{
  "id": 619309,
  "title": "Video how to use Kaggle server in VS Code",
  "url": "/competitions/vesuvius-challenge-surface-detection/discussion/619309",
  "author_name": "Samidullo",
  "post_date": "2025-11-13T22:34:46.011000",
  "votes": 27,
  "comment_count": 3,
  "views": 0,
  "content": "<p><strong>If you want to work with different AI models and still want to code in VS Code, check the below link</strong></p>\n<p>It helps you connect Kaggle Jupyter server into your local VS Code. </p>\n<p><strong><img src=\"https://drive.google.com/file/d/161RIawvtU1-V49GW60Vr4n0TAXqjwLG9/view?usp=drivesdk\" alt=\"Instruction video\"></strong></p>\n<hr>\n<hr>\n<p><strong><em>Detailed Guide: Connect VS Code Local to Kaggle Jupyter Server for Free GPU Access</em></strong></p>\n<p><strong>Step 1: Start Kaggle Jupyter Server with GPU</strong></p>\n<ol>\n<li>Go to Kaggle Notebooks (kaggle.com/notebooks) and create or open a notebook.</li>\n<li>Select Run &gt; Kaggle Jupyter Server to open the right-side panel.</li>\n</ol>\n<p>In the panel, enable the GPU option (if needed) and click Start to launch the server.</p>\n<p><strong>Step 2: Get the VS Code Compatible URL</strong></p>\n<ol>\n<li>In the Kaggle Jupyter Server panel, scroll to the Manually Connect section.</li>\n<li>Copy the VS Code Compatible URL (e.g., http://.kaggle.com:8888/?token=).</li>\n</ol>\n<p><strong>Step 3: Connect from Local VS Code</strong></p>\n<ol>\n<li>Open VS Code and load your .ipynb file (or create a new one).</li>\n<li>Click Select Kernel in the top-right corner. If you don’t see it, press Ctrl+Shift+P (or Cmd+Shift+P on Mac).</li>\n<li>Choose Select Notebook Kernel from the command list.</li>\n<li>Select Existing Jupyter Server, then paste the Kaggle URL into the “Enter the URL of the running Jupyter server” field.</li>\n<li>Rename the kernel (optional, e.g., “Kaggle GPU”).</li>\n<li>Select the newly named kernel from the list.</li>\n<li>Press Enter to connect. Create a new cell and run !nvidia-smi to check the GPU.</li>\n</ol>\n<p>Result\nI’ve successfully run heavy tasks like model training on Kaggle’s GPU without complex local setup. Coding stays in the familiar VS Code interface, while the computational power comes from Kaggle!</p>\n<p>Notes</p>\n<ul>\n<li>Manual Save: The .ipynb file in VS Code doesn’t sync with Kaggle, so save often.</li>\n<li>Server Timeout: If Kaggle shuts down the server due to inactivity, repeat the steps to reconnect.</li>\n</ul>\n<p>References:</p>\n<ul>\n<li>Kaggle Docs: <a href=\"https://lnkd.in/gpS5cTgB\" target=\"_blank\">https://lnkd.in/gpS5cTgB</a></li>\n<li>VS Code Docs: <a href=\"https://lnkd.in/gUcUXVnj\" target=\"_blank\">https://lnkd.in/gUcUXVnj</a></li>\n</ul>\n<hr>\n<h1><strong>Also check the article too:</strong> Google Colab &amp; VS Code Extension: <a href=\"https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0\" target=\"_blank\">https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0</a></h1>",
  "messages": [
    {
      "id": 3322967,
      "postDate": "2025-11-13T22:34:46.010Z",
      "content": "<p><strong>If you want to work with different AI models and still want to code in VS Code, check the below link</strong></p>\n<p>It helps you connect Kaggle Jupyter server into your local VS Code. </p>\n<p><strong><img src=\"https://drive.google.com/file/d/161RIawvtU1-V49GW60Vr4n0TAXqjwLG9/view?usp=drivesdk\" alt=\"Instruction video\"></strong></p>\n<hr>\n<hr>\n<p><strong><em>Detailed Guide: Connect VS Code Local to Kaggle Jupyter Server for Free GPU Access</em></strong></p>\n<p><strong>Step 1: Start Kaggle Jupyter Server with GPU</strong></p>\n<ol>\n<li>Go to Kaggle Notebooks (kaggle.com/notebooks) and create or open a notebook.</li>\n<li>Select Run &gt; Kaggle Jupyter Server to open the right-side panel.</li>\n</ol>\n<p>In the panel, enable the GPU option (if needed) and click Start to launch the server.</p>\n<p><strong>Step 2: Get the VS Code Compatible URL</strong></p>\n<ol>\n<li>In the Kaggle Jupyter Server panel, scroll to the Manually Connect section.</li>\n<li>Copy the VS Code Compatible URL (e.g., http://.kaggle.com:8888/?token=).</li>\n</ol>\n<p><strong>Step 3: Connect from Local VS Code</strong></p>\n<ol>\n<li>Open VS Code and load your .ipynb file (or create a new one).