{
  "id": 107862,
  "title": "Competition about to end",
  "url": "/competitions/aptos2019-blindness-detection/discussion/107862",
  "author_name": "",
  "post_date": "2019-09-07T13:20:34.853462800Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I am so far quite satisfied with my stand\nI was targeting for bronze hope fully should end with it but still dont know about the shake up as many are having apprehension about. \nFew takeaways\n1) Try new things incrementally from starting days. and observe the score, if it keep rising add more part to it.Do keep a track of good performances and save your models with right tag.</p>\n\n<p>2) dont jump from one model to another model until you have tried all possible bits with it</p>\n\n<p>3) Go through the discussions regularly  to watch for   any new updates in field of Deep learning eg. Efficientnet was new model i came across and i think top rankers  may be using it </p>\n\n<p>4 ) go to know about the new type of augmentations  combinations ,thanks to @drhb </p>",
  "messages": [
    {
      "id": "620412",
      "postDate": "09/07/2019 13:20:34",
      "content": "<p>I am so far quite satisfied with my stand\nI was targeting for bronze hope fully should end with it but still dont know about the shake up as many are having apprehension about. \nFew takeaways\n1) Try new things incrementally from starting days. and observe the score, if it keep rising add more part to it.Do keep a track of good performances and save your models with right tag.</p>\n\n<p>2) dont jump from one model to another model until you have tried all possible bits with it</p>\n\n<p>3) Go through the discussions regularly  to watch for   any new updates in field of Deep learning eg. Efficientnet was new model i came across and i think top rankers  may be using it </p>\n\n<p>4 ) go to know about the new type of augmentations  combinations ,thanks to @drhb </p>",
      "rawMarkdown": "I am so far quite satisfied with my stand\nI was targeting for bronze hope fully should end with it but still dont know about the shake up as many are having apprehension about. \nFew takeaways\n1) Try new things incrementally from starting days. and observe the score, if it keep rising add more part to it.Do keep a track of good performances and save your models with right tag.\n\n2) dont jump from one model to another model until you have tried all possible bits with it\n\n3) Go through the discussions regularly  to watch for   any new updates in field of Deep learning eg. Efficientnet was new model i came across and i think top rankers  may be using it \n\n4 ) go to know about the new type of augmentations  combinations ,thanks to @drhb",
      "votes": null
    },
    {
      "id": "620415",
      "postDate": "09/07/2019 13:29:09",
      "content": "<p>4 best advice mate,specially the 2nd one,,i see kagglers switch model to model for maximizing score,it is a wrong idea!!! thank you man <a href=\"/jaideepvalani\">@jaideepvalani</a> </p>",
      "rawMarkdown": "4 best advice mate,specially the 2nd one,,i see kagglers switch model to model for maximizing score,it is a wrong idea!!! thank you man @jaideepvalani",
      "votes": null
    },
    {
      "id": "620429",
      "postDate": "09/07/2019 13:54:50",
      "content": "<p>Great stuff =) Good luck for the private standings hope you can get a medal =) </p>",
      "rawMarkdown": "Great stuff =) Good luck for the private standings hope you can get a medal =)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 620415,
      "author_name": "mobassir",
      "author_url": "",
      "post_date": "09/07/2019 13:29:09",
      "content": "<p>4 best advice mate,specially the 2nd one,,i see kagglers switch model to model for maximizing score,it is a wrong idea!!! thank you man <a href=\"/jaideepvalani\">@jaideepvalani</a> </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 620429,
      "author_name": "drhabib",
      "author_url": "",
      "post_date": "09/07/2019 13:54:50",
      "content": "<p>Great stuff =) Good luck for the private standings hope you can get a medal =) </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "620412": "I am so far quite satisfied with my stand\nI was targeting for bronze hope fully should end with it but still dont know about the shake up as many are having apprehension about. \nFew takeaways\n1) Try new things incrementally from starting days. and observe the score, if it keep rising add more part to it.Do keep a track of good performances and save your models with right tag.\n\n2) dont jump from one model to another model until you have tried all possible bits with it\n\n3) Go through the discussions regularly  to watch for   any new updates in field of Deep learning eg. Efficientnet was new model i came across and i think top rankers  may be using it \n\n4 ) go to know about the new type of augmentations  combinations ,thanks to @drhb",
    "620415": "4 best advice mate,specially the 2nd one,,i see kagglers switch model to model for maximizing score,it is a wrong idea!!! thank you man @jaideepvalani",
    "620429": "Great stuff =) Good luck for the private standings hope you can get a medal =)"
  },
  "source": "meta"
}