LMstudio support for bottlenecks (vulkan llama.Cpp)

[rpi-rgb]

Just a translation layer between newelle tools and LMstudio engine.

it's a newelle extension that allow models loaded through openai api but hosted by LMstudio to actually use tools while running on Vulkan Llama.Cpp v1.44.0.

allowing decent use of gpt-oss-20b with a 680m iGPU & 32Go RAM (less could be good)

(some crash from Newelle using it, just reopen and regenerate answer)

https://github.com/rpi-rgb/Newelle-extension-tool-use-Lmstudio-vulkan-support

[FrancescoCaracciolo]

Add the repo to the newelle-extensions topic

[rpi-rgb]

lms just updated in beta channel and no need of extension now