Model Overview

This model, developed by tmmdev, is a specialized 7 billion parameter CodeLlama-based architecture designed for financial pattern recognition in trading charts. It processes OHLCV (Open, High, Low, Close, Volume) data to identify technical patterns with high precision, providing confidence scores for its analyses.

Key Capabilities

• Price Action & Chart Formations: Identifies channel patterns (>= 4 points), triangle formations (>= 3 points), and double top/bottom patterns (with <= 2% deviation).
• Support/Resistance & Trend Analysis: Detects support/resistance levels (>= 2 touches) and performs comprehensive trend line analysis.
• Volume Analysis: Incorporates volume profile analysis to enhance pattern recognition accuracy.
• Performance: Achieves a pattern accuracy of >85% and a false signal rejection rate of >90%, with a minimum confidence threshold of 0.8.

Good For

• Automated technical analysis in financial markets.
• Developing trading algorithms that rely on chart patterns.
• Applications requiring high-precision identification of financial market structures.

Technical Details

• Architecture: CodeLlama Base
• Context Window: 4096 tokens
• Quantization: 8-bit
• Memory: Requires a minimum of 8GB RAM.