finbert-tone-finetuned-fintwitter-classification
| Property | Value |
|---|---|
| Parameter Count | 110M |
| Model Type | Financial Sentiment Analysis |
| Architecture | FinBERT (BERT-based) |
| Accuracy | 88.40% |
| F1 Score | 0.8838 |
What is finbert-tone-finetuned-fintwitter-classification?
This is a specialized financial sentiment analysis model that has been fine-tuned on Twitter financial news data. Built upon the FinBERT architecture, it's specifically designed to analyze and classify financial tweets into bullish, bearish, or neutral sentiments. The model demonstrates robust performance with 88.4% accuracy on evaluation tasks.
Implementation Details
The model was trained using carefully selected hyperparameters, including a learning rate of 2e-05, batch size of 16, and implemented with PyTorch. Training spanned 20 epochs using the Adam optimizer with native AMP mixed precision training.
- Uses linear learning rate scheduler
- Implements class weight balancing for improved performance on underrepresented labels
- Trained on the Twitter Financial News dataset
- Achieves balanced metrics across precision, recall, and F1 score
Core Capabilities
- Financial sentiment classification of tweets
- Handles complex financial terminology and context
- Balanced performance across different sentiment categories
- Optimized for real-time social media analysis
Frequently Asked Questions
Q: What makes this model unique?
The model combines FinBERT's financial domain expertise with specific optimizations for Twitter content, achieving high accuracy while maintaining balanced performance across different sentiment categories. The class weight adjustment feature makes it particularly effective for real-world applications where sentiment distribution may be uneven.
Q: What are the recommended use cases?
This model is ideal for financial market sentiment analysis, automated trading signals based on social media sentiment, market research, and real-time monitoring of financial news sentiment on Twitter. It's particularly suited for applications requiring high-accuracy classification of financial tweets.
