Official Code Implementation of TaLK

Paper Information

• Title: TaLK: Text-attributed Graph Dataset Distillation via Coupling Language Model with Graph-Aware Kernel
• Description: Official implementation of the dataset distillation algorithm for text-attributed graphs (TAGs) with LM-GNN integrated learning.

---

Datasets

The statistics of the datasets used in this work are summarized below:

Dataset	$\vert V \vert$	$\vert E \vert$	# Class
Cora	2,708	5,429	7
Photo	48,362	500,928	12
Computers	87,229	721,081	10
Arxiv	169,343	1,166,243	40

• Cora / Arxiv: Citation network datasets.
  • The raw text and graph structures are sourced from here.
• Photo / Computers: Amazon co-purchase datasets.
  • The raw text and graph structures are sourced from here.

Note on Data Preparation: To download and set up the datasets, please follow the respective instructions provided in the source repositories linked above.

---

Run Instructions

1. Dataset Distillation

To distill a text-attributed graph dataset, run the following command:

python distillation.py --dataset [cora, Photo, Computers, arxiv] --data_root [ROOT FOLDER] --lm_name [bert, roberta, deberta] --seed [SEED] --cond_size [COND_SIZE] --gpu_id [GPU_ID]

2. Evaluation on Distilled Datasets

To train and evaluate models using the generated synthetic (distilled) dataset, use:

python evaluation.py --dataset [cora, Photo, Computers, arxiv] --data_root [ROOT FOLDER] --lm_name [bert, roberta, deberta] --seed [SEED] --cond_size [COND_SIZE] --gpu_id [GPU_ID] --cond_path [PATH FOR THE SAVED SYNTHETIC DATA]

---

Citation

If you find this code or our paper useful in your research, please cite:

[TBD]