3 ms·

Well it's not really support of llama 2 if it has to be extensively finetuned to "convert" the model.