Fix cells 8-10 to be self-contained

notebooks/00_environment_setup.ipynb

@@ -204,23 +204,7 @@
    "id": "776236f8",
    "metadata": {},
    "outputs": [],
-   "source": [
-    "# Create necessary directories\n",
-    "directories = [\n",
-    "    config.DATASETS_DIR,\n",
-    "    config.MODELS_DIR,\n",
-    "    config.ARTIFACTS_DIR,\n",
-    "    config.BASE_DIR / \"logs\",\n",
-    "    config.BASE_DIR / \"cache\",\n",
-    "]\n",
-    "\n",
-    "print(\"Creating directory structure...\")\n",
-    "for directory in directories:\n",
-    "    directory.mkdir(parents=True, exist_ok=True)\n",
-    "    print(f\"  \u2713 {directory}\")\n",
-    "\n",
-    "print(\"\\n\u2713 Directory structure ready!\")"
-   ]
+   "source": "from pathlib import Path\n\n# Define directories (self-contained)\nBASE_DIR = Path('..').resolve()\nDATASETS_DIR = BASE_DIR / 'datasets'\nMODELS_DIR = BASE_DIR / 'models'\nARTIFACTS_DIR = BASE_DIR / 'artifacts'\n\n# Create necessary directories\ndirectories = [\n    DATASETS_DIR,\n    MODELS_DIR,\n    ARTIFACTS_DIR,\n    BASE_DIR / 'logs',\n    BASE_DIR / 'cache',\n]\n\nprint('Creating directory structure...')\nfor directory in directories:\n    directory.mkdir(parents=True, exist_ok=True)\n    print(f'  \u2713 {directory}')\n\nprint('\\n\u2713 Directory structure ready!')\n"
   },
   {
    "cell_type": "markdown",
@@ -236,34 +220,7 @@
    "id": "6b854bac",
    "metadata": {},
    "outputs": [],
-   "source": [
-    "import json\n",
-    "\n",
-    "# Export configuration for other notebooks\n",
-    "notebook_config = {\n",
-    "    \"device\": str(DEVICE),\n",
-    "    \"python_version\": f\"{python_version.major}.{python_version.minor}.{python_version.micro}\",\n",
-    "    \"torch_version\": torch.__version__,\n",
-    "    \"cuda_available\": cuda_available,\n",
-    "    \"base_dir\": str(config.BASE_DIR),\n",
-    "    \"datasets_dir\": str(config.DATASETS_DIR),\n",
-    "    \"models_dir\": str(config.MODELS_DIR),\n",
-    "    \"artifacts_dir\": str(config.ARTIFACTS_DIR),\n",
-    "    \"random_state\": config.RANDOM_STATE,\n",
-    "    \"test_size\": config.TEST_SIZE,\n",
-    "    \"cv_folds\": config.CV_FOLDS,\n",
-    "    \"gemini_configured\": bool(config.GEMINI_API_KEY),\n",
-    "    \"huggingface_configured\": bool(config.HUGGINGFACE_TOKEN),\n",
-    "    \"created_at\": str(pd.Timestamp.now())\n",
-    "}\n",
-    "\n",
-    "config_path = config.BASE_DIR / \"notebook_config.json\"\n",
-    "with open(config_path, \"w\") as f:\n",
-    "    json.dump(notebook_config, f, indent=2)\n",
-    "\n",
-    "print(f\"\u2713 Configuration saved to: {config_path}\")\n",
-    "print(\"\\n\" + json.dumps(notebook_config, indent=2))"
-   ]
+   "source": "import json\nimport sys\nimport os\nfrom pathlib import Path\n\n# Get values (self-contained)\npython_version = sys.version_info\n\ntry:\n    import torch\n    torch_version = torch.