FinBERT-PT-BR
| Property | Value |
|---|---|
| License | Apache 2.0 |
| Language | Portuguese |
| Paper | View Paper |
| Downloads | 383,070 |
What is FinBERT-PT-BR?
FinBERT-PT-BR is a specialized BERT model designed for financial sentiment analysis in Brazilian Portuguese texts. The model underwent a two-stage training process: initial language modeling on 1.4 million financial news texts, followed by sentiment classification fine-tuning on 500 labeled examples. This approach has demonstrated superior performance compared to existing state-of-the-art models in Portuguese financial sentiment analysis.
Implementation Details
The model implements a BertForSequenceClassification architecture and can classify text into three sentiment categories: POSITIVE, NEGATIVE, and NEUTRAL. It can be easily integrated using the Transformers library and supports both direct model usage and pipeline implementation.
- Pre-trained on 1.4M financial texts
- Fine-tuned for sentiment analysis
- Supports batch processing
- Maximum sequence length of 512 tokens
Core Capabilities
- Financial sentiment analysis in Portuguese
- Market sentiment index generation
- Investment strategy support
- Macroeconomic data analysis
- Real-time market sentiment tracking
Frequently Asked Questions
Q: What makes this model unique?
This model is specifically trained on Brazilian Portuguese financial texts, making it highly specialized for financial sentiment analysis in this language. The two-stage training approach with extensive pre-training on domain-specific texts sets it apart from general-purpose sentiment models.
Q: What are the recommended use cases?
The model is ideal for analyzing financial news, market reports, and economic indicators in Portuguese. It can be used for building sentiment indices, developing investment strategies, and analyzing macroeconomic trends through textual data.
