Please help
I am a bit of a noob and am trying to get this text to image model working however whenever I start kobold and go to the image gen tab and put the flux1-dev.safetensors file and press launch I get an error. It says could not load image model. Also in my terminal it says:
ggml_cuda_init: failed to initialize CUDA: CUDA driver version is insufficient for CUDA runtime version
ggml_backend_cuda_init: invalid device 0
Error: KCPP SD Failed to create context!
If using Flux/SD3.5, make sure you have ALL files required (e.g. VAE, T5, Clip...) or baked in!
Load Image Model OK: False
Please help me out. I have no idea what files I need or what the problem is.
how generate image in 9:16 ratio?
Hello @go2mattdavis , it is likely an issue with "kobold." I can see this because you're seeing the CUDA errors, which are server-side. So try running it locally, or on a cloud platform if you can, noting that you likely need around 16 GB of VRAM to run it.
how generate image in 9:16 ratio?
You can do this by multiplying. The image model requires you to have both numbers divisible by 8. So try multiplying the ratios by 8, or 16... etc. Any multiple of 8. You can do it like this, 9:16 = 98, 916 (preserving ratio). Then you can do the math!