Add rough readme template for later

README.md

@@ -12,4 +12,92 @@ tags:
 - Akkadian
 - PaddleOCR
 - PaddlePaddle
----
+---
+
+# NabuOCR
+
+*Ancient Cuneiform Meets Modern AI*
+
+NabuOCR is a specialized OCR model for transliterating ancient cuneiform tablets directly from images to ATF (ASCII Transliteration Format). Named after Nabu, the Mesopotamian god of writing and scribes, this model bridges a 5,000-year gap between humanity's earliest writing system and cutting-edge computer vision.
+
+## Overview
+
+NabuOCR processes images of cuneiform tablets and automatically generates scholarly transliterations in ATF format, the standard used by assyriologists worldwide. Built by fine-tuning PaddleOCR-VL on cuneiform tablet images, it can handle multiple views of tablets and produce complete transliterations including metadata.
+
+## Features
+
+- **Multi-view Processing**: Handles obverse, reverse, and edge views of tablets
+- **ATF Output**: Generates standard ATF format used by CDLI and other digital cuneiform projects
+- **Robust Recognition**: Trained on diverse tablet conditions from multiple periods
+- **Lightweight**: Based on the efficient 0.9B parameter PaddleOCR-VL model
+
+## Example Output
+
+Given an image of a cuneiform tablet, NabuOCR generates:
+
+```
+#atf: lang sux
+@tablet
+@obverse
+1. 1(disz) geme2 u4 1(disz)-sze3
+2. ki dingir-ra-ta
+3. da-da-ga
+4. szu ba-ti
+@reverse
+1. mu ki-masz{ki} ba-hul
+```
+
+## Model Architecture
+
+NabuOCR is built on PaddleOCR-VL, fine-tuned with:
+- **Training Data**: [Specify dataset size] cuneiform tablet images from CDLI
+- **Input Resolution**: 4096 max axis (automatically resized)
+- **Output Format**: ATF standard transliteration
+- **Languages Supported**: Sumerian (sux), Akkadian (akk), and other ancient Near Eastern languages
+
+## Usage Tips
+
+### Best Practices
+- Provide high-resolution images when possible (minimum 800x800 recommended)
+- Include all visible sides of the tablet in a single image or provide multiple views
+- Ensure good lighting and contrast in photographs
+- Remove excessive background from images
+
+
+## Performance
+
+| Dataset | Character Accuracy | Line Accuracy | Full Tablet Accuracy |
+|---------|-------------------|---------------|---------------------|
+| Test Set | XX.X% | XX.X% | XX.X% |
+| Old Babylonian | XX.X% | XX.X% | XX.X% |
+| Neo-Assyrian | XX.X% | XX.X% | XX.X% |
+
+## Limitations
+
+- Best performance on well-preserved tablets with clear impressions
+- May struggle with heavily damaged or eroded sections
+- Currently optimized for administrative and economic texts
+- Limited support for complex literary texts with unusual sign variants
+
+## Citation
+
+If you use NabuOCR in your research, please cite:
+
+```bibtex
+@software{nabuocr2025,
+  title={NabuOCR: Neural Cuneiform Transliteration},
+  author={[Zack Williams]},
+  year={2025},
+  url={https://huggingface.co/boatbomber/NabuOCR}
+}
+```
+
+## Acknowledgments
+
+- Built on [PaddleOCR-VL](https://github.com/PaddlePaddle/PaddleOCR)
+- Training data courtesy of the [Cuneiform Digital Library Initiative (CDLI)](https://cdli.ucla.edu/)
+- ATF format specification from [ORACC](http://oracc.museum.upenn.edu/)
+
+---
+
+*Bringing the ancient art of cuneiform into the age of artificial intelligence*