llama.cpp sampler selection

[nsvrana]

Description

Please enable, through the CLI or UI, the ability to select different samplers and adjust their parameters when using llama.cpp. In particular, mirostat is showing a lot of promise for improving the quality of text generation.

Additional Context

Here's the pull request on the llama.cpp side: ggml-org/llama.cpp#1126

[oobabooga]

I have added mirostat in a recent commit, so I'll close this issue