</li>\n<li>Click Select Kernel in the top-right corner. If you don’t see it, press Ctrl+Shift+P (or Cmd+Shift+P on Mac).</li>\n<li>Choose Select Notebook Kernel from the command list.</li>\n<li>Select Existing Jupyter Server, then paste the Kaggle URL into the “Enter the URL of the running Jupyter server” field.</li>\n<li>Rename the kernel (optional, e.g., “Kaggle GPU”).</li>\n<li>Select the newly named kernel from the list.</li>\n<li>Press Enter to connect. Create a new cell and run !nvidia-smi to check the GPU.</li>\n</ol>\n<p>Result\nI’ve successfully run heavy tasks like model training on Kaggle’s GPU without complex local setup. Coding stays in the familiar VS Code interface, while the computational power comes from Kaggle!</p>\n<p>Notes</p>\n<ul>\n<li>Manual Save: The .ipynb file in VS Code doesn’t sync with Kaggle, so save often.</li>\n<li>Server Timeout: If Kaggle shuts down the server due to inactivity, repeat the steps to reconnect.</li>\n</ul>\n<p>References:</p>\n<ul>\n<li>Kaggle Docs: <a href=\"https://lnkd.in/gpS5cTgB\" target=\"_blank\">https://lnkd.in/gpS5cTgB</a></li>\n<li>VS Code Docs: <a href=\"https://lnkd.in/gUcUXVnj\" target=\"_blank\">https://lnkd.in/gUcUXVnj</a></li>\n</ul>\n<hr>\n<h1><strong>Also check the article too:</strong> Google Colab &amp; VS Code Extension: <a href=\"https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0\" target=\"_blank\">https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0</a></h1>",
      "rawMarkdown": "**If you want to work with different AI models and still want to code in VS Code, check the below link**\n\nIt helps you connect Kaggle Jupyter server into your local VS Code. \n\n**![Instruction video](https://drive.google.com/file/d/161RIawvtU1-V49GW60Vr4n0TAXqjwLG9/view?usp=drivesdk)**\n\n-----------------------------------------------------------------------------------------------------------------------------------------------------\n\n-----------------------------------------------------------------------------------------------------------------------------------------------------\n\n***Detailed Guide: Connect VS Code Local to Kaggle Jupyter Server for Free GPU Access***\n\n**Step 1: Start Kaggle Jupyter Server with GPU**\n1. Go to Kaggle Notebooks (kaggle.com/notebooks) and create or open a notebook.\n2. Select Run > Kaggle Jupyter Server to open the right-side panel.\n\nIn the panel, enable the GPU option (if needed) and click Start to launch the server.\n\n**Step 2: Get the VS Code Compatible URL**\n1. In the Kaggle Jupyter Server panel, scroll to the Manually Connect section.\n2. Copy the VS Code Compatible URL (e.g., http://<server-id>.kaggle.com:8888/?token=<your-token>).\n\n**Step 3: Connect from Local VS Code**\n1. Open VS Code and load your .ipynb file (or create a new one).\n2. Click Select Kernel in the top-right corner. If you don’t see it, press Ctrl+Shift+P (or Cmd+Shift+P on Mac).\n3. Choose Select Notebook Kernel from the command list.\n4. Select Existing Jupyter Server, then paste the Kaggle URL into the “Enter the URL of the running Jupyter server” field.\n5. Rename the kernel (optional, e.g., “Kaggle GPU”).\n6. Select the newly named kernel from the list.\n7. Press Enter to connect. Create a new cell and run !nvidia-smi to check the GPU.\n\nResult\nI’ve successfully run heavy tasks like model training on Kaggle’s GPU without complex local setup. Coding stays in the familiar VS Code interface, while the computational power comes from Kaggle!\n\nNotes\n- Manual Save: The .ipynb file in VS Code doesn’t sync with Kaggle, so save often.\n- Server Timeout: If Kaggle shuts down the server due to inactivity, repeat the steps to reconnect.\n\nReferences:\n- Kaggle Docs: https://lnkd.in/gpS5cTgB\n- VS Code Docs: https://lnkd.in/gUcUXVnj\n\n\n\n-----\n\n\n# **Also check the article too:** Google Colab & VS Code Extension: [https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0](https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0)",
      "votes": 27
    },
    {
      "id": 3393015,
      "postDate": "2026-01-18T05:02:13.123Z",
      "content": "<p>Always wondered if its possible to setup a link between kaggle and vscode for computational resources. Neat!</p>",
      "rawMarkdown": "Always wondered if its possible to setup a link between kaggle and vscode for computational resources. Neat!"
