File size: 2,793 Bytes
17f6d0d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
tags:
- generated_from_trainer
model-index:
- name: testc8-1
  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. -->

# testc8-1

This model is a fine-tuned version of [shafin/chemical-bert-uncased-finetuned-cust-c2](https://huggingface.co./shafin/chemical-bert-uncased-finetuned-cust-c2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1490

## 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: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 0.0415        | 1.0   | 16   | 0.1392          |
| 0.0443        | 2.0   | 32   | 0.1289          |
| 0.0471        | 3.0   | 48   | 0.1363          |
| 0.042         | 4.0   | 64   | 0.1598          |
| 0.0452        | 5.0   | 80   | 0.1571          |
| 0.0446        | 6.0   | 96   | 0.1733          |
| 0.0466        | 7.0   | 112  | 0.1301          |
| 0.0391        | 8.0   | 128  | 0.1359          |
| 0.0425        | 9.0   | 144  | 0.1324          |
| 0.0436        | 10.0  | 160  | 0.0939          |
| 0.0406        | 11.0  | 176  | 0.1495          |
| 0.0387        | 12.0  | 192  | 0.1592          |
| 0.0335        | 13.0  | 208  | 0.1118          |
| 0.0413        | 14.0  | 224  | 0.1508          |
| 0.0363        | 15.0  | 240  | 0.1471          |
| 0.0428        | 16.0  | 256  | 0.1721          |
| 0.0384        | 17.0  | 272  | 0.1853          |
| 0.0381        | 18.0  | 288  | 0.1578          |
| 0.0373        | 19.0  | 304  | 0.1707          |
| 0.0351        | 20.0  | 320  | 0.1241          |
| 0.0346        | 21.0  | 336  | 0.1602          |
| 0.0386        | 22.0  | 352  | 0.1207          |
| 0.0274        | 23.0  | 368  | 0.1642          |
| 0.0338        | 24.0  | 384  | 0.1169          |
| 0.0327        | 25.0  | 400  | 0.1461          |
| 0.026         | 26.0  | 416  | 0.1323          |
| 0.0315        | 27.0  | 432  | 0.1403          |
| 0.042         | 28.0  | 448  | 0.1056          |
| 0.0346        | 29.0  | 464  | 0.1186          |
| 0.0294        | 30.0  | 480  | 0.1490          |


### Framework versions

- Transformers 4.24.0
- Pytorch 1.12.1+cu113
- Datasets 2.7.1
- Tokenizers 0.13.2