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: 0.656
Support: 4875
Macro Avg:
Value: 0.638
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 model, trained on the 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 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 presented in Lacoste et al. (2019).