[Feature request] Support for Korean-Chinese sakura model

[YuiTH]

I've tried SakuraLLM/LN-Korean-14B-v0.2 with Q8 quantization. There are two major problem preventing me using it with auto-novel:

1. There is no Korean original text support, especially website hardcoded prompt as "你是一个轻小说翻译模型，可以流畅通顺地以日本轻小说的风格将日文翻译成简体中文"
2. Korean Model often generate past paragraph as part of output. I'm not sure if it's a model bug or website bug. But I will put it here. For example:

Set past_text as zero will prevent this issue. Let me know if you need more information or test case here.