Hi all! I’m graduating from undergrad in a few weeks and have been eyeing contributing to PyTorch for awhile. Through work I’ve gained exposure to a pretty decent chunk of the Python frontend including distributed as well as writing a few custom operators in C++/CUDA (haven’t dived into Triton yet). I have a pretty good background regarding computer performance and the low level details in general and was wondering if anybody could point me towards issues or features that might be appropriate for my skill set. Happy to provide any other information as requested!
The “good first issue” category could be a good starter to pick some tasks.
Let me know if you can find something interesting.
Found a few that I think would be of interest for me, I’m going to hold off on grabbing for now as I have outside stuff to finish up over the next few weeks. Param spec typing, or Reduce torch.compile default logs all sound great, especially as a way to get a good feel for the codebase. I appreciate the kind reply, looking forward to contributing!
1 Like
That sounds great! Let me know how it goes once you find the time.
1 Like