Update README.md

README.md

@@ -1,215 +1,8 @@
 ---
-license: mit
 task_categories:
-- image-generation
-- computer-vision
-- other
-language:
-- en
+- image-to-image
 tags:
-- animation
-- mathematical-patterns
-- procedural-generation
-- computer-graphics
-- fractals
-- generative-art
-- webdataset
-size_categories:
-- 1M<n<10M
+- WebDataset
 ---
 
-# GradientTile 1M Dataset
-
-A large-scale dataset of 1 million procedurally generated animated GIFs featuring mathematical patterns, fractals, and artistic visualizations.
-
-## Dataset Description
-
-GradientTile 1M is a comprehensive dataset of animated mathematical patterns generated using advanced procedural algorithms. Each sample is a 100x100 pixel animated GIF with 60 frames, showcasing various mathematical patterns including fractals, geometric shapes, organic forms, and abstract visualizations.
-
-### Dataset Summary
-
-- **Total Samples**: 100,000
-- **Total Size**: 9.44 GB
-- **Pattern Types**: 159 unique patterns
-- **Color Schemes**: 8 color schemes
-- **Format**: WebDataset (tar files)
-- **Resolution**: 100x100 pixels
-- **Frames**: 60 frames per animation
-- **Duration**: 6 seconds per animation (100ms per frame)
-
-## Pattern Categories
-
-### Mathematical Patterns
-- **Fractals**: Mandelbrot, Julia sets, and other fractal patterns
-- **Geometric**: Spirals, waves, concentric circles, ripples
-- **Curves**: Lissajous, rose curves, cardioids, lemniscates
-- **Complex**: Interference patterns, plasma, turbulence
-
-### 3D Projections
-- **Shapes**: Spheres, cubes, pyramids, diamonds, torus
-- **Biological**: DNA helix, molecular structures
-
-### Organic & Nature
-- **Natural**: Trees, leaves, veins, neural networks
-- **Atmospheric**: Aurora, northern lights, solar flares
-- **Oceanic**: Wave motion, underwater effects, coral
-
-### Technology & Aesthetics
-- **Digital**: Circuit boards, matrix effects, holograms
-- **Energy**: Laser, electric, magnetic, quantum effects
-- **Punk Aesthetics**: Cyberpunk, steampunk, biopunk styles
-
-### Cosmic & Abstract
-- **Space**: Galactic, stellar, planetary, black holes
-- **Abstract**: Infinity, eternity, consciousness, awareness
-- **Philosophical**: Present, now, everywhere, nowhere
-
-## Color Schemes
-
-1. **Rainbow** - Vibrant spectrum colors
-2. **Fire** - Warm red, orange, yellow tones
-3. **Ocean** - Cool blue, teal, aqua tones
-4. **Neon** - Bright electric colors
-5. **Aurora** - Northern lights green and purple
-6. **Cosmic** - Deep space blues and purples
-7. **Thermal** - Heat map red to blue
-8. **Random** - Procedurally generated colors
-
-## Dataset Structure
-
-The dataset is organized in WebDataset format for efficient loading:
-
-```
-gradienttile-1m/
-├── shard_000000.tar    # Samples 0-999
-├── shard_000001.tar    # Samples 1000-1999
-├── ...
-├── shard_009999.tar    # Samples 9990000-9999999
-└── dataset_info.json   # Dataset metadata
-```
-
-Each shard contains:
-- `sample_XXXXXX.gif` - Animated GIF file
-- `sample_XXXXXX.json` - Complete metadata
-
-## Sample Metadata
-
-Each sample includes comprehensive metadata:
-
-```json
-{
-  "sample_id": 42,
-  "pattern_type": "mandelbrot",
-  "color_scheme": "rainbow",
-  "seed": 42,
-  "parameters": { /* pattern-specific parameters */ },
-  "frequency_stats": { /* mathematical frequency analysis */ },
-  "reproducibility": { /* complete args for regeneration */ }
-}
-```
-
-## Usage
-
-### Loading with Hugging Face Datasets
-
-```python
-from datasets import load_dataset
-
-# Load the dataset
-dataset = load_dataset("your-username/gradienttile-1m")
-
-# Access samples
-sample = dataset["train"][0]
-gif_data = sample["gif"]
-metadata = sample["metadata"]
-```
-
-### Loading with WebDataset
-
-```python
-import webdataset as wds
-
-# Create WebDataset
-dataset = wds.WebDataset("gradienttile-1m/shard-{000000..000999}.tar")
-dataset = dataset.decode("pil").to_tuple("gif", "json")
-
-# Iterate through samples
-for gif, metadata in dataset:
-    # Process sample
-    pass
-```
-
-### Loading with PyTorch
-
-```python
-import torch
-from torch.utils.data import DataLoader
-
-# Create PyTorch dataset
-dataset = wds.WebDataset("gradienttile-1m/shard-{000000..000999}.tar")
-dataset = dataset.decode("pil").to_tuple("gif", "json")
-dataloader = DataLoader(dataset, batch_size=32, shuffle=True)
-```
-
-## Applications
-
-### Machine Learning
-- **Pattern Recognition**: Train models to classify mathematical patterns
-- **Style Transfer**: Learn to transfer color schemes between patterns
-- **Generation**: Train GANs to generate new animated patterns
-- **Compression**: Develop efficient video compression algorithms
-
-### Research
-- **Mathematical Analysis**: Study frequency properties and pattern relationships
-- **Procedural Generation**: Research new pattern generation algorithms
-- **Visualization**: Create tools for mathematical concept visualization
-
-### Creative Applications
-- **Art Generation**: Create unique animated artworks
-- **Education**: Teach mathematics through visual patterns
-- **Entertainment**: Generate dynamic backgrounds and visual effects
-
-## Generation Process
-
-The dataset was generated using the GradientTile animation generator with:
-
-1. **Random Pattern Selection**: Each sample uses a randomly selected pattern type
-2. **Parameter Generation**: Pattern-specific parameters are randomly generated within valid ranges
-3. **Color Scheme Assignment**: Random color scheme selection
-4. **Reproducibility**: Each sample uses a deterministic seed for full reproducibility
-5. **Quality Control**: All samples are validated for correct format and content
-
-## Technical Specifications
-
-- **Image Format**: Animated GIF
-- **Resolution**: 100x100 pixels
-- **Frame Rate**: 10 FPS (100ms per frame)
-- **Duration**: 6 seconds per animation
-- **Compression**: GIF LZW compression
-- **File Size**: 50-200KB per sample (average ~100KB)
-
-## Citation
-
-If you use this dataset in your research, please cite:
-
-```bibtex
-@dataset{gradienttile1m,
-  title={GradientTile 1M: A Large-Scale Dataset of Animated Mathematical Patterns},
-  author={Your Name},
-  year={2024},
-  url={https://huggingface.co/datasets/your-username/gradienttile-1m},
-  note={A dataset of 1 million procedurally generated animated mathematical patterns}
-}
-```
-
-## License
-
-This dataset is released under the MIT License. See the LICENSE file for details.
-
-## Contact
-
-For questions or issues, please open an issue on the [GitHub repository](https://github.com/your-username/gradienttile) or contact the maintainers.
-
----
-
-*Generated on 2025-10-26 20:43:07*
+A large-scale dataset of 100K procedurally generated animated GIFs featuring mathematical patterns, fractals, and artistic visualizations.