TheCrazyInsanity/nixos-configv3
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

ourgh

Browse Source
This commit is contained in:
TheCrazyInsanity
2026-06-21 21:34:23 -04:00
parent f3d27c2f33
commit 72747120fa
6 changed files with 59 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
machines/homepc/configuration.nix
View File

@@ -46,7 +46,7 @@
# And this config is getting a bit complicated for a single pc config
# Should be moved to it's own shit
environment.systemPackages = [
pkgs.ollama-cuda
# pkgs.ollama-cuda
pkgs.opencode
pkgs.llama-cpp
pkgs.llama-swap
@@ -57,16 +57,16 @@
inputs.llm-agents.packages.${pkgs.stdenv.hostPlatform.system}.rtk
];
services.ollama = {
enable = true;
package = pkgs.ollama-cuda;
environmentVariables = {
OLLAMA_NUM_PARALLEL = "1";
OLLAMA_FLASH_ATTENTION = "1";
OLLAMA_KV_CACHE_TYPE = "q4_0";
OLLAMA_CONTEXT_LENGTH = "16384";
};
};
# services.ollama = {
# enable = true;
# package = pkgs.ollama-cuda;
# environmentVariables = {
# OLLAMA_NUM_PARALLEL = "1";
# OLLAMA_FLASH_ATTENTION = "1";
# OLLAMA_KV_CACHE_TYPE = "q4_0";
# OLLAMA_CONTEXT_LENGTH = "16384";
# };
# };
# Configure llama-swap as a systemd service
systemd.services.llama-swap = {
@@ -124,11 +124,11 @@
"Qwen3.5-4B":
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-4B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256"
ttl: 300
"Gemma4-E4B":
cmd: "llama-server --port ''${PORT} -hf Abhiray/gemma-4-E4B-it-heretic-GGUF:Q4_K_M --ctx-size 64000 --temp 0.7 --top-p 0.8 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256"
"Gemma4-E2B":
cmd: "llama-server --port ''${PORT} -hf unsloth/gemma-4-E2B-it-qat-GGUF:UD-Q4_K_XL --spec-type draft-mtp --spec-draft-n-max 4 --ctx-size 64000 --temp 1.0 --top-p 0.95 --top-k 64 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256 --chat-template-kwargs '{\"enable_thinking\": true}'"
ttl: 300
"Qwen3.5-0.8B-Non-Thinking":
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.7 --top-p 0.8 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256"
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.7 --top-p 0.8 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256 --chat-template-kwargs '{\"enable_thinking\": false}'"
ttl: 300'';
# Set your time zone.
@@ -146,13 +146,13 @@
# useXkbConfig = true; # use xkb.options in tty.
# };
services.hermes-agent = {
enable = true;
# settings.model.default = "anthropic/claude-sonnet-4";
environmentFiles = [ config.sops.secrets."hermes-env".path ];
addToSystemPackages = true;
configFile = /etc/hermes/config.yaml;
};
# services.hermes-agent = {
# enable = true;
# # settings.model.default = "anthropic/claude-sonnet-4";
# environmentFiles = [ config.sops.secrets."hermes-env".path ];
# addToSystemPackages = true;
# configFile = /etc/hermes/config.yaml;
# };
services.xserver.enable = true; # On anything else modern this should be false.