This model is intended for binary sentiment classification tasks in Indonesian language texts. It predicts whether a given text expresses positive or negative sentiment.

Out-of-Scope Use

The model is not designed to classify neutral sentiments or handle languages other than Indonesian.

Bias, Risks, and Limitations

Bias: The model's performance is reliant on the quality and diversity of the training data. Biases in the dataset may influence predictions.
Limitations: The model is limited to binary sentiment analysis and may not perform well on ambiguous or mixed-sentiment texts.

Recommendations

Users should validate predictions on a case-by-case basis for high-stakes applications.

How to Get Started with the Model

python

from transformers import BertTokenizer, BertForSequenceClassification  

# Load the tokenizer and model  
tokenizer = BertTokenizer.from_pretrained("agufsamudra/indo-sentiment-analysis")  
model = BertForSequenceClassification.from_pretrained("agufsamudra/indo-sentiment-analysis")  

# Example usage  
text = "Saya sangat puas dengan pelayanan ini!"  
inputs = tokenizer(text, return_tensors="pt", padding="max_length", truncation=True, max_length=128)  
outputs = model(**inputs)  
logits = outputs.logits  
prediction = logits.argmax(-1).item()  

label = "Positive" if prediction == 1 else "Negative"  
print(f"Sentiment: {label}")  

Training Details

Training Data

The model was trained on a dataset of Indonesian text reviews from Play Store applications. The dataset was labeled for binary sentiment analysis (Positive and Negative). It contains an equal distribution of positive and negative examples to ensure balanced learning.

Training Procedure

Training Hyperparameters

Optimizer: AdamW
Learning Rate: 3e-6
Epochs: 3
Max Sequence Length: 128 tokens

Evaluation

Testing Data

The model was evaluated on a separate test dataset of 20,000 samples (10,000 Positive, 10,000 Negative).

Metrics

The model's performance was evaluated using standard metrics, including accuracy, precision, recall, and F1-score.

Results

Metric	Training Set	Testing Set
Accuracy	95.28%	95.56%
Precision	96%	96%
Recall	96%	96%
F1-Score	96%	96%

Technical Specifications

Model Architecture and Objective

The model is based on IndoBERT, a pre-trained transformer model tailored for Indonesian text. It was fine-tuned for binary classification tasks.

Compute Infrastructure

Hardware: Google Collab GPU
Software: Python, PyTorch, Transformers library