Depending on the language, personally, I haven't seen a great deal of hallucination with the models I use (llama 3.1 often). I wouldn't expect, say Spanish, to

Depending on the language, personally, I haven't seen a great deal of hallucination with the models I use (llama 3.1 often). I wouldn't expect, say Spanish, to be a problem. I think you'll find more potential issues when it comes to more domain-specific terminology.