{
  "id": 106326,
  "title": "Looking for teammates ",
  "url": "/competitions/aptos2019-blindness-detection/discussion/106326",
  "author_name": "",
  "post_date": "2019-08-29T05:46:45.990932600Z",
  "votes": 1,
  "comment_count": 11,
  "views": 0,
  "content": "<p>Hello all,</p>\n\n<p>I am looking for teammates, preferably with an LB score around 0.8. I know I am a little on the low side (0.793), but this is just a basic ResNet50 5-fold CV classification, which may be something you can add to an ensemble to improve results. I have contributed significantly to this competition (uploaded the previous dataset, the second-most upvoted kernel, three of the current hottest discussion posts are mine), in case you were worried about my dedication to the competition.</p>\n\n<p>Please let me know if you are interested!</p>",
  "messages": [
    {
      "id": "611123",
      "postDate": "08/29/2019 05:46:45",
      "content": "<p>Hello all,</p>\n\n<p>I am looking for teammates, preferably with an LB score around 0.8. I know I am a little on the low side (0.793), but this is just a basic ResNet50 5-fold CV classification, which may be something you can add to an ensemble to improve results. I have contributed significantly to this competition (uploaded the previous dataset, the second-most upvoted kernel, three of the current hottest discussion posts are mine), in case you were worried about my dedication to the competition.</p>\n\n<p>Please let me know if you are interested!</p>",
      "rawMarkdown": "Hello all,\n\nI am looking for teammates, preferably with an LB score around 0.8. I know I am a little on the low side (0.793), but this is just a basic ResNet50 5-fold CV classification, which may be something you can add to an ensemble to improve results. I have contributed significantly to this competition (uploaded the previous dataset, the second-most upvoted kernel, three of the current hottest discussion posts are mine), in case you were worried about my dedication to the competition.\n\nPlease let me know if you are interested!",
      "votes": null
    },
    {
      "id": "611167",
      "postDate": "08/29/2019 06:28:58",
      "content": "<p>hi, is your score:\n- from single model weights? (or ensemble of the 5 weights)\n- with or without TTA?\n- what is the approx. runtime? (on commit)</p>\n\n<p>(my score is 2x weights, TTA, commit time about 30 minutes)</p>",
      "rawMarkdown": "hi, is your score:\n- from single model weights? (or ensemble of the 5 weights)\n- with or without TTA?\n- what is the approx. runtime? (on commit)\n\n(my score is 2x weights, TTA, commit time about 30 minutes)",
      "votes": null
    },
    {
      "id": "612055",
      "postDate": "08/29/2019 15:25:49",
      "content": "<p>My score is 5 fold CV of ResNet50 without TTA. Commit time is like 10 min but submission time is like 2 hours</p>",
      "rawMarkdown": "My score is 5 fold CV of ResNet50 without TTA. Commit time is like 10 min but submission time is like 2 hours",
      "votes": null
    },
    {
      "id": "612195",
      "postDate": "08/29/2019 16:43:51",
      "content": "<p>ok, will think about it - but to be honest would also prefer someone &gt; 0.8</p>",
      "rawMarkdown": "ok, will think about it - but to be honest would also prefer someone &gt; 0.8",
      "votes": null
    },
    {
      "id": "612431",
      "postDate": "08/29/2019 19:40:18",
      "content": "<p>Are you still interested?</p>",
      "rawMarkdown": "Are you still interested?",
      "votes": null
    },
    {
      "id": "612455",
      "postDate": "08/29/2019 19:54:14",
      "content": "<p>The deadline is in a few hours...</p>",
      "rawMarkdown": "The deadline is in a few hours...",
      "votes": null
    },
    {
      "id": "612661",
      "postDate": "08/29/2019 23:37:41",
      "content": "<p>Hi,\nMy 1-fold model score is 0.794 (pretrained on 2015). And 0.788 trained on 2019. Classification only.\nAnd, 0.816 which is an ensemble of 3 models.\nI use EfficientNetB3 in Keras, simple TTA, one model runs approx 10 minutes (commit).</p>",
      "rawMarkdown": "Hi,\nMy 1-fold model score is 0.794 (pretrained on 2015). And 0.788 trained on 2019. Classification only.\nAnd, 0.816 which is an ensemble of 3 models.\nI use EfficientNetB3 in Keras, simple TTA, one model runs approx 10 minutes (commit).",
      "votes": null
    },
    {
      "id": "612665",
      "postDate": "08/29/2019 23:44:26",
      "content": "<p>Are you interested in teaming up with me? </p>",
      "rawMarkdown": "Are you interested in teaming up with me?",
      "votes": null
    },
    {
      "id": "612670",
      "postDate": "08/29/2019 23:46:38",
      "content": "<p>Yep! We have few minutes!</p>",
      "rawMarkdown": "Yep! We have few minutes!",
      "votes": null
    },
    {
      "id": "612672",
      "postDate": "08/29/2019 23:48:09",
      "content": "<p>I use fastai right now, but I think ensembling should still work...</p>",
      "rawMarkdown": "I use fastai right now, but I think ensembling should still work...",
      "votes": null
    },
    {
      "id": "612673",
      "postDate": "08/29/2019 23:51:11",
