update README

README.md

@@ -69,4 +69,41 @@ configs:
   - split: retain
     path: github/retain-*
 license: apache-2.0
----
+---
+
+# 📖 unlearn_dataset
+The unlearn_dataset serves as a benchmark for evaluating unlearning methodologies in pre-trained large language models across diverse domains, including arXiv, GitHub. 
+
+## 🔍 Loading the datasets
+
+To load the dataset:
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("llmunlearn/unlearn_dataset", name="arxiv", split="forget")
+```
+* Available configuration names and corresponding splits:
+  - `arxiv`: `forget, approximate, retain`
+  - `github`: `forget, approximate, retain`
+  - `general`: `evaluation, retain`
+
+
+## 🛠️ Codebase
+
+For evaluating unlearning methods on our datasets, visit our [GitHub repository](https://github.com/yaojin17/Unlearning_LLM).
+
+
+
+## ⭐ Citing our Work
+
+If you find our codebase or dataset useful, please consider citing our paper:
+
+```bib
+@article{yao2024machine,
+  title={Machine Unlearning of Pre-trained Large Language Models},
+  author={Yao, Jin and Chien, Eli and Du, Minxin and Niu, Xinyao and Wang, Tianhao and Cheng, Zezhou and Yue, Xiang},
+  journal={arXiv preprint arXiv:2402.15159},
+  year={2024}
+}
+```