druub

Ever since AlphaEvolve - the idea that if you build a harness which can evaluate solutions and give LLMs a database where they can keep storing their work and then sample from it - they do find non-trivial solutions over time leaning from their own past ideas.

It is the ultimate manifestation of test-time scaling. I think karpathy just popularised it.

What are reasonable alternatives to arXiv? It has become increasinbgly slow. Techrxiv?

sorry for your loss and thanks for sharing. At times I think I don't have the strength to go on after something like this happens to me. But so happy that you are able to still find joy.

What do people do with synthetically generated data? Use it to train models? I doubt if models will improve by appending redundant data since they use the training data to estimate distributions. Just curious.