Does anyone know how these models fare in terms of multilingual real-world usage? I’ve used previous iterations of llama models and they all seemed to be lackin

Does anyone know how these models fare in terms of multilingual real-world usage? I’ve used previous iterations of llama models and they all seemed to be lacking in that regard.