{
  "id": 178476,
  "title": "non-competitive",
  "url": "/competitions/gan-getting-started/discussion/178476",
  "author_name": "",
  "post_date": "2020-08-30T05:12:00.371548300Z",
  "votes": 3,
  "comment_count": 6,
  "views": 0,
  "content": "<p>I was reading the Overview and got to the FAQs and saw this…</p>\n<blockquote>\n  <p>Getting Started competitions are a non-competitive way to get familiar with Kaggle’s platform</p>\n</blockquote>\n<p>as if … still laughing 😂</p>\n<p>I'm sure going to try to beat Roman 😜</p>\n<p>Have fun. This is an interesting topic.</p>",
  "messages": [
    {
      "id": "991045",
      "postDate": "08/30/2020 05:12:00",
      "content": "<p>I was reading the Overview and got to the FAQs and saw this…</p>\n<blockquote>\n  <p>Getting Started competitions are a non-competitive way to get familiar with Kaggle’s platform</p>\n</blockquote>\n<p>as if … still laughing 😂</p>\n<p>I'm sure going to try to beat Roman 😜</p>\n<p>Have fun. This is an interesting topic.</p>",
      "rawMarkdown": "I was reading the Overview and got to the FAQs and saw this...\n\n> Getting Started competitions are a non-competitive way to get familiar with Kaggle’s platform\n\nas if ... still laughing 😂\n\nI'm sure going to try to beat Roman 😜\n\nHave fun. This is an interesting topic.",
      "votes": null
    },
    {
      "id": "1000667",
      "postDate": "09/06/2020 17:31:49",
      "content": "<p>still busy getting good melanoma scores</p>",
      "rawMarkdown": "still busy getting good melanoma scores",
      "votes": null
    },
    {
      "id": "1000921",
      "postDate": "09/06/2020 23:09:20",
      "content": "<p>I had a play for a while with Melanoma after the competition finished but my best score still remains at <strong>0.9456</strong>, which is <strong>15</strong>th on the Private leaderboard now. I'm currently happy with that and trying new stuff. </p>\n<p>The GAN comp has me <strong>2</strong>nd now but I'm running out of \"knobs\" to turn. </p>\n<p>As I'm originally a biologist, I'm a sucker for medical and scientific stuff, so I'm working through the Drug comp, but even though my local CV is getting better my LB pos is not. In fact the better it gets the worse my LB score is.</p>\n<p>I keep having a stab at the Lung comp but I think I may concentrate more on just the Drug one and not spread my time, effort and GPU/TPU quota to thinly.</p>\n<p>keeps me thinking 😃<br>\nYou know what I did all day today? Cut watermellon up into 1 inch chunks. 😓</p>",
      "rawMarkdown": "I had a play for a while with Melanoma after the competition finished but my best score still remains at **0.9456**, which is **15**th on the Private leaderboard now. I'm currently happy with that and trying new stuff. \n\nThe GAN comp has me **2**nd now but I'm running out of \"knobs\" to turn. \n\nAs I'm originally a biologist, I'm a sucker for medical and scientific stuff, so I'm working through the Drug comp, but even though my local CV is getting better my LB pos is not. In fact the better it gets the worse my LB score is.\n\nI keep having a stab at the Lung comp but I think I may concentrate more on just the Drug one and not spread my time, effort and GPU/TPU quota to thinly.\n\nkeeps me thinking 😃\nYou know what I did all day today? Cut watermellon up into 1 inch chunks. 😓",
      "votes": null
    },
    {
      "id": "1000979",
      "postDate": "09/07/2020 00:54:20",
      "content": "<p>Did not get 0.9456 until now but it's good training, to try out some techniques to beat you in the next challenge :)<br>\nNo - for real - I really can imagine an analyzing app to make sense (especially for ppl that don't have access to a laboratory to analyze it) The biggest problem is probably that one has to make a clinical study before releasing.<br>\nWill tell you when I got 0.946  - probably next week :)</p>",
      "rawMarkdown": "Did not get 0.9456 until now but it's good training, to try out some techniques to beat you in the next challenge :)\nNo - for real - I really can imagine an analyzing app to make sense (especially for ppl that don't have access to a laboratory to analyze it) The biggest problem is probably that one has to make a clinical study before releasing.\nWill tell you when I got 0.946  - probably next week :)",
      "votes": null
    },
    {
      "id": "1003193",
      "postDate": "09/08/2020 18:20:21",
      "content": "<p>I have found that: <a href=\"https://www.youtube.com/watch?v=GZVsONq3qtg\" target=\"_blank\">https://www.youtube.com/watch?v=GZVsONq3qtg</a></p>",
      "rawMarkdown": "I have found that: https://www.youtube.com/watch?v=GZVsONq3qtg",
      "votes": null
    },
    {
      "id": "1005172",
      "postDate": "09/10/2020 09:55:35",
      "content": "<p>this gan uses double discriminator - source code should be available</p>",
      "rawMarkdown": "this gan uses double discriminator - source code should be available",
      "votes": null
    },
    {
      "id": "1005572",
