File size: 3,311 Bytes
e82c80b
a85cfa3
c3d7593
 
 
 
 
 
 
 
e82c80b
 
 
 
 
c3d7593
 
 
 
 
 
 
3846d09
e82c80b
c3d7593
 
e82c80b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a85cfa3
 
c3d7593
a85cfa3
c3d7593
a85cfa3
c3d7593
a85cfa3
c3d7593
a85cfa3
c3d7593
a85cfa3
c3d7593
a85cfa3
c3d7593
a85cfa3
c3d7593
a85cfa3
6d5220c
a85cfa3
03f6b3d
 
e82c80b
03f6b3d
 
 
 
 
 
 
 
 
 
 
c3d7593
a85cfa3
e82c80b
 
 
 
a85cfa3
c3d7593
a85cfa3
 
 
 
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
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131

---
language:
- en
- ms
- zh
tags:
- sentiment-analysis
- text-classification
- multilingual
license: apache-2.0
datasets:
- tyqiangz/multilingual-sentiments
metrics:
- accuracy
model-index:
- name: xlm-roberta-base-sentiment-multilingual-finetuned
  results:
  - task:
      type: text-classification
      name: Text Classification
    dataset:
      type: tyqiangz/multilingual-sentiments
      name: Multilingual Sentiments
    metrics:
    - type: accuracy
      value: 0.7528205128205128


Baseline Scores:
  Classification Report:
    Negative:
      Precision: 0.6153
      Recall: 0.8292
      F1-score: 0.7064
      Support: 1680
    Neutral:
      Precision: 0.5381
      Recall: 0.3035
      F1-score: 0.3881
      Support: 1443
    Positive:
      Precision: 0.7607
      Recall: 0.7803
      F1-score: 0.7704
      Support: 1752
  Metrics:
    Accuracy:
      Value: 0.6560
      Support: 4875
    Macro Avg:
      Value: 0.6380
      Support: 4875
    Weighted Avg:
      Value: 0.6447
      Support: 4875

Finetuned Scores:
  Classification Report:
    Negative:
      Precision: 0.7487
      Recall: 0.7875
      F1-score: 0.7676
      Support: 1680
    Neutral:
      Precision: 0.6775
      Recall: 0.6216
      F1-score: 0.6484
      Support: 1443
    Positive:
      Precision: 0.8128
      Recall: 0.8276
      F1-score: 0.8201
      Support: 1752
  Metrics:
    Accuracy:
      Value: 0.7528
      Support: 4875
    Macro Avg:
      Value: 0.7463
      Support: 4875
    Weighted Avg:
      Value: 0.7507
      Support: 4875
---

# xlm-roberta-base-sentiment-multilingual-finetuned

## Model description

This is a fine-tuned version of the [cardiffnlp/twitter-xlm-roberta-base-sentiment-multilingual](https://huggingface.co/cardiffnlp/twitter-xlm-roberta-base-sentiment-multilingual) model, trained on the [tyqiangz/multilingual-sentiments](https://huggingface.co/datasets/tyqiangz/multilingual-sentiments) dataset. It's designed for multilingual sentiment analysis in English, Malay, and Chinese.

## Intended uses & limitations

This model is intended for sentiment analysis tasks in English, Malay, and Chinese. It can classify text into three sentiment categories: positive, negative, and neutral.

## Training and evaluation data

The model was trained and evaluated on the [tyqiangz/multilingual-sentiments](https://huggingface.co/datasets/tyqiangz/multilingual-sentiments) dataset, which includes data in English, Malay, and Chinese.

## Training procedure

The model was fine-tuned using the Hugging Face Transformers library.

training_args = TrainingArguments(
    output_dir="./results",
    num_train_epochs=5,
    per_device_train_batch_size=16,
    per_device_eval_batch_size=64,
    warmup_steps=500,
    weight_decay=0.01,
    logging_dir='./logs',
    logging_steps=10,
    evaluation_strategy="epoch",
    save_strategy="epoch",
    load_best_model_at_end=True,
)

## Evaluation results

'eval_accuracy': 0.7528205128205128, 'eval_f1': 0.7511924805177581, 'eval_precision': 0.7506612130427309, 'eval_recall': 0.7528205128205128


## Test Score : 

## Environmental impact

Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).