HN Reader

"Starting from a single base LLM"

Ok, zero data, except the data used in the teacher model.

For values of zero quite far above zero.

Perpetual Motion Machines were a thing at some point, too.

Terrible choice of name. DeepSeek developed a historically important model called “R-Zero” (this was the predecessor to R1 that was training without any coldstart SFT, and was very strong but difficult to read chain of thought because it code switches into Chinese and has no line breaks).

I think in formal domain like lean it should actually be possible to do it from zero--but seems like no major successes no far

OK but how do you ensure it's improving in a direction that aligns with reality?

I still don't understand what a "reasoning" LLM is

Now gamify it.

What could go wrong?

Conceptually, it's effectively a GAN