{
  "id": 39690,
  "title": "Hacking GTA V for Carvana Kaggle Challenge – Hacker Noon",
  "url": "/competitions/carvana-image-masking-challenge/discussion/39690",
  "author_name": "",
  "post_date": "2017-09-19T09:38:41.895957700Z",
  "votes": 6,
  "comment_count": 6,
  "views": 0,
  "content": "<p><a href=\"https://hackernoon.com/hacking-gta-v-for-carvana-kaggle-challenge-6d0b7fb4c781\">https://hackernoon.com/hacking-gta-v-for-carvana-kaggle-challenge-6d0b7fb4c781</a></p>\n\n<p>Check this out!</p>\n\n<p>This is not my work. It is very good work by \"deepsystems.ai.\" I think it is very impressive!</p>",
  "messages": [
    {
      "id": "222583",
      "postDate": "09/19/2017 09:38:41",
      "content": "<p><a href=\"https://hackernoon.com/hacking-gta-v-for-carvana-kaggle-challenge-6d0b7fb4c781\">https://hackernoon.com/hacking-gta-v-for-carvana-kaggle-challenge-6d0b7fb4c781</a></p>\n\n<p>Check this out!</p>\n\n<p>This is not my work. It is very good work by \"deepsystems.ai.\" I think it is very impressive!</p>",
      "rawMarkdown": "https://hackernoon.com/hacking-gta-v-for-carvana-kaggle-challenge-6d0b7fb4c781\n\nCheck this out!\n\nThis is not my work. It is very good work by \"deepsystems.ai.\" I think it is very impressive!",
      "votes": null
    },
    {
      "id": "222593",
      "postDate": "09/19/2017 10:36:15",
      "content": "<p>I believe the data was posted too late (according to the competition rules), so we are not allowed to use the model.</p>\n\n<p>Looked great, though.</p>",
      "rawMarkdown": "I believe the data was posted too late (according to the competition rules), so we are not allowed to use the model.\n\nLooked great, though.",
      "votes": null
    },
    {
      "id": "222800",
      "postDate": "09/20/2017 06:01:14",
      "content": "<p>Yeah, I just saw his post leading me to it. It looks very interesting but we cannot use it.</p>",
      "rawMarkdown": "Yeah, I just saw his post leading me to it. It looks very interesting but we cannot use it.",
      "votes": null
    },
    {
      "id": "222950",
      "postDate": "09/20/2017 17:07:45",
      "content": "<p>Wow, they have even considered shadows! It would look so fake to just copy the cars into new scenes!</p>",
      "rawMarkdown": "Wow, they have even considered shadows! It would look so fake to just copy the cars into new scenes!",
      "votes": null
    },
    {
      "id": "222976",
      "postDate": "09/20/2017 19:00:19",
      "content": "<p>This is a really good dataset. I have played around with it by training in the resolution of 512*512 and 1024*1024 and found out that it improves a lot for my 512*512 model, but gets overfitting in my 1024*1024 model. I am curious why this happens. </p>",
      "rawMarkdown": "This is a really good dataset. I have played around with it by training in the resolution of 512*512 and 1024*1024 and found out that it improves a lot for my 512*512 model, but gets overfitting in my 1024*1024 model. I am curious why this happens.",
      "votes": null
    },
    {
      "id": "223460",
      "postDate": "09/22/2017 07:40:36",
      "content": "<p>so.it is too late?</p>",
      "rawMarkdown": "so.it is too late?",
      "votes": null
    },
    {
      "id": "223740",
      "postDate": "09/23/2017 09:07:51",
      "content": "<p>yes. But this data doesn't give any improvement anyway.</p>",
      "rawMarkdown": "yes. But this data doesn't give any improvement anyway.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 222593,
      "author_name": "antorsae",
      "author_url": "",
      "post_date": "09/19/2017 10:36:15",
      "content": "<p>I believe the data was posted too late (according to the competition rules), so we are not allowed to use the model.</p>\n\n<p>Looked great, though.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222800,
      "author_name": "sheriytm",
      "author_url": "",
      "post_date": "09/20/2017 06:01:14",
      "content": "<p>Yeah, I just saw his post leading me to it. It looks very interesting but we cannot use it.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222950,
      "author_name": "wappyotoole",
      "author_url": "",
      "post_date": "09/20/2017 17:07:45",
      "content": "<p>Wow, they have even considered shadows! It would look so fake to just copy the cars into new scenes!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 222976,
      "author_name": "junhongxu",
      "author_url": "",
      "post_date": "09/20/2017 19:00:19",
      "content": "<p>This is a really good dataset. I have played around with it by training in the resolution of 512*512 and 1024*1024 and found out that it improves a lot for my 512*512 model, but gets overfitting in my 1024*1024 model. I am curious why this happens. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 223460,
      "author_name": "zhenghang",
      "author_url": "",
      "post_date": "09/22/2017 07:40:36",
      "content": "<p>so.it is too late?</p>",
      "votes": null,
      "replies": [
        {
          "id": 223740,
          "author_name": "asanakoev",
          "author_url": "",
          "post_date": "09/23/2017 09:07:51",
          "content": "<p>yes. But this data doesn't give any improvement anyway.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "222583": "https://hackernoon.com/hacking-gta-v-for-carvana-kaggle-challenge-6d0b7fb4c781\n\nCheck this out!\n\nThis is not my work. It is very good work by \"deepsystems.ai.\" I think it is very impressive!",
    "222593": "I believe the data was posted too late (according to the competition rules), so we are not allowed to use the model.\n\nLooked great, though.",
    "222800": "Yeah, I just saw his post leading me to it. It looks very interesting but we cannot use it.",
    "222950": "Wow, they have even considered shadows! It would look so fake to just copy the cars into new scenes!",
    "222976": "This is a really good dataset. I have played around with it by training in the resolution of 512*512 and 1024*1024 and found out that it improves a lot for my 512*512 model, but gets overfitting in my 1024*1024 model. I am curious why this happens.",
    "223460": "so.it is too late?",
    "223740": "yes. But this data doesn't give any improvement anyway."
  },
  "source": "meta"
}