{
  "id": 205144,
  "title": "Long time ⏳ model training",
  "url": "/competitions/ranzcr-clip-catheter-line-classification/discussion/205144",
  "author_name": "",
  "post_date": "2020-12-18T17:29:04.555377Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>As we all know the model training takes 20-30 Min for just single epoch (for <strong>GPU</strong>).<br>\nAlso, after 40 Min (run notebook and commit) the kaggle cross-check the user availability <strong>( Are You Still here ?)</strong>. </p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4684168%2F758cc3a827147400b4e5b5c99ee8d1a4%2F20201220_204042.jpg?generation=1608477066050288&amp;alt=media\" alt=\"\"></p>\n<p><strong>Can anybody suggest me a hack or a trick to believe kaggle kernel for user availability for long time (i.e. 5-6 hours)</strong></p>",
  "messages": [
    {
      "id": "1118059",
      "postDate": "12/18/2020 17:29:04",
      "content": "<p>As we all know the model training takes 20-30 Min for just single epoch (for <strong>GPU</strong>).<br>\nAlso, after 40 Min (run notebook and commit) the kaggle cross-check the user availability <strong>( Are You Still here ?)</strong>. </p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4684168%2F758cc3a827147400b4e5b5c99ee8d1a4%2F20201220_204042.jpg?generation=1608477066050288&amp;alt=media\" alt=\"\"></p>\n<p><strong>Can anybody suggest me a hack or a trick to believe kaggle kernel for user availability for long time (i.e. 5-6 hours)</strong></p>",
      "rawMarkdown": "As we all know the model training takes 20-30 Min for just single epoch (for **GPU**).\nAlso, after 40 Min (run notebook and commit) the kaggle cross-check the user availability **( Are You Still here ?)**. \n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4684168%2F758cc3a827147400b4e5b5c99ee8d1a4%2F20201220_204042.jpg?generation=1608477066050288&alt=media)\n\n\n**Can anybody suggest me a hack or a trick to believe kaggle kernel for user availability for long time (i.e. 5-6 hours)**",
      "votes": null
    },
    {
      "id": "1118119",
      "postDate": "12/18/2020 18:44:10",
      "content": "<p>So if your notebook is running, the \"are you still there\" shouldn't pop up, however, for me, especially when using TPUs, there are times even when the code is running, it still automatically turns the session off for me. In that case, just try reloading the page, it usually fixes it.</p>",
      "rawMarkdown": "So if your notebook is running, the \"are you still there\" shouldn't pop up, however, for me, especially when using TPUs, there are times even when the code is running, it still automatically turns the session off for me. In that case, just try reloading the page, it usually fixes it.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1118119,
      "author_name": "andy1010",
      "author_url": "",
      "post_date": "12/18/2020 18:44:10",
      "content": "<p>So if your notebook is running, the \"are you still there\" shouldn't pop up, however, for me, especially when using TPUs, there are times even when the code is running, it still automatically turns the session off for me. In that case, just try reloading the page, it usually fixes it.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1118059": "As we all know the model training takes 20-30 Min for just single epoch (for **GPU**).\nAlso, after 40 Min (run notebook and commit) the kaggle cross-check the user availability **( Are You Still here ?)**. \n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4684168%2F758cc3a827147400b4e5b5c99ee8d1a4%2F20201220_204042.jpg?generation=1608477066050288&alt=media)\n\n\n**Can anybody suggest me a hack or a trick to believe kaggle kernel for user availability for long time (i.e. 5-6 hours)**",
    "1118119": "So if your notebook is running, the \"are you still there\" shouldn't pop up, however, for me, especially when using TPUs, there are times even when the code is running, it still automatically turns the session off for me. In that case, just try reloading the page, it usually fixes it."
  },
  "source": "meta"
}