README.md

metadata

title: Candlestick Chart Diffusion
emoji: 📈
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit

Candlestick Chart Diffusion Generator

Generate candlestick chart images from text descriptions using a conditional diffusion model.

Features

• 🎨 Generate candlestick charts from text prompts
• 🏋️ Train your own model on custom data
• 📂 Load pre-trained checkpoints

Example Prompts

• "bullish trend with high volatility"
• "bearish reversal pattern"
• "double bottom formation"
• "sideways market consolidation"
• "head and shoulders pattern"

How to Use

1. Generate Dataset: Use the data generator script to create training data
2. Train Model: Upload dataset and train for 50+ epochs
3. Generate Charts: Enter a text prompt and generate!

Model Architecture

• U-Net with cross-attention for text conditioning
• Diffusion with cosine noise schedule
• Text Encoder with transformer layers