# TEG Datasets

## Dataset Format

Each dataset is a [PyG Data object](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.data.Dataset.html#torch_geometric.data.Dataset)  and is stored in the `processed` subdir following a unified format with each attribute defined as follows:

- `edge_index`: Graph connectivity in COO format with shape [2, num_edges] and type `torch.long`.
- `text_nodes`: `List` contains textual information for each node in the graph.
- `text_edges`: `List` contains textual information for each edge in the graph.
- `node_labels`: `List` contains text labels for each node in the graph. We use `-1` to represent nodes without labels

## Embedding Data Format
The embedding data is thrived from `text_nodes` and `text_edges` through PLM including:
- [GPT](https://platform.openai.com/docs/guides/embeddings)
- [BERT-base](https://huggingface.co/bert-base-uncased)
- [BERT-large](https://huggingface.co/bert-large-uncased)

**We will provide more TEG datasets and PLM embedding in the futrue!**