      "postDate": "09/10/2020 15:25:31",
      "content": "<p>0.9471 private with an ensemble of 10 models - think I can do even better next week</p>",
      "rawMarkdown": "0.9471 private with an ensemble of 10 models - think I can do even better next week",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1000667,
      "author_name": "romanweilguny",
      "author_url": "",
      "post_date": "09/06/2020 17:31:49",
      "content": "<p>still busy getting good melanoma scores</p>",
      "votes": null,
      "replies": [
        {
          "id": 1000921,
          "author_name": "mutantspore",
          "author_url": "",
          "post_date": "09/06/2020 23:09:20",
          "content": "<p>I had a play for a while with Melanoma after the competition finished but my best score still remains at <strong>0.9456</strong>, which is <strong>15</strong>th on the Private leaderboard now. I'm currently happy with that and trying new stuff. </p>\n<p>The GAN comp has me <strong>2</strong>nd now but I'm running out of \"knobs\" to turn. </p>\n<p>As I'm originally a biologist, I'm a sucker for medical and scientific stuff, so I'm working through the Drug comp, but even though my local CV is getting better my LB pos is not. In fact the better it gets the worse my LB score is.</p>\n<p>I keep having a stab at the Lung comp but I think I may concentrate more on just the Drug one and not spread my time, effort and GPU/TPU quota to thinly.</p>\n<p>keeps me thinking 😃<br>\nYou know what I did all day today? Cut watermellon up into 1 inch chunks. 😓</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1000979,
          "author_name": "romanweilguny",
          "author_url": "",
          "post_date": "09/07/2020 00:54:20",
          "content": "<p>Did not get 0.9456 until now but it's good training, to try out some techniques to beat you in the next challenge :)<br>\nNo - for real - I really can imagine an analyzing app to make sense (especially for ppl that don't have access to a laboratory to analyze it) The biggest problem is probably that one has to make a clinical study before releasing.<br>\nWill tell you when I got 0.946  - probably next week :)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1003193,
          "author_name": "romanweilguny",
          "author_url": "",
          "post_date": "09/08/2020 18:20:21",
          "content": "<p>I have found that: <a href=\"https://www.youtube.com/watch?v=GZVsONq3qtg\" target=\"_blank\">https://www.youtube.com/watch?v=GZVsONq3qtg</a></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1005172,
          "author_name": "romanweilguny",
          "author_url": "",
          "post_date": "09/10/2020 09:55:35",
          "content": "<p>this gan uses double discriminator - source code should be available</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1005572,
          "author_name": "romanweilguny",
          "author_url": "",
          "post_date": "09/10/2020 15:25:31",
          "content": "<p>0.9471 private with an ensemble of 10 models - think I can do even better next week</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "991045": "I was reading the Overview and got to the FAQs and saw this...\n\n> Getting Started competitions are a non-competitive way to get familiar with Kaggle’s platform\n\nas if ... still laughing 😂\n\nI'm sure going to try to beat Roman 😜\n\nHave fun. This is an interesting topic.",
    "1000667": "still busy getting good melanoma scores",
    "1000921": "I had a play for a while with Melanoma after the competition finished but my best score still remains at **0.9456**, which is **15**th on the Private leaderboard now. I'm currently happy with that and trying new stuff. \n\nThe GAN comp has me **2**nd now but I'm running out of \"knobs\" to turn. \n\nAs I'm originally a biologist, I'm a sucker for medical and scientific stuff, so I'm working through the Drug comp, but even though my local CV is getting better my LB pos is not. In fact the better it gets the worse my LB score is.\n\nI keep having a stab at the Lung comp but I think I may concentrate more on just the Drug one and not spread my time, effort and GPU/TPU quota to thinly.\n\nkeeps me thinking 😃\nYou know what I did all day today? Cut watermellon up into 1 inch chunks. 😓",
    "1000979": "Did not get 0.9456 until now but it's good training, to try out some techniques to beat you in the next challenge :)\nNo - for real - I really can imagine an analyzing app to make sense (especially for ppl that don't have access to a laboratory to analyze it) The biggest problem is probably that one has to make a clinical study before releasing.\nWill tell you when I got 0.946  - probably next week :)",
    "1003193": "I have found that: https://www.youtube.com/watch?v=GZVsONq3qtg",
    "1005172": "this gan uses double discriminator - source code should be available",
    "1005572": "0.9471 private with an ensemble of 10 models - think I can do even better next week"
  },
  "source": "meta"
}