News for hackers. Stuff that matters.
Re:Open-R1: an open reproduction of DeepSeek-R1 by brianstrimp on Tuesday January 28, 2025 @02:43AM (#42849855)
Only meaningful if code+data deterministically reproduce the weights.

At that point, the weights are just the cached output. Which has value since it's costly to produce from code+data.

Re:Open-R1: an open reproduction of DeepSeek-R1 by frabcus on Tuesday January 28, 2025 @03:54AM (#42850232)
I don't think it needs to be deterministic - and if it isn't, having the data and code becomes even more important!

Compilers generally aren't deterministic (see the reproduceable build movement), yet we still use their output binaries.