about 2 years ago
Did you know?!
The Global PyTorch Summer Hackathon had some amazing submissions including
Even though I have been an avid user of PyTorch for years, I learned so much during the PyTorch Summer Hackathon. It all started with a simple problem that popped up in my own research, that quickly grew into a project I felt the community might enjoy as well. The hackathon gave me the opportunity to look at how PyTorch works under the hood, especially data-loaders for my entry Torchmeta. PyTorch is moving fast and has an incredible support, so don't be afraid to get your hands dirty.
Another lesson I learned from this project is how important the design choices are, and how you should learn from well-established projects. In the case of Torchmeta, I was deeply inspired by the amazing work the teams behind Torchvision and OpenAI Gym did, and how they designed their API to make it as easy as possible to use these (now standard) tools. Make sure to get feedback from your team and your friends throughout the creative process: I had the chance to work with friends who gave me valuable feedback during every step of the development. This can surprisingly take more time than actually writing code, but it will be very helpful.
Do you have a tip you’d like to share with your fellow devs? Be sure to share it in the Hackathon Slack Channel, we know they’ll appreciate it!
If you have any questions about the hackathon, please post on the discussion forum.