Bert Base Chinese Finetuned Sentiment

🚀 BERT Chinese Sentiment Analysis Model

This is a BERT-based Chinese sentiment analysis model that can be used to determine the sentiment tendency (positive, negative, or neutral) of text.

🚀 Quick Start

This BERT-based Chinese sentiment analysis model can judge the sentiment tendency (positive, negative, or neutral) of text.

✨ Features

• The model is fine-tuned based on bert-base-chinese.
• It is suitable for sentiment analysis of Chinese app reviews.
• Output labels: 0 (negative), 1 (positive), 2 (neutral).
• It is trained using Focal Loss to handle the class imbalance problem.

💻 Usage Examples

Basic Usage

from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch

# Load the model and tokenizer
model = AutoModelForSequenceClassification.from_pretrained("jackietung/bert-base-chinese-sentiment-finetuned")
tokenizer = AutoTokenizer.from_pretrained("jackietung/bert-base-chinese-sentiment-finetuned")

# Prepare the input
text = "這個App使用體驗很差！"
inputs = tokenizer(text, return_tensors="pt")

# Make predictions
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
    
    # Get the prediction result
    label_names = ["負面", "正面", "中性"]
    predicted_class = torch.argmax(predictions, dim=1).item()
    
    print(f"預測類別: {label_names[predicted_class]}")
    print(f"預測分數: {predictions[0][predicted_class].item():.4f}")
    
    # Show scores for all classes
    for i, label in enumerate(label_names):
        print(f"{label} 分數: {predictions[0][i].item():.4f}")

📄 License

This project is licensed under the MIT License.