__version__\n    cuda_available = torch.cuda.is_available()\n    if cuda_available:\n        DEVICE = 'cuda'\n    elif hasattr(torch.backends, 'mps') and torch.backends.mps.is_available():\n        DEVICE = 'mps'\n    else:\n        DEVICE = 'cpu'\nexcept ImportError:\n    torch_version = 'not installed'\n    cuda_available = False\n    DEVICE = 'cpu'\n\n# Load config\nconfig_json_path = Path('notebook_config.json')\nif config_json_path.exists():\n    with open(config_json_path, 'r') as f:\n        loaded_config = json.load(f)\nelse:\n    loaded_config = {}\n\nBASE_DIR = Path('..').resolve()\nDATASETS_DIR = BASE_DIR / 'datasets'\nMODELS_DIR = BASE_DIR / 'models'\nARTIFACTS_DIR = BASE_DIR / 'artifacts'\nRANDOM_STATE = loaded_config.get('random_state', 42)\nTEST_SIZE = loaded_config.get('test_size', 0.2)\nCV_FOLDS = loaded_config.get('cv_folds', 5)\n\n# Export configuration for other notebooks\nnotebook_config = {\n    'device': str(DEVICE),\n    'python_version': f'{python_version.major}.{python_version.minor}.{python_version.micro}',\n    'torch_version': torch_version,\n    'cuda_available': cuda_available,\n    'base_dir': str(BASE_DIR),\n    'datasets_dir': str(DATASETS_DIR),\n    'models_dir': str(MODELS_DIR),\n    'artifacts_dir': str(ARTIFACTS_DIR),\n    'random_state': RANDOM_STATE,\n    'test_size': TEST_SIZE,\n    'cv_folds': CV_FOLDS,\n}\n\nconfig_path = Path('notebook_runtime_config.json')\nwith open(config_path, 'w') as f:\n    json.dump(notebook_config, f, indent=2)\n\nprint(f'\u2713 Configuration exported to: {config_path.absolute()}')\nprint(json.dumps(notebook_config, indent=2))\n"
   },
   {
    "cell_type": "markdown",
@@ -279,24 +236,7 @@
    "id": "f409be56",
    "metadata": {},
    "outputs": [],
-   "source": [
-    "print(\"\\n\" + \"=\" * 60)\n",
-    "print(\"ENVIRONMENT SETUP COMPLETE\")\n",
-    "print(\"=\" * 60)\n",
-    "print(f\"\"\"\n",
-    "\u2705 Python: {python_version.major}.{python_version.minor}.{python_version.micro}\n",
-    "\u2705 Device: {DEVICE}\n",
-    "\u2705 PyTorch: {torch.__version__}\n",
-    "\u2705 Gemini API: {'Ready' if config.GEMINI_API_KEY else 'Not configured'}\n",
-    "\u2705 HuggingFace: {'Ready' if config.HUGGINGFACE_TOKEN else 'Not configured'}\n",
-    "\u2705 WebScraper API: Ready\n",
-    "\u2705 Directories: Created\n",
-    "\n",
-    "You can now proceed to the next notebook:\n",
-    "  \u2192 01_data_acquisition.ipynb\n",
-    "\"\"\")\n",
-    "print(\"=\" * 60)"
-   ]
+   "source": "import sys\nimport json\nimport os\nfrom pathlib import Path\n\npython_version = sys.version_info\n\ntry:\n    import torch\n    torch_version = torch.__version__\n    if torch.cuda.is_available():\n        DEVICE = 'cuda'\n    elif hasattr(torch.backends, 'mps') and torch.backends.mps.is_available():\n        DEVICE = 'mps'\n    else:\n        DEVICE = 'cpu'\nexcept ImportError:\n    torch_version = 'not installed'\n    DEVICE = 'cpu'\n\n# Load config\nconfig_json_path = Path('notebook_config.json')\nif config_json_path.exists():\n    with open(config_json_path, 'r') as f:\n        loaded_config = json.load(f)\nelse:\n    loaded_config = {}\n\nGEMINI_API_KEY = loaded_config.get('gemini_api_key') or os.getenv('GEMINI_API_KEY', '')\nHUGGINGFACE_TOKEN = os.getenv('HF_TOKEN', '')\n\nprint('\\n' + '=' * 60)\nprint('ENVIRONMENT SETUP COMPLETE')\nprint('=' * 60)\nprint(f'''\n\u2705 Python: {python_version.major}.{python_version.minor}.{python_version.micro}\n\u2705 Device: {DEVICE}\n\u2705 PyTorch: {torch_version}\n\u2705 Gemini API: {'Ready' if GEMINI_API_KEY else 'Not configured'}\n\u2705 HuggingFace: {'Ready' if HUGGINGFACE_TOKEN else 'Using public access'}\n\u2705 WebScraper API: Ready\n\u2705 Directories: Created\n\nYou can now proceed to the next notebook:\n  \u2192 01_data_acquisition.ipynb\n''')\nprint('=' * 60)\n"
   }
  ],
  "metadata": {