Merge pull request NVIDIA-AI-Blueprints#374 from NVIDIA-AI-Blueprints/rkharwar-nv-patch-4

Add files via upload

Author: rkharwar-nv

notebooks/launchable.ipynb

@@ -127,6 +127,16 @@
     "RAG_BASE_URL = f\"http://{IPADDRESS}:{RAG_SERVER_PORT}\"\n",
     "INGESTOR_BASE_URL = f\"http://{IPADDRESS}:{INGESTOR_SERVER_PORT}\"\n",
     "\n",
+    "# NIM services to deploy (excludes nim-llm and vlm-ms since we use NVIDIA-hosted endpoints)\n",
+    "NIM_SERVICES = (\n",
+    "    \"nemoretriever-embedding-ms \"\n",
+    "    \"nemoretriever-ranking-ms \"\n",
+    "    \"page-elements \"\n",
+    "    \"graphic-elements \"\n",
+    "    \"table-structure \"\n",
+    "    \"nemoretriever-ocr\"\n",
+    ")\n",
+    "\n",
     "\n",
     "# =============================================================================\n",
     "# DOCKER COMPOSE HELPERS\n",
@@ -520,9 +530,9 @@
     "    print(\"DEPLOYING NVIDIA RAG BLUEPRINT\")\n",
     "    print(\"=\" * 60)\n",
     "   \n",
-    "    print(\"\\n[1/4] NIM Microservices...\")\n",
-    "    docker_compose(\"deploy/compose/nims.yaml\", \"pull\", \"-q\")\n",
-    "    docker_compose(\"deploy/compose/nims.yaml\", \"up\", \"-d\")\n",
+    "    print(\"\\n[1/4] NIM Microservices (no local LLM/VLM)...\")\n",
+    "    docker_compose(\"deploy/compose/nims.yaml\", \"pull\", f\"-q {NIM_SERVICES}\")\n",
+    "    docker_compose(\"deploy/compose/nims.yaml\", \"up\", f\"-d {NIM_SERVICES}\")\n",
     "    print(\"-\" * 60)\n",
     "    print(\"\\n[2/4] Vector Database...\")\n",
     "    docker_compose(\"deploy/compose/vectordb.yaml\", \"pull\", \"-q\")\n",
@@ -667,42 +677,42 @@
     "    print(\"    ❌ FAIL: nvidia-smi not found\")\n",
     "\n",
     "# ─────────────────────────────────────────────────────────────────────────────\n",
-    "# [3] NVIDIA Driver Version (need 560+)\n",
+    "# [3] NVIDIA Driver Version (need 580+)\n",
     "# ─────────────────────────────────────────────────────────────────────────────\n",
-    "print(\"\\n[3] NVIDIA Driver Version (need 560+):\")\n",
+    "print(\"\\n[3] NVIDIA Driver Version (need 580+):\")\n",
     "try:\n",
     "    result = subprocess.run([\"nvidia-smi\", \"-q\"], capture_output=True, text=True)\n",
     "    if result.returncode == 0:\n",
     "        match = re.search(r\"Driver Version\\s*:\\s*(\\d+)\", result.stdout)\n",
     "        if match:\n",
     "            driver_version = int(match.group(1))\n",
     "            print(f\"    Driver Version: {driver_version}\")\n",
-    "            if driver_version >= 560:\n",
+    "            if driver_version >= 580:\n",
     "                print(\"    ✅ PASS\")\n",
     "            else:\n",
-    "                errors.append(f\"Driver version {driver_version} < 560 required\")\n",
-    "                print(f\"    ❌ FAIL: Version {driver_version} < 560\")\n",
+    "                errors.append(f\"Driver version {driver_version} < 580 required\")\n",
+    "                print(f\"    ❌ FAIL: Version {driver_version} < 580\")\n",
     "        else:\n",
     "            print(\"    Unable to parse driver version\")\n",
     "except FileNotFoundError:\n",
     "    print(\"    ❌ FAIL: nvidia-smi not found\")\n",
     "\n",
     "# ─────────────────────────────────────────────────────────────────────────────\n",
-    "# [4] CUDA Version (need 12.4+)\n",
+    "# [4] CUDA Version (need 13+)\n",
     "# ─────────────────────────────────────────────────────────────────────────────\n",
-    "print(\"\\n[4] CUDA Version (need 12.4+):\")\n",
+    "print(\"\\n[4] CUDA Version (need 13+):\")\n",
     "try:\n",
     "    result = subprocess.run([\"nvidia-smi\", \"-q\"], capture_output=True, text=True)\n",
     "    if result.returncode == 0:\n",
     "        match = re.search(r\"CUDA Version\\s*:\\s*(\\d+\\.\\d+)\", result.stdout)\n",
     "        if match:\n",
     "            cuda_version = float(match.group(1))\n",
     "            print(f\"    CUDA Version: {cuda_version}\")\n",
-    "            if cuda_version >= 12.4:\n",
+    "            if cuda_version >= 13:\n",
     "                print(\"    ✅ PASS\")\n",
     "            else:\n",
-    "                errors.append(f\"CUDA version {cuda_version} < 12.4 required\")\n",
-    "                print(f\"    ❌ FAIL: Version {cuda_version} < 12.4\")\n",
+    "                errors.append(f\"CUDA version {cuda_version} < 13 required\")\n",
+    "                print(f\"    ❌ FAIL: Version {cuda_version} < 13\")\n",
     "        else:\n",
     "            print(\"    Unable to parse CUDA version\")\n",
     "except FileNotFoundError:\n",
@@ -1525,6 +1535,7 @@
    "metadata": {},
    "source": [
     "### 4.5 Study Extracted Results\n",
+    "⚠️ **Only run this cell if you ran cell 2.3a Save Extracted Content**\n",
     "\n",
     "When `APP_NVINGEST_SAVETODISK=True` is enabled, the ingestion pipeline saves the extracted results to:\n",
     "```\n",