DeBERTa-v3 Financial Sentiment Classifier
| Property | Value |
|---|---|
| Base Model | microsoft/deberta-v3-base |
| Task | Financial Sentiment Classification |
| Training Data | 4,840 Financial News Articles |
| Accuracy | 89.13% |
| F1 Score | 89.12% |
| Author | nickmuchi |
What is deberta-v3-base-finetuned-finance-text-classification?
This is a specialized financial sentiment analysis model built on Microsoft's DeBERTa-v3 architecture. It's been fine-tuned on a comprehensive dataset of 4,840 financial news articles, combining data from financial-phrasebank and a COVID-19 sentiment dataset. The model classifies text into three sentiment categories: negative, neutral, and positive, making it particularly valuable for financial market analysis and research.
Implementation Details
The model was trained using a carefully optimized process with the following specifications: 15 epochs of training, using Adam optimizer with a learning rate of 2e-05, batch sizes of 16, and implementing native AMP for mixed precision training. The training showed consistent improvement, reaching its peak performance in the final epochs with remarkable stability.
- Training utilized linear learning rate scheduling
- Implemented with PyTorch 1.11.0 and Transformers 4.19.2
- Achieved high-quality metrics across all evaluation parameters
- Optimized for financial domain-specific text analysis
Core Capabilities
- Accurate sentiment classification with 89.27% precision
- Robust performance on financial news and reports
- Balanced performance across positive, negative, and neutral classifications
- Effective handling of COVID-19 related financial sentiment
Frequently Asked Questions
Q: What makes this model unique?
This model combines the powerful DeBERTa-v3 architecture with specialized financial domain training, achieving exceptional accuracy (89.13%) and F1-score (89.12%) on financial text classification tasks. Its training on both traditional financial news and COVID-19 related content makes it particularly robust for contemporary financial analysis.
Q: What are the recommended use cases?
The model is ideal for automated financial sentiment analysis, market research, news monitoring, and trading signal generation. It's particularly well-suited for applications requiring real-time analysis of financial news and reports, with reliable classification across positive, neutral, and negative sentiments.
