Prerequisites

• I am running the latest code. Mention the version if possible as well.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new and useful enhancement to share.

Feature Description

As of 0.9.1, there are several options in the help that are not recognized.

--seed
--min-p
--top-k
--top-p
--samplers

Additionally, some args are partially supported:

• --prompt is unknown but -p works.
• --file is unknown but -f works.

I assume the help is forked from llama.cpp or is somehow linked.

Motivation

Models like QwQ-32B have recommended samplers from Qwen so I would like to be able to use their recommended settings to stop looping at the end of a reply.

Possible Implementation

I noted #692 is about --repeat-penalty wasn't supported in the 0.8 stream but is fixed as of 0.9.1 so I assume these args need to be added to some list. I think that issue can be closed.

There may be more but these are the args that are not supported at all that I have found.