File size: 3,268 Bytes
48451f4
 
 
 
 
 
 
 
 
 
 
 
 
2e6e050
5b0f246
2e6e050
48451f4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: mit
tags:
- generated_from_trainer
model-index:
- name: mlcovid19-classifier
  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. -->

# mlcovid19-classifier
- [Mulit-lingual COVID-19 Fake News Detection and Intervention](https://counterinfodemic.org/)

This model is a fine-tuned version of [xlm-roberta-base](https://huggingface.co/xlm-roberta-base) on Multi-lingual COVID19 Fake News dataset. Please visite our project [website](https://counterinfodemic.org/) for more info.
It achieves the following results on the evaluation set:
- Loss: 0.4116
- F1 Macro: 0.6750
- F1 Misinformation: 0.9407
- F1 Factual: 0.8529
- F1 Other: 0.2315
- Prec Macro: 0.7057
- Prec Misinformation: 0.9229
- Prec Factual: 0.8958
- Prec Other: 0.2983

## 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: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 4367
- num_epochs: 30

### Training results

| Training Loss | Epoch | Step | Validation Loss | F1 Macro | F1 Misinformation | F1 Factual | F1 Other | Prec Macro | Prec Misinformation | Prec Factual | Prec Other |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-----------------:|:----------:|:--------:|:----------:|:-------------------:|:------------:|:----------:|
| 0.8111        | 3.67  | 500  | 0.4101          | 0.5506   | 0.9162            | 0.7356     | 0.0      | 0.5421     | 0.8969              | 0.7295       | 0.0        |
| 0.3688        | 7.35  | 1000 | 0.3397          | 0.5770   | 0.9321            | 0.7988     | 0.0      | 0.5694     | 0.9111              | 0.7972       | 0.0        |
| 0.3012        | 11.03 | 1500 | 0.3011          | 0.5912   | 0.9415            | 0.8322     | 0.0      | 0.5955     | 0.9104              | 0.8761       | 0.0        |
| 0.249         | 14.7  | 2000 | 0.3020          | 0.5931   | 0.9404            | 0.8388     | 0.0      | 0.5841     | 0.9206              | 0.8317       | 0.0        |
| 0.1957        | 18.38 | 2500 | 0.3308          | 0.6402   | 0.9406            | 0.8433     | 0.1365   | 0.7126     | 0.9234              | 0.8445       | 0.3699     |
| 0.1438        | 22.06 | 3000 | 0.3502          | 0.6615   | 0.9406            | 0.8529     | 0.1911   | 0.6952     | 0.9283              | 0.8543       | 0.3030     |
| 0.0996        | 25.73 | 3500 | 0.4116          | 0.6750   | 0.9407            | 0.8529     | 0.2315   | 0.7057     | 0.9229              | 0.8958       | 0.2983     |
| 0.0657        | 29.41 | 4000 | 0.4413          | 0.6422   | 0.9428            | 0.8497     | 0.1342   | 0.7126     | 0.9269              | 0.8453       | 0.3655     |


### Framework versions

- Transformers 4.23.0
- Pytorch 1.12.1+cu113
- Datasets 2.5.2
- Tokenizers 0.13.1