README is in my opinion (author here) the most interesting - I wrote it to help others build useful mental model to be able to recreate the project yourself, without need to even read my code
Thanks for sharing this. As someone currently researching LLMs, I'm sure I'll be referencing this quite a bit going forward.
The lesson-style README is a great approach. Breaking down LLM inference into digestible steps makes the codebase approachable even for people who haven't touched CUDA before.
Looks interesting, it reminds me of the first llama.cpp, but better documented.
I love the documentation formatted in lessons. I can't wait to read through it.
Very nice job on read me.<p>>>Physically, LLM is a file which contains a lot of float numbers.<p>aka atoms of the LLM.
Wanted to add that the author has an amazing blog with lots of interesting papers: <a href="https://jedrzej.maczan.pl/" rel="nofollow">https://jedrzej.maczan.pl/</a>
It seems the author believes checking the return values of CUDA API calls is not "tiny" enough :-(
[dead]
[dead]