yeah so training was done for first two chapters, started to gather resources to have in each chapter, probably going to move DiT to be last chapter, because it’s only useful in bigger projects and is just implementation of architecture from ml paper. Maybe today I will finish DiT, latent and start to give some notes to learning it in readme(maybe, but I have to do my mandarin HW too)
So I done the first folder(solution), debug all of it, trained model, then I fixed file names in rest of repo, and put some work on readme
So summing up work I did through research and ai olympiad camp(10+ hours💀). Learned math behind basics of flow matching, implemented a view variants, created an outline to work on to master the tool. In future I plan on making full on tutorial to flow matching in one place.