    },
    {
      "id": 3382902,
      "postDate": "2025-12-28T19:22:27.873Z",
      "content": "<p>thanks for this</p>",
      "rawMarkdown": "thanks for this"
    },
    {
      "id": 3382918,
      "postDate": "2025-12-28T20:15:33.300Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 3393015,
      "author_name": "Daisy Arce",
      "author_url": "",
      "post_date": "2026-01-18T05:02:13.123000",
      "content": "<p>Always wondered if its possible to setup a link between kaggle and vscode for computational resources. Neat!</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 3382902,
      "author_name": "Oblivix",
      "author_url": "",
      "post_date": "2025-12-28T19:22:27.873000",
      "content": "<p>thanks for this</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 3382918,
      "author_name": "",
      "author_url": "",
      "post_date": "2025-12-28T20:15:33.300000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3322967": "**If you want to work with different AI models and still want to code in VS Code, check the below link**\n\nIt helps you connect Kaggle Jupyter server into your local VS Code. \n\n**![Instruction video](https://drive.google.com/file/d/161RIawvtU1-V49GW60Vr4n0TAXqjwLG9/view?usp=drivesdk)**\n\n-----------------------------------------------------------------------------------------------------------------------------------------------------\n\n-----------------------------------------------------------------------------------------------------------------------------------------------------\n\n***Detailed Guide: Connect VS Code Local to Kaggle Jupyter Server for Free GPU Access***\n\n**Step 1: Start Kaggle Jupyter Server with GPU**\n1. Go to Kaggle Notebooks (kaggle.com/notebooks) and create or open a notebook.\n2. Select Run > Kaggle Jupyter Server to open the right-side panel.\n\nIn the panel, enable the GPU option (if needed) and click Start to launch the server.\n\n**Step 2: Get the VS Code Compatible URL**\n1. In the Kaggle Jupyter Server panel, scroll to the Manually Connect section.\n2. Copy the VS Code Compatible URL (e.g., http://<server-id>.kaggle.com:8888/?token=<your-token>).\n\n**Step 3: Connect from Local VS Code**\n1. Open VS Code and load your .ipynb file (or create a new one).\n2. Click Select Kernel in the top-right corner. If you don’t see it, press Ctrl+Shift+P (or Cmd+Shift+P on Mac).\n3. Choose Select Notebook Kernel from the command list.\n4. Select Existing Jupyter Server, then paste the Kaggle URL into the “Enter the URL of the running Jupyter server” field.\n5. Rename the kernel (optional, e.g., “Kaggle GPU”).\n6. Select the newly named kernel from the list.\n7. Press Enter to connect. Create a new cell and run !nvidia-smi to check the GPU.\n\nResult\nI’ve successfully run heavy tasks like model training on Kaggle’s GPU without complex local setup. Coding stays in the familiar VS Code interface, while the computational power comes from Kaggle!\n\nNotes\n- Manual Save: The .ipynb file in VS Code doesn’t sync with Kaggle, so save often.\n- Server Timeout: If Kaggle shuts down the server due to inactivity, repeat the steps to reconnect.\n\nReferences:\n- Kaggle Docs: https://lnkd.in/gpS5cTgB\n- VS Code Docs: https://lnkd.in/gUcUXVnj\n\n\n\n-----\n\n\n# **Also check the article too:** Google Colab & VS Code Extension: [https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0](https://medium.com/@lucamassaron/google-colab-vs-code-extension-7d84f55d20e0)",
    "3393015": "Always wondered if its possible to setup a link between kaggle and vscode for computational resources. Neat!",
    "3382902": "thanks for this",
    "3382918": ""
  }
}