{
  "id": 237125,
  "title": "We need rich computation resource?",
  "url": "/competitions/indoor-location-navigation/discussion/237125",
  "author_name": "",
  "post_date": "2021-05-07T08:47:42.485959400Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>In discussions or notebooks, some of top LB teams seem to use RNN based models.<br>\nFor many participants in this competition (including me), it is not easy to build Deep Learning models because of insufficient computing resource.</p>\n<p>But I think rich computing resource is not important in this competition.<br>\nI've been using only Kaggle Notebook (CPU)  in this competition and reached public score 2.x recently.</p>\n<p>I'm looking forward to hear top teams' solutions after 18th, May. :)</p>",
  "messages": [
    {
      "id": "1296422",
      "postDate": "05/07/2021 08:47:42",
      "content": "<p>In discussions or notebooks, some of top LB teams seem to use RNN based models.<br>\nFor many participants in this competition (including me), it is not easy to build Deep Learning models because of insufficient computing resource.</p>\n<p>But I think rich computing resource is not important in this competition.<br>\nI've been using only Kaggle Notebook (CPU)  in this competition and reached public score 2.x recently.</p>\n<p>I'm looking forward to hear top teams' solutions after 18th, May. :)</p>",
      "rawMarkdown": "In discussions or notebooks, some of top LB teams seem to use RNN based models.\nFor many participants in this competition (including me), it is not easy to build Deep Learning models because of insufficient computing resource.\n\nBut I think rich computing resource is not important in this competition.\nI've been using only Kaggle Notebook (CPU)  in this competition and reached public score 2.x recently.\n\nI'm looking forward to hear top teams' solutions after 18th, May. :)",
      "votes": null
    },
    {
      "id": "1296787",
      "postDate": "05/07/2021 13:51:42",
      "content": "<p>I've just been using google colab (pro - $10/mo), which has similar resources as kaggle notebooks. There is a lot of data in this competition, but after pre-processing it's not <em>that</em> big - it's not like the image challenges which have many many GB of images :)</p>",
      "rawMarkdown": "I've just been using google colab (pro - $10/mo), which has similar resources as kaggle notebooks. There is a lot of data in this competition, but after pre-processing it's not _that_ big - it's not like the image challenges which have many many GB of images :)",
      "votes": null
    },
    {
      "id": "1296851",
      "postDate": "05/07/2021 14:29:52",
      "content": "<p>I agree. In my case preprocessed data (containing also sensor data) takes around 10 GBs using .npy format</p>",
      "rawMarkdown": "I agree. In my case preprocessed data (containing also sensor data) takes around 10 GBs using .npy format",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1296787,
      "author_name": "chris62",
      "author_url": "",
      "post_date": "05/07/2021 13:51:42",
      "content": "<p>I've just been using google colab (pro - $10/mo), which has similar resources as kaggle notebooks. There is a lot of data in this competition, but after pre-processing it's not <em>that</em> big - it's not like the image challenges which have many many GB of images :)</p>",
      "votes": null,
      "replies": [
        {
          "id": 1296851,
          "author_name": "olaf2000",
          "author_url": "",
          "post_date": "05/07/2021 14:29:52",
          "content": "<p>I agree. In my case preprocessed data (containing also sensor data) takes around 10 GBs using .npy format</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1296422": "In discussions or notebooks, some of top LB teams seem to use RNN based models.\nFor many participants in this competition (including me), it is not easy to build Deep Learning models because of insufficient computing resource.\n\nBut I think rich computing resource is not important in this competition.\nI've been using only Kaggle Notebook (CPU)  in this competition and reached public score 2.x recently.\n\nI'm looking forward to hear top teams' solutions after 18th, May. :)",
    "1296787": "I've just been using google colab (pro - $10/mo), which has similar resources as kaggle notebooks. There is a lot of data in this competition, but after pre-processing it's not _that_ big - it's not like the image challenges which have many many GB of images :)",
    "1296851": "I agree. In my case preprocessed data (containing also sensor data) takes around 10 GBs using .npy format"
  },
  "source": "meta"
}