server-config/services/llama-cpp.nix

36 lines
1.1 KiB
Nix

{
pkgs,
service_configs,
config,
inputs,
lib,
...
}:
{
services.llama-cpp = {
enable = true;
model = builtins.toString (
pkgs.fetchurl {
url = "https://huggingface.co/rodrigomt/Qwen3-30B-A3B-Thinking-Deepseek-Distill-2507-v3.1-V2-GGUF/resolve/main/Qwen3-30B-A3B-Thinking-Deepseek-Distill-2507-v3.1-V2-UD-Q4_K_XL.gguf";
sha256 = "1a3abffc8463041e24cdc43af26c99b6cfab1d2ee78fef0d793033ec0e5b58aa";
}
);
port = service_configs.ports.llama_cpp;
host = "0.0.0.0";
# vulkan broken: https://github.com/ggml-org/llama.cpp/issues/13801
package = (lib.optimizePackage inputs.llamacpp.packages.${pkgs.system}.default);
extraFlags = [
# "-ngl"
# "9999"
];
};
# have to do this in order to get vulkan to work
systemd.services.llama-cpp.serviceConfig.DynamicUser = lib.mkForce false;
services.caddy.virtualHosts."llm.${service_configs.https.domain}".extraConfig = ''
${builtins.readFile ../secrets/caddy_auth}
reverse_proxy :${builtins.toString config.services.llama-cpp.port}
'';
}