README.md

metadata

tags:
  - sentiment-analysis
  - finance
  - stock-market
  - english
pipeline_tag: text-classification
license: mit
language: en

🧠 FinancialBERT Sentiment Analysis (FinNews Binary)

This is a fine-tuned BERT model for binary sentiment classification of financial news headlines, adapted for real-time stock market sentiment prediction.

🔍 Model Details

• Architecture: BERT (12-layer, 768-hidden, 12-heads)
• Base model: ahmedrachid/FinancialBERT-Sentiment-Analysis
• Fine-tuned task: Binary classification – Positive or Negative
• Problem type: single_label_classification
• Special tokens: [CLS], [SEP], [PAD], [MASK], [UNK]

Neutral headlines are mapped to Positive to simplify binary output.

🧾 Training Summary

• Dataset: 5,000+ manually labeled financial news headlines
• Tokenizer: Custom WordPiece tokenizer
• Max sequence length: 128
• Framework: Transformers v4.51.3 (PyTorch backend)
• Output labels:
  • LABEL_0 = Negative
  • LABEL_1 = Positive

📈 Intended Use

Ideal for:

• Real-time market sentiment dashboard
• Trading signal pipelines
• Event-driven NLP analysis

🚀 Usage (Example)

from transformers import pipeline

classifier = pipeline("text-classification", model="your-username/your-model-name")
classifier("Apple's Q4 earnings beat expectations amid strong iPhone sales")
# Output: [{'label': 'LABEL_1', 'score': 0.98}]