{
  "id": 79255,
  "title": "CNN Kernel dies",
  "url": "/competitions/quora-insincere-questions-classification/discussion/79255",
  "author_name": "BarrelChester",
  "post_date": "2019-02-01T18:56:33.584000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I'm trying to run a deep CNN model but the kernel dies as soon as I start training. When I run a shallow CNN in the notebook it always succeeds, but when I commit it I get a time out message in the kernel. Is there some way to \"ballpark\" the size of a network that a Kaggle kernel can manage?</p>\n\n<p>I've reduced the parameters to ~450,000 and it still dies instantly.</p>\n\n<p>Thanks!</p>",
  "messages": [
    {
      "id": 464905,
      "postDate": "2019-02-01T18:56:33.583Z",
      "content": "<p>I'm trying to run a deep CNN model but the kernel dies as soon as I start training. When I run a shallow CNN in the notebook it always succeeds, but when I commit it I get a time out message in the kernel. Is there some way to \"ballpark\" the size of a network that a Kaggle kernel can manage?</p>\n\n<p>I've reduced the parameters to ~450,000 and it still dies instantly.</p>\n\n<p>Thanks!</p>",
      "rawMarkdown": "I'm trying to run a deep CNN model but the kernel dies as soon as I start training. When I run a shallow CNN in the notebook it always succeeds, but when I commit it I get a time out message in the kernel. Is there some way to \"ballpark\" the size of a network that a Kaggle kernel can manage?\n\nI've reduced the parameters to ~450,000 and it still dies instantly.\n\nThanks!"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "464905": "I'm trying to run a deep CNN model but the kernel dies as soon as I start training. When I run a shallow CNN in the notebook it always succeeds, but when I commit it I get a time out message in the kernel. Is there some way to \"ballpark\" the size of a network that a Kaggle kernel can manage?\n\nI've reduced the parameters to ~450,000 and it still dies instantly.\n\nThanks!"
  }
}