Update README.md

README.md

@@ -1,3 +1,52 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+---
+
+HME100K – Hugging Face Version
+
+Dataset Name: HME100K (Hugging Face Version)
+
+Original Source: https://www.kaggle.com/datasets/cutedeadu/hme100k
+
+Original Author (Kaggle): cutedeadu
+
+## Overview
+HME100K is a dataset of handwritten mathematical expressions.
+Each sample consists of:
+- An image containing a handwritten math expression
+- A corresponding LaTeX-style text annotation
+
+## This dataset is useful for:
+- Handwritten Mathematical Expression Recognition (HMER)
+- OCR research for mathematical formulas
+- Sequence-to-sequence modeling (image-to-LaTeX)
+- Vision-language modeling tasks
+
+## Dataset Structure
+This Hugging Face version provides the dataset in a structured format with 3 columns:
+- id: Unique identifier of each sample
+- image: Handwritten math expression image
+- text: Ground truth LaTeX-style annotation
+
+The dataset is stored in Apache Arrow (PyArrow) format for efficient loading with the Hugging Face datasets library.
+
+## Why This Version?
+This version has been converted into Hugging Face Dataset format to:
+1. Make it easier to load using the datasets library
+2. Enable fast and memory-efficient access
+3. Allow seamless integration into Hugging Face training pipelines
+4. Provide convenient usage for researchers already working in the Hugging Face ecosystem
+
+This conversion does NOT modify the original data content.
+It only restructures the dataset for better accessibility.
+
+## Credit
+All original data credit goes to:
+- cutedeadu (Kaggle Author)
+- Original dataset: HME100K
+- Source: https://www.kaggle.com/datasets/cutedeadu/hme100k
+
+If you use this dataset in your research, please consider crediting the original author and Kaggle source accordingly.
+
+Hugging Face Conversion
+This Hugging Face formatted version was created to improve accessibility and usability for the open research community.