Improve dataset card: Add project overview, sample usage, and refine metadata

This PR enhances the dataset card for `SlimPajama-Meta-rater-Cleanliness-30B` by:
- Adding `license: apache-2.0` and additional tags (`data-selection`, `data-quality`) to the YAML metadata for better discoverability and accuracy.
- Including a direct link to the project's ACL Anthology page.
- Integrating the "Overview" and "PRRC Framework" sections from the original GitHub repository to provide richer context about the Meta-rater project and its methodology.
- Providing a clear sample code snippet for loading the dataset using the Hugging Face `datasets` library.
- Updating the citation to the official ACL Anthology BibTeX entry.

These updates aim to provide users with a more comprehensive understanding of the dataset and facilitate its usage.

README.md

@@ -1,12 +1,15 @@
 ---
-task_categories:
-- text-generation
 language:
 - en
-tags:
-- pretrain
 size_categories:
 - 10B<n<100B
+task_categories:
+- text-generation
+tags:
+- pretrain
+- data-quality
+- data-selection
+license: apache-2.0
 ---
 
 # Top 30B token SlimPajama Subset selected by the Cleanliness rater
@@ -14,6 +17,30 @@ size_categories:
 This repository contains the dataset described in the paper [Meta-rater: A Multi-dimensional Data Selection Method for Pre-training Language Models](https://huggingface.co/papers/2504.14194).
 
 Code: https://github.com/opendatalab/Meta-rater
+Project page: https://aclanthology.org/2025.acl-long.533/
+
+## 🎯 Overview
+
+The composition of pre-training datasets for large language models (LLMs) remains largely undisclosed, hindering transparency and efforts to optimize data quality—a critical driver of model performance. **Meta-rater** introduces a groundbreaking multi-dimensional data selection framework that **doubles convergence speed** and improves downstream task performance by **3.23%** compared to random selection.
+
+### 🏆 Key Achievements
+
+- **📈 2x Faster Convergence**: Meta-rater achieves equivalent performance using only 15B tokens compared to 30B tokens with random selection
+- **🎯 3.23% Performance Gain**: Significant improvement over random sampling on downstream tasks
+- **🔍 Multi-dimensional Quality Assessment**: Novel PRRC framework (Professionalism, Readability, Reasoning, Cleanliness)
+- **📊 Scalable Framework**: Benefits persist and increase from 1.3B to 7.2B parameter models
+- **🏗️ Comprehensive Dataset**: First fully annotated 627B-token SlimPajama with 25 quality metrics
+
+## 🧠 PRRC Framework
+
+We introduce four novel evaluation dimensions to comprehensively assess data quality:
+
+| Dimension | Description | F1 Score |
+|-----------|-------------|----------|
+| **🎓 Professionalism** | Degree of expertise and technical knowledge required | 91.57% |
+| **📖 Readability** | Ease of understanding and text clarity | 87.47% |
+| **🧮 Reasoning** | Complexity of logical thinking and analysis | 89.59% |
+| **✨ Cleanliness** | Format quality and noise-free content | 87.88% |
 
 ## Dataset Description
 
@@ -24,6 +51,23 @@ This dataset contains the top 30B tokens from the SlimPajama-627B corpus, select
 - **Quality metric**: Cleanliness (0–5 scale, see below)
 - **Annotation coverage**: 100% of selected subset
 
+## Sample Usage
+
+You can load this dataset using the Hugging Face `datasets` library:
+
+```python
+from datasets import load_dataset
+
+# Load the dataset
+dataset = load_dataset("opendatalab/SlimPajama-Meta-rater-Cleanliness-30B")
+
+# Print dataset information
+print(dataset)
+
+# Access a sample from the training split
+# print(dataset["train"][0])
+```
+
 ## Dataset Statistics
 
 - **Total tokens**: 30B (subset of SlimPajama-627B)
@@ -49,14 +93,35 @@ Scores are assigned by a ModernBERT model fine-tuned on Llama-3.3-70B-Instruct a
 
 ## Citation
 
-If you use this dataset, please cite:
+If you use Meta-rater in your research, please cite our paper:
 
 ```bibtex
-@article{zhuang2025meta,
-  title={Meta-rater: A Multi-dimensional Data Selection Method for Pre-training Language Models},
-  author={Zhuang, Xinlin and Peng, Jiahui and Ma, Ren and Wang, Yinfan and Bai, Tianyi and Wei, Xingjian and Qiu, Jiantao and Zhang, Chi and Qian, Ying and He, Conghui},
-  journal={arXiv preprint arXiv:2504.14194},
-  year={2025}
+@inproceedings{zhuang-etal-2025-meta,
+    title = "Meta-rater: A Multi-dimensional Data Selection Method for Pre-training Language Models",
+    author = "Zhuang, Xinlin  and
+      Peng, Jiahui  and
+      Ma, Ren  and
+      Wang, Yinfan  and
+      Bai, Tianyi  and
+      Wei, Xingjian  and
+      Jiantao, Qiu  and
+      Zhang, Chi  and
+      Qian, Ying  and
+      He, Conghui",
+    editor = "Che, Wanxiang  and
+      Nabende, Joyce  and
+      Shutova, Ekaterina  and
+      Pilehvar, Mohammad Taher",
+    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
+    month = jul,
+    year = "2025",
+    address = "Vienna, Austria",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2025.acl-long.533/",
+    doi = "10.18653/v1/2025.acl-long.533",
+    pages = "10856--10896",
+    ISBN = "979-8-89176-251-0",
+    abstract = "The composition of pre-training datasets for large language models (LLMs) remains largely undisclosed, hindering transparency and efforts to optimize data quality{---}a critical driver of model performance. Current data selection methods, such as natural language quality assessments, diversity-based filters, and classifier-based approaches, are limited by single-dimensional evaluation or redundancy-focused strategies. To address these gaps, we propose four dimensions to evaluate data quality: professionalism, readability, reasoning, and cleanliness. We further introduce \textbf{Meta-rater}, a multi-dimensional data selection method that integrates these dimensions with existing quality metrics through learned optimal weightings. Meta-rater employs proxy models to train a regression model that predicts validation loss, enabling the identification of optimal combinations of quality scores. Experiments demonstrate that Meta-rater \textbf{doubles convergence speed} for 1.3B parameter models and improves downstream task performance by \textbf{3.23{\%}}, with advantages that scale to models as large as 7.2B parameters. Our work establishes that holistic, multi-dimensional quality integration significantly outperforms conventional single-dimension approaches, offering a scalable paradigm for enhancing pre-training efficiency and model capability. To advance future research, we release scripts, data, and models at \url{https://github.com/opendatalab/Meta-rater}."
 }
 ```
 
@@ -72,4 +137,4 @@ This dataset is released under the same license as the original SlimPajama datas
 
 ---
 
-**Made with ❤️ by the OpenDataLab team** 
+**Made with ❤️ by the OpenDataLab team**