuned version of [demdecuong/vihealthbert-base-word](https://huggingface.co/demdecuong/vihealthbert-base-word) on the tmnam20/pretrained-vn-med-nli all dataset.
It achieves the following results on the evaluation set:
- Loss: 1.5768
- Accuracy: 0.6862

## 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: 3e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 21363
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10.0

### Training results

| Training Loss | Epoch  | Step  | Validation Loss | Accuracy |
|:-------------:|:------:|:-----:|:---------------:|:--------:|
| 7.0234        | 0.8616 | 5000  | 2.5909          | 0.5576   |
| 5.2736        | 1.7232 | 10000 | 2.1890          | 0.5962   |
| 4.9126        | 2.5849 | 15000 | 1.9095          | 0.6381   |
| 4.791         | 3.4465 | 20000 | 1.8286          | 0.6469   |
| 4.6538        | 4.3081 | 25000 | 1.7144          | 0.6644   |
| 4.5846        | 5.1697 | 30000 | 1.6779          | 0.6704   |
| 4.5568        | 6.0314 | 35000 | 1.6362          | 0.6766   |
| 4.5079        | 6.8930 | 40000 | 1.6008          | 0.6814   |
| 4.469         | 7.7546 | 45000 | 1.6064          | 0.6805   |
| 4.4514        | 8.6162 | 50000 | 1.5800          | 0.6852   |
| 4.4317        | 9.4779 | 55000 | 1.5540          | 0.6880   |


### Framework versions

- Transformers 4.40.2
- Pytorch 2.0.1+cu118
- Datasets 2.19.1
- Tokenizers 0.19.1