Quick caveat: It's a task that the model isn't explicitly trained on, so test it thoroughly before us We'll run more benchmarks this week and share our findings soon! Help us test it better? 🙂

→ View original post on X — @reach_vb