Deep learning works. Any other facts about this era or about me are relatively unimportant,¹The only close contender, I think, being the 1998 determination of the cosmological constant's positivity. But any cosmological discovery comes with troves of model uncertainty, which significantly diminishes its importance. but sometimes I write about them anyway. Hi, I'm Lovre!

Last few years I have mostly spent fascinated by the large language models.²The riches! Such simple structures holding so much complexity! Had I but a thousand years to spend on studying a model as simple as Llama 3 8b, it would not be enough. Of my public output, see my early investigations into in-context learning,³My work got cited in a few papers, and the later literature (see the many papers that cite Garg et al.) has been quite interesting, but I still feel a lot mystery in in-context learning. I am still thinking about what exactly would dispel the mystery for me. and a bit of benchmarking visual LLMs on Tetris.

Currently I'm working on a hard benchmark for LLM reasoning, and various little experiments with AI-to-AI communication, synthetic data, and am training some small language models.