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

i haven't committed in a while lmaooo

Browse Source
This commit is contained in:
TheCrazyInsanity
2026-07-16 10:52:29 -04:00
parent a4ec52164e
commit bd99251883
7 changed files with 47 additions and 66 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
machines/homepc/configuration.nix
View File

@@ -54,6 +54,7 @@
pkgs.nvidia-container-toolkit
inputs.llm-agents.packages.${pkgs.stdenv.hostPlatform.system}.pi
inputs.llm-agents.packages.${pkgs.stdenv.hostPlatform.system}.omp
inputs.llm-agents.packages.${pkgs.stdenv.hostPlatform.system}.rtk
];
@@ -94,17 +95,20 @@
# As long as this is here the models are declarative. llama-server will grab them if not downloaded already.
environment.etc."llama-swap/config.yaml".text = ''
models:
"Bonsai-27B":
cmd: "llama-server -hf prism-ml/Bonsai-27B-gguf:Q1_0 --port ''${PORT} -np 1 --ctx-size 64000 -ngl 999 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 -ctk q8_0 -ctv q8_0"
ttl: 2400
"Qwen3.6-35B-A3B-GGUF":
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q4_K_S --ctx-size 128000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -np 1 --fit-target 256 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 --chat-template-kwargs '{\"preserve_thinking\": true}'"
ttl: 2400
"Qwen3.6-35B-A3B-FLASH":
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q3_K_M --ctx-size 128000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -np 1 --fit-target 256 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 --chat-template-kwargs '{\"preserve_thinking\": true}'"
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q3_K_M --ctx-size 128000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -np 1 --fit-target 256 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 --no-mmap --chat-template-kwargs '{\"preserve_thinking\": true}'"
ttl: 2400
"Qwen3.5-35B-A3B-GGUF":
cmd: llama-server --port ''${PORT} -hf mudler/Qwen3.5-35B-A3B-APEX-GGUF:Mini --ctx-size 128000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -np 1 --fit-target 256 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 --no-kv-offload
ttl: 2400
"Gemma4-26B-A4B":
cmd: llama-server --port ''${PORT} -hf mudler/gemma-4-26B-A4B-it-heretic-APEX-GGUF:Mini --ctx-size 128000 --temp 1.0 --top-p 0.95 --top-k 64 --min-p 0.00 -np 1 --fit-target 256 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 --no-kv-offload
cmd: llama-server --port ''${PORT} -hf unsloth/gemma-4-26B-A4B-it-qat-GGUF:UD-Q4_K_XL --ctx-size 128000 --temp 1 --top-p 0.95 --top-k 64 --min-p 0.00 -np 1 --fit-target 512 --ubatch-size 1024 -fa on --slots --slot-save-path /home/laythe/llamapcache --jinja -kvu --no-mmproj --swa-checkpoints 32 --no-mmap
ttl: 2400
"Qwen3-1.7B-GGUF":
cmd: llama-server --port ''${PORT} -hf unsloth/Qwen3-1.7B-GGUF:Q4_K_S --ctx-size 32768 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu
@@ -122,7 +126,7 @@
cmd: "llama-server --port ''${PORT} -hf Jackrong/Qwen3.5-9B-Claude-4.6-Opus-Reasoning-Distilled-GGUF:Q4_K_S --ctx-size 32000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 512 --chat-template-kwargs '{\"enable_thinking\": true}' --no-mmproj"
ttl: 300
"Qwen3.5-4B":
cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-4B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256"
cmd: "llama-server --port ''${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 --chat-template-kwargs '{\"enable_thinking\": false}'"
ttl: 300
"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}'"
@@ -146,14 +150,6 @@
# 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.xserver.enable = true; # On anything else modern this should be false.
# Configure keymap in X11