Merge branch 'master' of https://gitea.thecrazyinsanity.win/thecrazyinsanity/nixos-configv3
This commit is contained in:
@@ -48,6 +48,7 @@
|
||||
pkgs.opencode
|
||||
pkgs.llama-cpp
|
||||
pkgs.llama-swap
|
||||
# pkgs.agent-browser
|
||||
|
||||
inputs.llm-agents.packages.${pkgs.stdenv.hostPlatform.system}.pi
|
||||
inputs.llm-agents.packages.${pkgs.stdenv.hostPlatform.system}.rtk
|
||||
@@ -90,6 +91,12 @@
|
||||
# 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:
|
||||
"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}'"
|
||||
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
|
||||
|
||||
Reference in New Issue
Block a user