{
  "id": 103075,
  "title": "The kaggle kernel close after running over 1 hour, how could I train the model?",
  "url": "/competitions/aptos2019-blindness-detection/discussion/103075",
  "author_name": "lawrence",
  "post_date": "2019-08-07T00:20:40.715000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>As title, even with GPU on, for the large scale dataset to train the whole data cost over one hour to run the machine. For me, in each training epoch it cost 20minutes to run though whole training set one time. After three epochs it exceed one hour, and the kaggle kernel stops automatically. How could I train with the kernel for long running time?</p>",
  "messages": [
    {
      "id": 593662,
      "postDate": "2019-08-07T00:20:40.717Z",
      "content": "<p>As title, even with GPU on, for the large scale dataset to train the whole data cost over one hour to run the machine. For me, in each training epoch it cost 20minutes to run though whole training set one time. After three epochs it exceed one hour, and the kaggle kernel stops automatically. How could I train with the kernel for long running time?</p>",
      "rawMarkdown": "As title, even with GPU on, for the large scale dataset to train the whole data cost over one hour to run the machine. For me, in each training epoch it cost 20minutes to run though whole training set one time. After three epochs it exceed one hour, and the kaggle kernel stops automatically. How could I train with the kernel for long running time?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "593662": "As title, even with GPU on, for the large scale dataset to train the whole data cost over one hour to run the machine. For me, in each training epoch it cost 20minutes to run though whole training set one time. After three epochs it exceed one hour, and the kaggle kernel stops automatically. How could I train with the kernel for long running time?"
  }
}