Indobert Analisis Sentimen Review Produk

🚀 IndoBERT Sentiment Classification ID - 2 Classes

This model is fine - tuned from indobenchmark/indobert - base - p2 for the task of sentiment classification of product reviews in Indonesian with two labels (positive and negative). The dataset used consists of 10,600 product reviews on the e - commerce platform Tokopedia.

🚀 Quick Start

This model is designed to classify comments or reviews into two sentiment classes: POSITIVE and NEGATIVE.

✨ Features

• Metrics: The model is evaluated using metrics such as Accuracy and F1 Score.
• Base Model: It is based on the pre - trained model [indobenchmark/indobert - base - p2](https://huggingface.co/indobenchmark/indobert - base - p2).

📦 Installation

No specific installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

# Load model dan tokenizer
from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
import torch

# Load dari Hugging Face Hub
model = AutoModelForSequenceClassification.from_pretrained("siRendy/indobert-analisis-sentimen-review-produk")
tokenizer = AutoTokenizer.from_pretrained("siRendy/indobert-analisis-sentimen-review-produk")

# Fungsi prediksi
def predict_sentiment(text):
    classifier = pipeline(
        "text-classification",
        model=model,
        tokenizer=tokenizer,
        device=0 if torch.cuda.is_available() else -1
    )

    result = classifier(text)[0]
    return {
        "sentiment": str(result["label"]),
        "confidence": round(result["score"], 4)
    }

# Contoh penggunaan
text = "Produk ini sangat buruk dan tidak layak dibeli."
prediction = predict_sentiment(text)
print(prediction)

📚 Documentation

🔧 Technical Details

Training Arguments

The following is the TrainingArguments configuration used during training:

from transformers import TrainingArguments

training_args = TrainingArguments(
    output_dir="./results",
    num_train_epochs=3,
    per_device_train_batch_size=4,
    gradient_accumulation_steps=4,
    per_device_eval_batch_size=4,
    weight_decay=0.05,
    eval_strategy="epoch",
    save_strategy="epoch",
    seed=42,
    load_best_model_at_end=True,
    metric_for_best_model="f1",
    logging_dir="./logs",
    report_to="tensorboard",
    logging_steps=100,
    warmup_ratio=0.05,
)

Evaluation Results

The model is evaluated over 3 epochs using Accuracy and F1 Score metrics. The following is the model's performance at each epoch:

The best model is selected from the 3rd epoch based on the highest F1 value.

📄 License

This project is licensed under the MIT license.