- Dec 19, 2022
- 31
- 52
Leaving this here for anyone that uses LMstudio+Gemma+Vulkan llama.cpp with Luna that is dealing with models crashing w/ exitCode=3221226505 after updating because I was ripping my hair out figuring why my machine suddenly couldn't run them or was translating super slow:
Go into configuration -> runtime -> runtime selection and in there select an older version of Vulkan Llama.cpp, if you're like me who has the bare minimum and avoids updating in fear it'll break everything, and can only run up to 2.28, for some reason it's causing models to not load (I've seen other people have this problem too), 2.27 runs just fine.
(as an addenum, if you're not using Vulkan as a default for GGUF and have some vram to offload to, highly rec to switch to Vulkan, it severely cuts down answer time.)
Go into configuration -> runtime -> runtime selection and in there select an older version of Vulkan Llama.cpp, if you're like me who has the bare minimum and avoids updating in fear it'll break everything, and can only run up to 2.28, for some reason it's causing models to not load (I've seen other people have this problem too), 2.27 runs just fine.
(as an addenum, if you're not using Vulkan as a default for GGUF and have some vram to offload to, highly rec to switch to Vulkan, it severely cuts down answer time.)
