Harald Hoyer ac70c57c15 chore(halo): preload both llama models and tune preset ... Preload Qwen3.6-27B and Qwen3.6-35B-A3B at startup (load-on-startup) so both are warm immediately under --models-max 2, set parallel = 1 as the [*] fallback for any other model, and adjust per-model context size and draft depth.		2026-05-20 07:14:26 +02:00
..
amd	feat(amd): enable Wake-on-LAN on enp7s0	2026-05-16 13:40:25 +02:00
attic	feat(headscale): add ACL policy, isolate mx, make mx an exit node	2026-05-13 09:06:40 +02:00
halo	chore(halo): preload both llama models and tune preset	2026-05-20 07:14:26 +02:00
mx	feat(mx): increase halo bot timeout	2026-05-19 23:52:46 +02:00
nixtee1	refactor(nix): extract common system configs into reusable modules	2026-01-30 10:42:09 +01:00
sgx	refactor(opencode): extract serve service into shared NixOS module	2026-05-05 13:43:27 +02:00
t15	refactor(nix): extract common system configs into reusable modules	2026-01-30 10:42:09 +01:00
x1	chore(x1,amd): disable cratedocs-mcp service	2026-05-13 11:35:59 +02:00