chore(halo): max models

2026-05-22 10:37:00 +02:00 · 2026-05-22 10:37:00 +02:00 · 87dfe74daa
commit 87dfe74daa
parent c845f7c385
2 changed files with 2 additions and 1 deletions
--- a/systems/x86_64-linux/halo/llama-server.nix
+++ b/systems/x86_64-linux/halo/llama-server.nix
@ -29,7 +29,7 @@
        "--host 0.0.0.0"
        "--port 8000"
        "--models-preset ${./models.ini}"
-        "--models-max 2"
+        "--models-max 3"
      ];
      Restart = "on-failure";
      RestartSec = 10;
--- a/systems/x86_64-linux/halo/models.ini
+++ b/systems/x86_64-linux/halo/models.ini
@ -54,5 +54,6 @@ min-p               = 0.0
 presence-penalty    = 1.5
 parallel            = 1
 c                   = 131072
+load-on-startup     = true
 chat-template-kwargs = {"preserve_thinking": true}