Oh, that means its a llama architecture model! Is the tokenizer the same? It may "work" without actually working optimally until llama.cpp patches it in. And

Oh, that means its a llama architecture model! Is the tokenizer the same? It may "work" without actually working optimally until llama.cpp patches it in. And the instruct model was just uploaded.