FinBERT-Tone Financial Topic Classification
| Property | Value |
|---|---|
| Base Model | yiyanghkust/finbert-tone |
| Task | Financial Topic Classification |
| Accuracy | 91.06% |
| F1 Score | 91.06% |
| Training Framework | PyTorch 1.13.0 |
What is finbert-tone-finetuned-finance-topic-classification?
This model is a specialized fine-tuned version of FinBERT-Tone, specifically adapted for classifying financial topics in Twitter content across 20 different categories. It incorporates intelligent class weighting to handle imbalanced data distribution, resulting in robust performance across all topic categories.
Implementation Details
The model was trained using a carefully tuned learning process with Adam optimizer (learning rate: 2e-05) and linear scheduler over 20 epochs. It employs native AMP mixed precision training for optimal performance and efficiency. The training process showed consistent improvement, starting from 85.52% accuracy in epoch 1 and reaching 90.79% by the final epoch.
- Batch size: 64 for both training and evaluation
- Optimization: Adam with betas=(0.9,0.999)
- Learning rate: 2e-05 with linear scheduling
- Mixed precision training implementation
Core Capabilities
- Multi-class classification across 20 financial topics
- Optimized for Twitter financial news content
- Balanced performance through weighted class handling
- High precision (91.13%) and recall (91.06%) metrics
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its specialized focus on financial Twitter content and its ability to handle class imbalance through weighted training. The high accuracy and balanced precision-recall metrics make it particularly reliable for real-world financial topic classification tasks.
Q: What are the recommended use cases?
The model is ideal for analyzing financial news and discussions on social media, particularly Twitter. It can be used for content categorization, trend analysis, and automated financial news routing systems.
