Apertus is comparable to the 2024 Llama 3 model from Meta, according to SWI.
I mean that’s not great, but considering this seems to be one of the fairly trained and truly open models, it’s not too terrible either and should definitely be the path forward with LLMs.
It’s not terrible if it doesn’t answer in swiss german lol. It’s close enough that it makes me think I am at fault for not understanding what it’s saying before not making any sense on a second read.
I mean that’s not great, but considering this seems to be one of the fairly trained and truly open models, it’s not too terrible either and should definitely be the path forward with LLMs.
It’s not terrible if it doesn’t answer in swiss german lol. It’s close enough that it makes me think I am at fault for not understanding what it’s saying before not making any sense on a second read.
But german responses are perfectly fine.