Closed jacohend closed 8 months ago
gpu_split: 10, # interpreted as divided by 10f for the gpu split cuda_device: 0,
gpu_backend: ggml::accelerator::Backend::Gpu,