      "content": "<p>I am on kagglenoobs.slack.com and I think that is the best way to communicate.</p>",
      "rawMarkdown": "I am on kagglenoobs.slack.com and I think that is the best way to communicate.",
      "votes": null
    },
    {
      "id": "612677",
      "postDate": "08/29/2019 23:54:29",
      "content": "<p><a href=\"/agscin\">@agscin</a> you can join here: <a href=\"https://kagglenoobs.herokuapp.com/\">https://kagglenoobs.herokuapp.com/</a></p>",
      "rawMarkdown": "agscin you can join here: https://kagglenoobs.herokuapp.com/",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 611167,
      "author_name": "steelrose",
      "author_url": "",
      "post_date": "08/29/2019 06:28:58",
      "content": "<p>hi, is your score:\n- from single model weights? (or ensemble of the 5 weights)\n- with or without TTA?\n- what is the approx. runtime? (on commit)</p>\n\n<p>(my score is 2x weights, TTA, commit time about 30 minutes)</p>",
      "votes": null,
      "replies": [
        {
          "id": 612055,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 15:25:49",
          "content": "<p>My score is 5 fold CV of ResNet50 without TTA. Commit time is like 10 min but submission time is like 2 hours</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612195,
          "author_name": "steelrose",
          "author_url": "",
          "post_date": "08/29/2019 16:43:51",
          "content": "<p>ok, will think about it - but to be honest would also prefer someone &gt; 0.8</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612431,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 19:40:18",
          "content": "<p>Are you still interested?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612455,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 19:54:14",
          "content": "<p>The deadline is in a few hours...</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 612661,
      "author_name": "agscin",
      "author_url": "",
      "post_date": "08/29/2019 23:37:41",
      "content": "<p>Hi,\nMy 1-fold model score is 0.794 (pretrained on 2015). And 0.788 trained on 2019. Classification only.\nAnd, 0.816 which is an ensemble of 3 models.\nI use EfficientNetB3 in Keras, simple TTA, one model runs approx 10 minutes (commit).</p>",
      "votes": null,
      "replies": [
        {
          "id": 612665,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 23:44:26",
          "content": "<p>Are you interested in teaming up with me? </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612670,
          "author_name": "agscin",
          "author_url": "",
          "post_date": "08/29/2019 23:46:38",
          "content": "<p>Yep! We have few minutes!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612672,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 23:48:09",
          "content": "<p>I use fastai right now, but I think ensembling should still work...</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612673,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 23:51:11",
          "content": "<p>I am on kagglenoobs.slack.com and I think that is the best way to communicate.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612677,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 23:54:29",
          "content": "<p><a href=\"/agscin\">@agscin</a> you can join here: <a href=\"https://kagglenoobs.herokuapp.com/\">https://kagglenoobs.herokuapp.com/</a></p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "611123": "Hello all,\n\nI am looking for teammates, preferably with an LB score around 0.8. I know I am a little on the low side (0.793), but this is just a basic ResNet50 5-fold CV classification, which may be something you can add to an ensemble to improve results. I have contributed significantly to this competition (uploaded the previous dataset, the second-most upvoted kernel, three of the current hottest discussion posts are mine), in case you were worried about my dedication to the competition.\n\nPlease let me know if you are interested!",
    "611167": "hi, is your score:\n- from single model weights? (or ensemble of the 5 weights)\n- with or without TTA?\n- what is the approx. runtime? (on commit)\n\n(my score is 2x weights, TTA, commit time about 30 minutes)",
    "612055": "My score is 5 fold CV of ResNet50 without TTA. Commit time is like 10 min but submission time is like 2 hours",
    "612195": "ok, will think about it - but to be honest would also prefer someone &gt; 0.8",
    "612431": "Are you still interested?",
    "612455": "The deadline is in a few hours...",
    "612661": "Hi,\nMy 1-fold model score is 0.794 (pretrained on 2015). And 0.788 trained on 2019. Classification only.\nAnd, 0.816 which is an ensemble of 3 models.\nI use EfficientNetB3 in Keras, simple TTA, one model runs approx 10 minutes (commit).",
    "612665": "Are you interested in teaming up with me?",
    "612670": "Yep! We have few minutes!",
    "612672": "I use fastai right now, but I think ensembling should still work...",
    "612673": "I am on kagglenoobs.slack.com and I think that is the best way to communicate.",
    "612677": "agscin you can join here: https://kagglenoobs.herokuapp.com/"
  },
  "source": "meta"
}