update readme

README.md

@@ -15,16 +15,45 @@ dataset_info:
     num_examples: 6315233
   download_size: 16370689925
   dataset_size: 56651995057
+  license: odc-by
+  task_categories:
+    - text-generation
+  language:
+    - en
+  pretty_name: OpenWebMath
+  size_categories:
+    - 10B<n<100B
 ---
 # <img align="left" width="50" height="50" src="imgs/openwebmath_logo.png" style="margin-top: -12px; margin-bottom: 0px;"> OpenWebMath
 
 
-**OpenWebMath** is a dataset containing the majority of the high-quality, mathematical text from the internet. It is filtered and extracted from over 200B HTML files on Common Crawl down to a set of 6.3 million documents containing 14.7B tokens. OpenWebMath is intended for use in *pretraining and finetuning large language models*.
+**OpenWebMath** is a dataset containing the majority of the high-quality, mathematical text from the internet. It is filtered and extracted from over 200B HTML files on Common Crawl down to a set of **6.3 million documents** containing **14.7B tokens**. OpenWebMath is intended for use in *pretraining and finetuning large language models*.
+
+Important Links:
+- [ArXiv Paper](...)
+- [GitHub Repository](...)
+- [Paper Twitter Thread](...)
+
+### Citation Information
+
+```
+@article{openwebmath,
+  title={OpenWebMath: An Open Dataset of High-Quality Mathematical Web Text},
+  author={Keiran Paster and Marco Dos Santos and Zhangir Azerbayev and Jimmy Ba},
+  journal={arXiv preprint arXiv:????.?????},
+  eprint={????.?????},
+  eprinttype = {arXiv},
+  url={https://arxiv.org/abs/????.?????},
+  year={2023}
+}
+```
 
 ## Dataset Contents
 
 Our aim with OpenWebMath was to build a dataset of as many mathematical documents sourced from the web as possible while preserving the formatting of mathematical content such as LaTeX equations. OpenWebMath contains documents from over 130k different domains, ranging from forums, educational and reference content, and blogs. The dataset contains documents covering mathematics, physics, statistics, computer science, and more.
 
+We provide automatically collected annotations for the source `url`, `timestamp` of the crawl and various information regarding the webpage and the text extraction.
+
 **Table 1:** Most Common Domains in OpenWebMath by Character Count
 | Domain                | # Characters | % Characters |
 |-----------------------|--------------|--------------|
@@ -38,16 +67,6 @@ Our aim with OpenWebMath was to build a dataset of as many mathematical document
 | groundai.com          | 545,214,990  | 1.12%        |
 | blogspot.com          | 520,392,333  | 1.07%        |
 | mathoverflow.net      | 499,102,560  | 1.02%        |
-| gmatclub.com          | 442,611,169  | 0.91%        |
-| gamedev.net           | 426,478,461  | 0.88%        |
-| ac.uk                 | 402,111,665  | 0.83%        |
-| aimsciences.org       | 344,716,386  | 0.71%        |
-| mathhelpforum.com     | 319,215,756  | 0.65%        |
-| deepai.org            | 313,512,520  | 0.64%        |
-| libretexts.org        | 282,014,149  | 0.58%        |
-| readthedocs.io        | 269,816,413  | 0.55%        |
-| tib.eu                | 199,714,017  | 0.41%        |
-| mit.edu               | 198,487,362  | 0.41%        |
 
 **Figure 1:** Content Distribution of OpenWebMath by Type and Subject
 <img src="imgs/contents_pie_chart.png" alt="Content Distribution of OpenWebMath by Type and Subject">
@@ -59,4 +78,11 @@ Our aim with OpenWebMath was to build a dataset of as many mathematical document
 
 OpenWebMath builds on the massive Common Crawl dataset, which contains over 200B HTML documents. We filtered the data to only include documents that are: (1) in English, (2) contain mathematical content, and (3) are of high quality. We also put a strong emphasis on extracting LaTeX content from the HTML documents as well as reducing boilerplate in comparison to other web datasets.
 
-As shown in Figure 2, the processing pipeline for OpenWebMath falls into five stages. First, we apply a prefilter to all HTML documents in Common Crawl to quickly judge whether they have mathematical content, skipping those that do not before doing the extensive processing needed to extract text and equations and remove boilerplate. Second, we extract the text, including mathematical content, from the HTML documents. Third, we apply language identification filters, perplexity-based quality filtering, and a mathematical content classifier filter. Fourth, we deduplicate the dataset using SimHash. Finally, we manually inspect the documents gathered in the previous steps and view documents from the most popular domains by document-count and character-count, removing domains that are not high quality.
+As shown in Figure 2, the processing pipeline for OpenWebMath falls into five stages. First, we apply a prefilter to all HTML documents in Common Crawl to quickly judge whether they have mathematical content, skipping those that do not before doing the extensive processing needed to extract text and equations and remove boilerplate. Second, we extract the text, including mathematical content, from the HTML documents. Third, we apply language identification filters, perplexity-based quality filtering, and a mathematical content classifier filter. Fourth, we deduplicate the dataset using SimHash. Finally, we manually inspect the documents gathered in the previous steps and view documents from the most popular domains by document-count and character-count, removing domains that are not high quality. Please find more information about the processing pipeline in the paper.
+
+## License
+
+OpenWebMath is made available under an ODC-By 1.0 license; users should also abide by the CommonCrawl ToU: [https://commoncrawl.org/terms-of-use/](https://commoncrawl.org/terms-of-use/). We do not alter the license of any of the underlying data.
+
+## Contact
+Keiran Paster ([keirp@cs.toronto.edu](mailto:keirp@cs.toronto.edu)) & Marco Dos Santos ([mjad3@cam.ac.uk](mailto:mjad3@cam.ac.uk))