Create README.md

README.md

@@ -0,0 +1,32 @@
+# CryptoTrendPredictor
+
+## Overview
+
+CryptoTrendPredictor is a BERT-based model fine-tuned to predict short-term Bitcoin price movement direction (up or down) within the next 24 hours based on crypto-related news headlines and social media text. It outputs a binary prediction: **up** or **down**.
+
+This model is intended for research and educational purposes only. It is not financial advice.
+
+## Model Architecture
+
+- **Base model**: BERT-base-uncased
+- **Task head**: Binary classification head
+- **Hidden size**: 768
+- **Number of layers**: 12
+- **Parameters**: ~110M
+
+Built using `BertForSequenceClassification` from the Transformers library.
+
+## Usage
+
+```python
+from transformers import pipeline
+
+predictor = pipeline(
+    "text-classification",
+    model="your-username/CryptoTrendPredictor"
+)
+
+text = "Bitcoin ETF approved by SEC, major institutions entering market"
+result = predictor(text)
+print(result)
+# [{'label': 'up', 'score': 0.92}]