PyTorch Docathon 2025 Kickoff

Key Takeaways

[Music] [Music] Hey, hey, hey. [Music] Hello and welcome to the Docathon kickoff video. We'll be sharing the information you need to know to this year's PieTorch Docathon. I'm Elena, a documentation engineer in the PyTorch team. I'm Ivan, another documentation engineer in the PyTorch team. And I'm Satlana, a doc engineer as well. We're excited to have you join us in improving the documentation for PyTorch and its related libraries. Community involvement in open source is essential for the health and longevity of the project. We can't do it without you. This time around, we'll be focusing on two main repositories, the PyTorch PyTorch repo and the PyTorch tutorials repo. In addition, we have some libraries that will also be part of this event. The docathon will start on June 3rd and run for two weeks. The final pull request should be submitted by June 15th at 5:00 p.m. Pacific time. Issues will be posted on GitHub and participants are expected to assign an issue to themselves and submit a PR within a couple of days for review. The issues are categorized into three levels of difficulty, easy, medium, and advanced. If this is your first time contributing to PyTorch, we recommend starting with an issue at the easy level. Before submitting your first PR, you will need to fill out the contributor license agreement or CLA which will pop up before you submit your first PR. You'll also need to review the code of conduct which can be found on the docathon issues in the particular repos on GitHub. All participants need to abide by the code of conduct. We'll also be using Discord for communication during the event in addition to comments on the PRs. This will allow us to stay connected and work together more effectively. We'll also have a kickoff event on the Discord on June 3rd at 10:00 a.m. Pacific. So, be sure to join us there for a live kickoff event and Q&A session. To make things more exciting, we'll be keeping track of contributions with the leaderboard. Top contributors will be recognized for their efforts, so bring your best work. Next, I'll hand it over to Ivan to talk more about PyTorch tutorials. Hi, we've seen a lot of good work done in past Docathons in the tutorials repo. So, we figured it would be a good idea to bring it back for this Docathon. I'm just going to go over how to do contributions for this reple. Head over to the PyTorch tutorials reple, find the issue labeled Docathon H12025. This issue contains information you need to know on how to pick an issue and submit it. You should focus on issues with the Docathon H1 2025 label as issues without this label will not be considered for this Dockathon. Each issue has a difficulty label as my colleague Alena mentioned. And if this is the first time, we suggest picking an issue with a lower difficulty. Once you find an issue, assign it to yourself by adding an assigned tome comment and that is slash assigned to me. If the issue is already assigned, move on to another. Once you have successfully assigned yourself, get working on the issue as soon as possible and get your PR in. Once you send in a PR, someone from the PyTorch team will review it. Please respond to any feedback you get in a timely manner in a timely manner. Please only take on two issues at a time. We'll unassign extras if you have more than two or if we do not see any activity on an issue in a while. I will now hand over to Fedlana to talk about the PyTorch PyTorch repo. Thanks, Simon. I will share how you can tribute to the main PyTorch rep during the Docathon. This time we are focusing on moving our docs from restructured text to MIS flavored markdown. We've heard it's tricky to write in restructured text. So we are switching to make it easier. We'll still use sphinx and mism markdown uh lets us keep sphinx features we need such as autod do and auto summary. This task might seem simple but p passing pytor doc test can be quite challenging. We ask you to submit one pr at a time to keep it manageable for it and feel free to use lms to help with this work. Apart from RST Migration, we have a number of issues in our regular backlog uh that you can check out. Look for issues with the Dakathon H1 2025 label. We'll accept full request until June uh 15 at uh 5:00 p.m. Pacific time. Uh after that, we'll take a couple of days to tally points and announce the winners in a blog post. Um, and of course there will be both digital and physical swag for the winners, including PyTorch conference tickets. Make sure you join our Discord server so you can communicate with us and ask questions as needed. Thank you so much for contributing to PyTorch