File size: 4,145 Bytes
fea6f65
eb03a78
 
 
9c2ac56
 
eb03a78
 
 
fea6f65
 
eb03a78
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
base_model: Anwaarma/Improved-mBERT-attempt2
metrics:
- accuracy
tags:
- generated_from_trainer
model-index:
- name: robust-mbert
  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. -->

# robust-mbert

This model is a fine-tuned version of [Anwaarma/Improved-mBERT-attempt2](https://huggingface.co/Anwaarma/Improved-mBERT-attempt2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2410
- Accuracy: 0.92

## 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: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Accuracy |
|:-------------:|:------:|:----:|:---------------:|:--------:|
| No log        | 0.0546 | 50   | 0.3348          | 0.88     |
| No log        | 0.1092 | 100  | 0.3140          | 0.89     |
| No log        | 0.1638 | 150  | 0.4226          | 0.84     |
| No log        | 0.2183 | 200  | 0.2552          | 0.92     |
| No log        | 0.2729 | 250  | 0.3494          | 0.85     |
| No log        | 0.3275 | 300  | 0.2387          | 0.94     |
| No log        | 0.3821 | 350  | 0.3383          | 0.87     |
| No log        | 0.4367 | 400  | 0.3088          | 0.9      |
| No log        | 0.4913 | 450  | 0.3561          | 0.89     |
| 0.3057        | 0.5459 | 500  | 0.3598          | 0.85     |
| 0.3057        | 0.6004 | 550  | 0.2880          | 0.89     |
| 0.3057        | 0.6550 | 600  | 0.2306          | 0.92     |
| 0.3057        | 0.7096 | 650  | 0.3648          | 0.88     |
| 0.3057        | 0.7642 | 700  | 0.2796          | 0.9      |
| 0.3057        | 0.8188 | 750  | 0.3100          | 0.88     |
| 0.3057        | 0.8734 | 800  | 0.2689          | 0.91     |
| 0.3057        | 0.9279 | 850  | 0.2707          | 0.89     |
| 0.3057        | 0.9825 | 900  | 0.2684          | 0.87     |
| 0.3057        | 1.0371 | 950  | 0.4417          | 0.86     |
| 0.2777        | 1.0917 | 1000 | 0.3980          | 0.88     |
| 0.2777        | 1.1463 | 1050 | 0.3233          | 0.9      |
| 0.2777        | 1.2009 | 1100 | 0.2857          | 0.9      |
| 0.2777        | 1.2555 | 1150 | 0.3229          | 0.89     |
| 0.2777        | 1.3100 | 1200 | 0.2364          | 0.92     |
| 0.2777        | 1.3646 | 1250 | 0.3015          | 0.87     |
| 0.2777        | 1.4192 | 1300 | 0.2713          | 0.89     |
| 0.2777        | 1.4738 | 1350 | 0.3839          | 0.87     |
| 0.2777        | 1.5284 | 1400 | 0.3173          | 0.9      |
| 0.2777        | 1.5830 | 1450 | 0.2690          | 0.91     |
| 0.2138        | 1.6376 | 1500 | 0.3804          | 0.89     |
| 0.2138        | 1.6921 | 1550 | 0.3020          | 0.88     |
| 0.2138        | 1.7467 | 1600 | 0.2702          | 0.89     |
| 0.2138        | 1.8013 | 1650 | 0.2815          | 0.9      |
| 0.2138        | 1.8559 | 1700 | 0.2867          | 0.89     |
| 0.2138        | 1.9105 | 1750 | 0.2861          | 0.87     |
| 0.2138        | 1.9651 | 1800 | 0.2585          | 0.89     |
| 0.2138        | 2.0197 | 1850 | 0.3170          | 0.9      |
| 0.2138        | 2.0742 | 1900 | 0.2928          | 0.9      |
| 0.2138        | 2.1288 | 1950 | 0.2635          | 0.93     |
| 0.1966        | 2.1834 | 2000 | 0.2695          | 0.93     |
| 0.1966        | 2.2380 | 2050 | 0.3348          | 0.9      |
| 0.1966        | 2.2926 | 2100 | 0.3577          | 0.91     |
| 0.1966        | 2.3472 | 2150 | 0.3360          | 0.92     |
| 0.1966        | 2.4017 | 2200 | 0.3721          | 0.91     |
| 0.1966        | 2.4563 | 2250 | 0.2410          | 0.92     |


### Framework versions

- Transformers 4.42.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1