File size: 2,089 Bytes
f54ee5c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4fe8e51
 
 
 
 
 
 
 
f54ee5c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d8908a5
 
f54ee5c
 
 
cbc630e
f54ee5c
 
 
 
cbc630e
 
4fe8e51
 
 
 
 
f54ee5c
 
 
 
4fe8e51
 
f54ee5c
4fe8e51
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
---
base_model: vinai/bertweet-base
tags:
- generated_from_trainer
metrics:
- f1
- recall
model-index:
- name: bertweet-base
  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. -->

# bertweet-base

This model is a fine-tuned version of [vinai/bertweet-base](https://huggingface.co/vinai/bertweet-base) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6796
- F1 Macro: 0.8476
- F1: 0.8811
- F1 Neg: 0.8141
- Acc: 0.855
- Prec: 0.9267
- Recall: 0.8398
- Mcc: 0.7020

## 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: 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: 5
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | F1 Macro | F1     | F1 Neg | Acc    | Prec   | Recall | Mcc    |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|:------:|:------:|:------:|:------:|:------:|
| 0.6221        | 1.0   | 1161 | 0.5233          | 0.7216   | 0.8315 | 0.6116 | 0.765  | 0.7682 | 0.9062 | 0.4689 |
| 0.4332        | 2.0   | 2322 | 0.4843          | 0.7862   | 0.8680 | 0.7045 | 0.8175 | 0.8081 | 0.9375 | 0.5946 |
| 0.3714        | 3.0   | 3483 | 0.5872          | 0.8405   | 0.8963 | 0.7846 | 0.86   | 0.8521 | 0.9453 | 0.6914 |
| 0.2856        | 4.0   | 4644 | 0.5511          | 0.8589   | 0.8984 | 0.8194 | 0.87   | 0.8984 | 0.8984 | 0.7179 |
| 0.2199        | 5.0   | 5805 | 0.6796          | 0.8476   | 0.8811 | 0.8141 | 0.855  | 0.9267 | 0.8398 | 0.7020 |


### Framework versions

- Transformers 4.40.1
- Pytorch 2.3.0+cu121
- Datasets 2.18.0
- Tokenizers 0.19.1