pls pr to official lcpp

[DarkReaperBoy]

https://huggingface.co/IEITYuan/Yuan2-M32-hf/discussions/3

[chong000]

Thank you for your suggestions.

Compared with the llama structure, the yuan2.0-models has some differences in model structure and computational implementation. Some of these differences cannot be implemented using native operatos(https://github.com/ggerganov/llama.cpp). Therefore, the implementation of some operators in this project has been changed;

Next, we will try to use the operators from the original project in order to merge it into the main branch of llama.cpp;