Merge pull request 'remove qwen3.5-coder-next' (#200) from feature/ollama-local into main

Reviewed-on: #200

modules/kubernetes.nix

@@ -46,10 +46,10 @@
       kubernetes
     ];
 
-    # Enable containerd for Kubernetes
-    virtualisation.containerd.enable = true;
+    ## Enable containerd for Kubernetes
+    #virtualisation.containerd.enable = true;
 
-    # Enable kubelet
+    ## Enable kubelet
     #services.kubelet = {
     #  enable = true;
     #  extraFlags = {

systems/artemision/configuration.nix

@@ -1,6 +1,7 @@
 {
   lib,
   pkgs,
+  config,
   ...
 }:
 {
@@ -17,6 +18,7 @@
     ./stylix.nix
     ./wifi.nix
     ./zerotier.nix
+    ../palatine-hill/ollama.nix
   ];
 
   time.timeZone = "America/New_York";
@@ -39,6 +41,19 @@
   sops.age.sshKeyPaths = [ "/etc/ssh/ssh_host_ed25519_key" ];
 
   services = {
+    ollama = {
+      package = lib.mkForce pkgs.ollama-rocm;
+      models = lib.mkForce "${config.services.ollama.home}/models";
+      loadModels = lib.mkForce [
+        "deepseek-r1:1.5b"
+        "lennyerik/zeta"
+        "nomic-embed-text:latest"
+        "glm-4.7-flash"
+        "magistral"
+        "devstral-small-2"
+        "starcoder2:7b"
+      ];
+    };
     flatpak.enable = true;
     calibre-web = {
       # temp disable this

systems/artemision/programs.nix

@@ -27,6 +27,7 @@
     fd
     file
     firefox
+
     # gestures replacement
     git
     glances
@@ -34,12 +35,8 @@
     grim
     htop
     hwloc
-    ipmiview
     iperf3
     # ipscan
-    javaPackages.compiler.temurin-bin.jdk-25
-    javaPackages.compiler.temurin-bin.jdk-21
-    javaPackages.compiler.temurin-bin.jdk-17
     jp2a
     jq
     kdePackages.kdenlive

systems/palatine-hill/configuration.nix

@@ -17,7 +17,6 @@
     ./minio.nix
     ./networking.nix
     ./nextcloud.nix
-    ./ollama.nix
     #./plex
     ./postgresql.nix
     ./samba.nix
@@ -26,10 +25,8 @@
 
   programs.git.lfs.enable = false;
 
-  nixpkgs.config = {
-    packageOverrides = pkgs: {
-      vaapiIntel = pkgs.vaapiIntel.override { enableHybridCodec = true; };
-    };
+  nixpkgs.config.packageOverrides = pkgs: {
+    vaapiIntel = pkgs.vaapiIntel.override { enableHybridCodec = true; };
   };
 
   boot = {
@@ -100,13 +97,6 @@
     smartd.enable = true;
     calibre-server.enable = false;
 
-    # Kubernetes example configuration
-    # To enable Kubernetes, uncomment the following:
-    # kubernetes = {
-    #   enable = true;
-    #   clusterName = "palatine-hill-cluster";
-    #   controlPlaneEndpoint = "localhost:6443";
-    # };
   };
 
   nix.gc.options = "--delete-older-than 150d";

systems/palatine-hill/ollama.nix

@@ -15,27 +15,26 @@ in
         "deepseek-r1:1.5b"
         "deepseek-r1:32b"
         "deepseek-r1:70b"
-        "qwen3"
-        #"qwen3-coder-next"
+        #"qwen3"
+        #"qwen3.5:latest"
         "qwen3-coder-next"
         "lennyerik/zeta"
-        "qwen2.5-coder:14b"
-        "qwen2.5-coder:32b"
         "nomic-embed-text:latest"
-        "llama4:scout"
-        "mistral:7b"
-        "minimax-m2.7:cloud"
         "lfm2:24b"
         "glm-4.7-flash"
+        "nemotron-cascade-2:30b"
+        "magistral"
+        "devstral-small-2"
+        "starcoder2:15b"
       ];
       models = vars.primary_ollama;
       environmentVariables = {
         FLASH_ATTENTION = "1";
-        OLLAMA_KV_CACHE_TYPE = "q8_0";
+        OLLAMA_KV_CACHE_TYPE = "q4_0";
         # Ollama memory configuration
         OLLAMA_MAX_LOADED_MODELS = "3";
         OLLAMA_MAX_QUEUE = "512";
-        OLLAMA_NUM_PARALLEL = "3";
+        OLLAMA_NUM_PARALLEL = "1";
 
         # ROCm memory optimization
         #HIP_VISIBLE_DEVICES = "0";

systems/selinunte/programs.nix

@@ -31,7 +31,6 @@
     grim
     htop
     hwloc
-    ipmiview
     iperf3
     # ipscan
     jp2a

users/alice/non-server.nix

@@ -63,8 +63,81 @@
         "latex"
         "terraform"
         "log"
+        "context7-mcp-server"
+        "github-mcp-server"
       ];
       userSettings = {
+        context_servers = {
+          nixos = {
+            command = "nix";
+            args = [
+              "run"
+              "github:utensils/mcp-nixos"
+              "--"
+            ];
+          };
+        };
+        language_models = {
+          ollama = {
+            api_url = "http://192.168.76.2:11434";
+            context_window = 128000;
+            # global keep alive doesnt work
+            #keep_alive = "15m";
+            available_models = [
+              {
+                name = "deepseek-r1:1.5b";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "deepseek-r1:32b";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "deepseek-r1:70b";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "qwen3-coder-next";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "lennyerik/zeta";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "nomic-embed-text:latest";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "lfm2:24b";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "glm-4.7-flash";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "nemotron-cascade-2:30b";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+              {
+                name = "magistral";
+                max_tokens = 128000;
+                keep_alive = "15m";
+              }
+
+            ];
+          };
+        };
         colorize_brackets = true;
         hard_tabs = false;
         vim_mode = true;
@@ -77,7 +150,7 @@
         agent = {
           default_model = {
             provider = "ollama";
-            model = "qwen2.5-coder:latest";
+            model = "glm-4.7-flash";
           };
           favorite_models = [ ];
           model_parameters = [ ];
@@ -89,13 +162,16 @@
         journal = {
           hour_format = "hour24";
         };
-        edit_preditions = {
-          provider = "open_ai_compatible_api";
-          open_ai_compatible_api = {
-            api_url = "http://localhost:11434/v1/completions";
-            model = "zeta:latest";
-            prompt_format = "infer";
+        edit_predictions = {
+          provider = "ollama";
+          ollama = {
+            #api_url = "http://192.168.76.2:11434/v1/completions";
+            api_url = "http://192.168.76.2:11434";
+            context_window = 128000;
+            model = "lennyerik/zeta";
+            prompt_format = "qwen";
             max_requests = 64;
+            max_output_tokens = 256;
           };
         };
         texlab = {
@@ -144,6 +220,8 @@
     # markdown
     nodePackages.markdownlint-cli
 
+    # insert essential rust dependencies
+
     # doom emacs dependencies
     yaml-language-server
     nodePackages.typescript-language-server
@@ -184,5 +262,6 @@
 
     # arch zed deps
     nixd
+    uv
   ];
 }