2 ms·

Our primary focus is on RL post-training. We think that is the best way to get the model to be a strong interactive agent.

• comex 11mo ago
  So, yes, but you won’t say what the base model is? :)
  • typpilol 11mo ago
    It seems like a sort of sonnet model as a lot of people are reporting it like to spam documentation on Twitter like sonnet 4.5