yeah, RLVR is still nascent and hence there's lots of noise.
How can these spurious rewards possibly work? Can we get similar gains on other models with broken rewards?
it's because in those cases, RLVR merely elicits the reasoning strategies already contained in the model through pre-training
this paper, which uses Reasoning gym, shows that you need to train for way longer than those papers you mentioned to actually uncover novel reasoning strategies: https://arxiv.org/abs/2505.24864
I personally think that Gemini 2.5 Pro's superiority comes from having hundreds or thousands RL tasks (without any proof whatsoever, so rather a feeling).
Given that GDM pioneered RL, that's a reasonable assumption
hot take: i don't think you even need to understand much linear algebra/calculus to understand what a transformer does. like the math for that could probably be learned within a week of focused effort.