From 5801e5f800c3c6b2acb442b65ec6c3416d12e06a Mon Sep 17 00:00:00 2001
From: crowetic <jason@crowetic.com>
Date: Mon, 19 May 2025 09:43:39 -0700
Subject: [PATCH] Fixed incorrect localai setup

---
 setup-ai-stack.sh | 136 +++++++++++++++++++++-------------------------
 1 file changed, 62 insertions(+), 74 deletions(-)

diff --git a/setup-ai-stack.sh b/setup-ai-stack.sh
index 2d08ed3..f2a30e2 100644
--- a/setup-ai-stack.sh
+++ b/setup-ai-stack.sh
@@ -2,6 +2,8 @@
 set -e
 
 mkdir -p ~/ai-stack
+mkdir -p ~/ai-stack/localai/models
+mkdir -p ~/ai-stack/localai/config
 LOG_FILE=~/ai-stack/setup.log
 exec > >(tee -a "$LOG_FILE") 2>&1
 
@@ -41,71 +43,45 @@ if [[ "$1" == "--update" ]]; then
   else
     echo "⚠️  JupyterLab is not running, skipping update."
   fi
-  
-  # # Whisper (whisper.cpp)
-  # if check_container_running "whisper"; then
-  #   echo "🧼 Updating Whisper..."
-  #   docker stop whisper
-  #   docker rm whisper
-  #   docker pull ggerganov/whisper.cpp:latest
-  #   docker run -d \
-  #     --name whisper \
-  #     --network ai-stack-net \
-  #     --gpus all \
-  #     -v whisper_data:/app/data \
-  #     ggerganov/whisper.cpp:latest
-  #   echo "✅ Whisper updated."
-  # else
-  #   echo "⚠️  Whisper is not running, skipping update."
-  # fi
-  
-  # # Stable Diffusion (A1111)
-  # if check_container_running "stable-diffusion"; then
-  #   echo "🧼 Updating Stable Diffusion..."
-  #   docker stop stable-diffusion
-  #   docker rm stable-diffusion
-  #   docker pull abraham-ai/automatic1111-webui:latest
-  #   docker run -d \
-  #     --name stable-diffusion \
-  #     --gpus all \
-  #     -p 7860:7860 \
-  #     -v sd_models:/data \
-  #     abraham-ai/automatic1111-webui:latest
-  #   echo "✅ Stable Diffusion updated."
-  # else
-  #   echo "⚠️  Stable Diffusion is not running, skipping update."
-  # fi
-  
-  # echo "🔁 Updating ComfyUI..."
-  # docker stop comfyui 2>/dev/null || true
-  # docker rm comfyui 2>/dev/null || true
-  # docker pull ghcr.io/cmdr2/comfyui:latest
-  # docker run -d --gpus all \
-  #   --name comfyui \
-  #   --network ai-stack-net \
-  #   -p 8188:8188 \
-  #   -v comfyui_data:/workspace \
-  #   ghcr.io/cmdr2/comfyui:latest
-  # echo "✅ ComfyUI updated and restarted."
 
-  # echo "🔁 Updating LocalAI..."
-  # docker stop localai 2>/dev/null || true
-  # docker rm localai 2>/dev/null || true
 
-  mkdir -p ~/ai-stack/localai
+  # 🔁 Updating LocalAI (GPU build)
+  if check_container_running "localai"; then
+    echo "🧠 Updating LocalAI..."
+    docker stop localai
+    docker rm localai
+  else
+    echo "➕ Installing LocalAI..."
+  fi
 
-  docker pull quay.io/go-skynet/local-ai:latest
-  docker run -d \
-    --name localai \
-    --network ai-stack-net \
-    -p 8080:8080 \
-    -v ~/ai-stack/localai:/models \
-    -e MODELS_PATH=/models \
-    -e ENABLE_OOLLAMA_BACKEND=true \
-    -v /var/run/docker.sock:/var/run/docker.sock \
-    quay.io/go-skynet/local-ai:latest
+  docker pull localai/localai:latest-aio-gpu-nvidia-cuda-12
+
+  # Ensure default models.yaml exists
+  MODELS_YAML=~/ai-stack/localai/config/models.yaml
+  if [ ! -f "$MODELS_YAML" ]; then
+    echo "📄 Creating default models.yaml..."
+cat > "$MODELS_YAML" <<EOF
+- name: phi
+  backend: llama-cuda
+  parameters:
+    model: /models/phi-2.gguf
+EOF
+  fi
+
+docker run -d \
+  --name localai \
+  --gpus all \
+  -p 8080:8080 \
+  -v ~/ai-stack/localai/models:/models \
+  -v ~/ai-stack/localai/config:/etc/local-ai \
+  --health-cmd="curl --fail http://localhost:8080/health || exit 1" \
+  --health-interval=30s \
+  --health-retries=3 \
+  --restart unless-stopped \
+  localai/localai:latest-aio-gpu-nvidia-cuda-12
+
+echo "✅ LocalAI running at http://localhost:8080"
 
-  echo "✅ LocalAI updated and restarted."
 
   
   	echo "🔁 Updating Whisper.cpp (custom GPU build)..."
@@ -447,26 +423,37 @@ if [[ "$1" != "--update" ]]; then
   if check_container_running "localai"; then
     echo "✅ Skipping LocalAI install (already running)."
   else
-    read -p "➕ Install LocalAI (OpenAI-compatible API proxy)? [y/N]: " lai
-    if [[ "$lai" =~ ^[Yy]$ ]]; then
-      echo "🧠 Setting up LocalAI..."
+    read -p "➕ Install LocalAI (GPU-accelerated)? [y/N]: " localai_prompt
+    if [[ "$localai_prompt" =~ ^[Yy]$ ]]; then
+      echo "🧠 Installing LocalAI (NVIDIA CUDA 12)..."
+      
+      mkdir -p ~/ai-stack/localai/models
+      mkdir -p ~/ai-stack/localai/config
 
-      docker stop localai 2>/dev/null || true
-      docker rm localai 2>/dev/null || true
+      echo "📄 Creating default models.yaml..."
+      cat > ~/ai-stack/localai/config/models.yaml <<EOF
+- name: phi
+  backend: llama-cuda
+  parameters:
+    model: /models/phi-2.gguf
+EOF
 
-      mkdir -p ~/ai-stack/localai
+      docker pull localai/localai:latest-aio-gpu-nvidia-cuda-12
 
       docker run -d \
         --name localai \
-        --network ai-stack-net \
+        --gpus all \
         -p 8080:8080 \
-        -v ~/ai-stack/localai:/models \
-        -e MODELS_PATH=/models \
-        -e ENABLE_OOLLAMA_BACKEND=true \
-        -v /var/run/docker.sock:/var/run/docker.sock \
-        quay.io/go-skynet/local-ai:latest
+        -v ~/ai-stack/localai/models:/models \
+        -v ~/ai-stack/localai/config:/etc/local-ai \
+        --env ENABLE_BACKENDS=llama-cuda,ollama
+        --health-cmd="curl --fail http://localhost:8080/health || exit 1" \
+        --health-interval=30s \
+        --health-retries=3 \
+        --restart unless-stopped \
+        localai/localai:latest-aio-gpu-nvidia-cuda-12
 
-      echo "🌐 LocalAI running at http://localhost:8080"
+      echo "🌐 LocalAI is now running at http://localhost:8080"
     fi
   fi
 fi
@@ -474,5 +461,6 @@ fi
 
 
 
+
 echo "✅ AI stack setup complete."