{
  "id": 113481,
  "title": "Question for a Neural Network",
  "url": "/competitions/ashrae-energy-prediction/discussion/113481",
  "author_name": "",
  "post_date": "2019-10-19T22:06:03.262990100Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I tried to predict the test data with a simple Neural Network. The kernel is here: <a href=\"https://www.kaggle.com/drcapa/ashrae-datagenerator-neuralnetwork\">https://www.kaggle.com/drcapa/ashrae-datagenerator-neuralnetwork</a>  (Version3)</p>\n\n<p>I work with a data generator. Unfortunately I can only train on a little subset of the train data becauce of memory and calculation time. An other problem is to predict the the big test data. My kernel will got out of memory and need too much calculation time. </p>\n\n<p>I tried to reduce the memory and switch on GPU without success.</p>\n\n<p>Are there any techniques i can try out or helpful advices?</p>",
  "messages": [
    {
      "id": "653073",
      "postDate": "10/19/2019 22:06:03",
      "content": "<p>I tried to predict the test data with a simple Neural Network. The kernel is here: <a href=\"https://www.kaggle.com/drcapa/ashrae-datagenerator-neuralnetwork\">https://www.kaggle.com/drcapa/ashrae-datagenerator-neuralnetwork</a>  (Version3)</p>\n\n<p>I work with a data generator. Unfortunately I can only train on a little subset of the train data becauce of memory and calculation time. An other problem is to predict the the big test data. My kernel will got out of memory and need too much calculation time. </p>\n\n<p>I tried to reduce the memory and switch on GPU without success.</p>\n\n<p>Are there any techniques i can try out or helpful advices?</p>",
      "rawMarkdown": "I tried to predict the test data with a simple Neural Network. The kernel is here: https://www.kaggle.com/drcapa/ashrae-datagenerator-neuralnetwork  (Version3)\n\nI work with a data generator. Unfortunately I can only train on a little subset of the train data becauce of memory and calculation time. An other problem is to predict the the big test data. My kernel will got out of memory and need too much calculation time. \n\nI tried to reduce the memory and switch on GPU without success.\n\nAre there any techniques i can try out or helpful advices?",
      "votes": null
    },
    {
      "id": "653075",
      "postDate": "10/19/2019 22:09:58",
      "content": "<p>Use a virtual machine with high ram and work on a IDE instead of kaggle kernel.</p>",
      "rawMarkdown": "Use a virtual machine with high ram and work on a IDE instead of kaggle kernel.",
      "votes": null
    },
    {
      "id": "653546",
      "postDate": "10/20/2019 16:07:47",
      "content": "<p>Many thanks for your response.</p>",
      "rawMarkdown": "Many thanks for your response.",
      "votes": null
    },
    {
      "id": "653550",
      "postDate": "10/20/2019 16:09:52",
      "content": "<p>Finally i improved my workflow with great results. Since version 4 the kernel can train on the whole data and predict all test data in acceptable time. </p>",
      "rawMarkdown": "Finally i improved my workflow with great results. Since version 4 the kernel can train on the whole data and predict all test data in acceptable time.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 653075,
      "author_name": "cemsaftekin92",
      "author_url": "",
      "post_date": "10/19/2019 22:09:58",
      "content": "<p>Use a virtual machine with high ram and work on a IDE instead of kaggle kernel.</p>",
      "votes": null,
      "replies": [
        {
          "id": 653546,
          "author_name": "drcapa",
          "author_url": "",
          "post_date": "10/20/2019 16:07:47",
          "content": "<p>Many thanks for your response.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 653550,
      "author_name": "drcapa",
      "author_url": "",
      "post_date": "10/20/2019 16:09:52",
      "content": "<p>Finally i improved my workflow with great results. Since version 4 the kernel can train on the whole data and predict all test data in acceptable time. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "653073": "I tried to predict the test data with a simple Neural Network. The kernel is here: https://www.kaggle.com/drcapa/ashrae-datagenerator-neuralnetwork  (Version3)\n\nI work with a data generator. Unfortunately I can only train on a little subset of the train data becauce of memory and calculation time. An other problem is to predict the the big test data. My kernel will got out of memory and need too much calculation time. \n\nI tried to reduce the memory and switch on GPU without success.\n\nAre there any techniques i can try out or helpful advices?",
    "653075": "Use a virtual machine with high ram and work on a IDE instead of kaggle kernel.",
    "653546": "Many thanks for your response.",
    "653550": "Finally i improved my workflow with great results. Since version 4 the kernel can train on the whole data and predict all test data in acceptable time."
  },
  "source": "meta"
}