Shoriful025's picture
Create README.md
2a216c7 verified
metadata
tags:
  - crypto-prediction
  - time-series
  - bert
license: apache-2.0
datasets:
  - custom-crypto-news
metrics:
  - f1-score
model-index:
  - name: crypto-trend-predictor
    results:
      - task:
          type: text-classification
        dataset:
          name: custom-crypto-news
          type: custom
        metrics:
          - name: F1-Score
            type: f1
            value: 0.85

Crypto Trend Predictor

Overview

This BERT-based model predicts cryptocurrency market trends (bearish, bullish, or neutral) based on news articles, tweets, or market summaries. It was fine-tuned on a dataset of historical crypto news and price movements.

Model Architecture

  • Base Model: BERT-base-uncased
  • Layers: 12
  • Hidden Size: 768
  • Attention Heads: 12
  • Fine-tuned for multi-class classification (bearish/bullish/neutral)

Intended Use

Ideal for analyzing crypto-related text to forecast short-term market trends, assisting traders or analysts in decision-making.

Limitations

  • Predictions are based on text sentiment and may not account for external factors like regulations or economic events.
  • Performance degrades on non-English text or highly technical jargon not seen in training.
  • Not financial advice; use at your own risk.

Example Code

from transformers import pipeline

predictor = pipeline("text-classification", model="user/crypto-trend-predictor")
result = predictor("Bitcoin surges after ETF approval.")
print(result)
# [{'label': 'BULLISH', 'score': 0.95}]