server: one shadow per mind on the serving context: each shadow keeps its own optimizer, /learn trains any loaded shadow, and unload releases it - #85

server: one shadow per mind on the serving context: each shadow keeps its own optimizer, /learn trains any loaded shadow, and unload releases it#85

joelteply wants to merge 5 commits into

codex/s4-resident-microstepsCambrianTech/llama.cpp:codex/s4-resident-microstepsfrom

cormac/per-mind-shadowCambrianTech/llama.cpp:cormac/per-mind-shadowCopy head branch name to clipboard