Feature: Easily modifying KV Cache and other settings for models and backends

[jxs714]

Hey guys,

Great job on the software. So far, to me this is the best local AI setup offering many tools and options compared to other similar softwares. The only issue I am having is the lack of options per model and engines without having to modify the YAML or manually modifying configuration files. I think you guys should 100% consider offering all available options as an easy toggle and text box to modify the settings per model and per backend such as KV Cache, Context Length, etc. Modifying these things manually in Yaml / config files is tedious if using many different models for testing and productivity.