My understanding is if something is entirely absent from the dataset, good luck trying to teach a model that through post training
Presumably if it knows how to find and fix bugs it can be told how to exploit them. But who knows what it will actually be capable of doing in the real world as opposed to Kiki’s benchmarks
Given that most of their improvements here are positioned as RL on their last base train, seems like just doing some additional RL would work fine?