Add comprehensive dataset card for AIGer_Dataset

This PR adds a comprehensive dataset card for the `AIGer_Dataset`.

It includes:
- A link to the associated paper: https://huggingface.co/papers/2508.11991
- A link to the GitHub repository: https://github.com/ichont/AIGer
- Relevant `task_categories` (`graph-ml`) and descriptive `tags`.
- A detailed description of the dataset's content and purpose.
- Practical usage information, including environmental configuration and experimental parameters, drawn from the project's GitHub README, to facilitate reproducibility.

README.md

@@ -0,0 +1,63 @@
+---
+task_categories:
+- graph-ml
+tags:
+- electronic-design-automation
+- eda
+- logic-circuits
+- and-inverter-graphs
+- aigs
+- graph-neural-networks
+- signal-probability-prediction
+- truth-table-distance-prediction
+---
+
+# AIGer Dataset
+
+This dataset provides the experimental data associated with the paper [Modeling Relational Logic Circuits for And-Inverter Graph Convolutional Network](https://huggingface.co/papers/2508.11991).
+
+And-Inverter Graphs (AIGs) are fundamental in Electronic Design Automation (EDA) for efficiently representing, optimizing, and verifying the functional characteristics of digital circuits. This dataset contains the AIGs used to evaluate the AIGer model, a GNN-based approach designed to enhance the expression of AIGs by jointly modeling functional and structural characteristics and improving message passing.
+
+The data supports research in critical EDA tasks such as Signal Probability Prediction (SSP) and Truth Table Distance Prediction (TTDP), which are crucial for improving chip performance, energy efficiency, and reliability.
+
+**Paper:** [Modeling Relational Logic Circuits for And-Inverter Graph Convolutional Network](https://huggingface.co/papers/2508.11991)
+
+**Code:** [https://github.com/ichont/AIGer](https://github.com/ichont/AIGer)
+
+## Dataset Description
+
+This dataset consists of various And-Inverter Graph (AIG) samples, structured for training and evaluating graph neural networks in the context of logic circuit analysis. It is specifically designed to be used with the AIGer model for tasks like Signal Probability Prediction (SSP) and Truth Table Distance Prediction (TTDP), demonstrating improved performance compared to existing methods.
+
+## Environmental Configuration
+
+To utilize the AIGer model with this dataset, the following environment setup is recommended, as per the associated GitHub repository:
+
+**Experimental Environment Installation Package Version Requirements:**
+
+```
+torch==2.2.1+cu118
+torch-sparse==0.6.18
+torch_scatter==2.1.2
+numpy==1.26.4
+```
+
+**Experimental Equipment:**
+
+Ubuntu 22.04.5 LTS
+Nvidia A6000
+
+## Sample Usage
+
+This dataset is essential for training and evaluating the AIGer model. Below are the key experimental parameters provided in the GitHub repository for configuring experiments:
+
+- `--task 'prob'`: Select signal probability prediction (SPP) task.
+- `--model 'AIGer'`: Use AIGer model.
+- `--batch_size 256`: Set the batch size required for training.
+- `--split_file 0.05-0.05-0.9`: Indicates the data split proportions for training, validation, and testing.
+- `--layer_num 9`: Set the number of layers of the AIGer network to 9.
+
+You can run the AIGer training process by ensuring proper bash environment configuration and executing:
+
+```bash
+bash train.sh
+```