{
  "id": 106479,
  "title": "looking for team members (preferably top 100)",
  "url": "/competitions/aptos2019-blindness-detection/discussion/106479",
  "author_name": "",
  "post_date": "2019-08-29T16:50:46.670689300Z",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>as the topic title says - looking for someone (preferably top 100) to form a team</p>\n\n<p>my current score is:\n- 2x Effnet-B4 (classification) + TTA; no Kappa optimization\n- commit runtime around 20-30 minutes</p>\n\n<p>I'm using PyTorch, have 2x RTX2080ti so mostly I only submit when I see good CV</p>\n\n<p>looking for:\n- either different models (Densenet, Resnext etc.) or different Effnet\n- not too many submits (so I can assume you're not trying to overfit to the LB :)\n- runtime (for commit) also at most 30 minutes\n- probably PyTorch also? not sure if it will be easy to combine predictions from PyTorch + Keras / TF in one notebook</p>\n\n<p>P.S. I assume that submit time should be 7x commit runtime so I would like to stay under 70 minutes combined commit time</p>",
  "messages": [
    {
      "id": "612205",
      "postDate": "08/29/2019 16:50:46",
      "content": "<p>as the topic title says - looking for someone (preferably top 100) to form a team</p>\n\n<p>my current score is:\n- 2x Effnet-B4 (classification) + TTA; no Kappa optimization\n- commit runtime around 20-30 minutes</p>\n\n<p>I'm using PyTorch, have 2x RTX2080ti so mostly I only submit when I see good CV</p>\n\n<p>looking for:\n- either different models (Densenet, Resnext etc.) or different Effnet\n- not too many submits (so I can assume you're not trying to overfit to the LB :)\n- runtime (for commit) also at most 30 minutes\n- probably PyTorch also? not sure if it will be easy to combine predictions from PyTorch + Keras / TF in one notebook</p>\n\n<p>P.S. I assume that submit time should be 7x commit runtime so I would like to stay under 70 minutes combined commit time</p>",
      "rawMarkdown": "as the topic title says - looking for someone (preferably top 100) to form a team\n\nmy current score is:\n- 2x Effnet-B4 (classification) + TTA; no Kappa optimization\n- commit runtime around 20-30 minutes\n\nI'm using PyTorch, have 2x RTX2080ti so mostly I only submit when I see good CV\n\nlooking for:\n- either different models (Densenet, Resnext etc.) or different Effnet\n- not too many submits (so I can assume you're not trying to overfit to the LB :)\n- runtime (for commit) also at most 30 minutes\n- probably PyTorch also? not sure if it will be easy to combine predictions from PyTorch + Keras / TF in one notebook\n\nP.S. I assume that submit time should be 7x commit runtime so I would like to stay under 70 minutes combined commit time",
      "votes": null
    },
    {
      "id": "612319",
      "postDate": "08/29/2019 18:06:55",
      "content": "<p>Don't fit all of the criteria but want to give it a shot. \n- My best score is 0.816 with Eff B0(Regression). Current LB score is an ensemble of couple other Efficientnet models. \n- Around 150 submits (Hope you are okay with it.)\n- Didn't keep track of time but I guess they are less than 30 minutes.\n- Using Fastai. But I believe it won't be an issue.  </p>",
      "rawMarkdown": "Don't fit all of the criteria but want to give it a shot. \n- My best score is 0.816 with Eff B0(Regression). Current LB score is an ensemble of couple other Efficientnet models. \n- Around 150 submits (Hope you are okay with it.)\n- Didn't keep track of time but I guess they are less than 30 minutes.\n- Using Fastai. But I believe it won't be an issue.",
      "votes": null
    },
    {
      "id": "612496",
      "postDate": "08/29/2019 20:28:51",
      "content": "<p>Are you interested?  The deadline is in a few hours and it's 2.30 in the morning in my country :/ </p>",
      "rawMarkdown": "Are you interested?  The deadline is in a few hours and it's 2.30 in the morning in my country :/",
      "votes": null
    },
    {
      "id": "612552",
      "postDate": "08/29/2019 21:22:25",
      "content": "<p>Hello... I am interested in forming a team... I know I am on the low end, but I am highly motivated to do well. My current model is just a ResNet50 5fold CV so it might be something useful to add to your ensemble. I also use fastai</p>",
      "rawMarkdown": "Hello... I am interested in forming a team... I know I am on the low end, but I am highly motivated to do well. My current model is just a ResNet50 5fold CV so it might be something useful to add to your ensemble. I also use fastai",
      "votes": null
    },
    {
      "id": "612593",
      "postDate": "08/29/2019 22:13:46",
      "content": "<p>Are you still interested?</p>",
      "rawMarkdown": "Are you still interested?",
      "votes": null
    },
    {
      "id": "612965",
      "postDate": "08/30/2019 05:52:00",
      "content": "<p>sorry guys, somehow wasn't getting email notifications on the replies</p>\n\n<p>good luck to everyone</p>",
      "rawMarkdown": "sorry guys, somehow wasn't getting email notifications on the replies\n\ngood luck to everyone",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 612319,
      "author_name": "tahsin",
      "author_url": "",
      "post_date": "08/29/2019 18:06:55",
      "content": "<p>Don't fit all of the criteria but want to give it a shot. \n- My best score is 0.816 with Eff B0(Regression). Current LB score is an ensemble of couple other Efficientnet models. \n- Around 150 submits (Hope you are okay with it.)\n- Didn't keep track of time but I guess they are less than 30 minutes.\n- Using Fastai. But I believe it won't be an issue.  </p>",
      "votes": null,
      "replies": [
        {
          "id": 612496,
          "author_name": "tahsin",
          "author_url": "",
          "post_date": "08/29/2019 20:28:51",
          "content": "<p>Are you interested?  The deadline is in a few hours and it's 2.30 in the morning in my country :/ </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 612552,
          "author_name": "tanlikesmath",
          "author_url": "",
          "post_date": "08/29/2019 21:22:25",
          "content": "<p>Hello... I am interested in forming a team... I know I am on the low end, but I am highly motivated to do well. My current model is just a ResNet50 5fold CV so it might be something useful to add to your ensemble. I also use fastai</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 612593,
      "author_name": "tanlikesmath",
      "author_url": "",
      "post_date": "08/29/2019 22:13:46",
      "content": "<p>Are you still interested?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 612965,
      "author_name": "steelrose",
      "author_url": "",
      "post_date": "08/30/2019 05:52:00",
      "content": "<p>sorry guys, somehow wasn't getting email notifications on the replies</p>\n\n<p>good luck to everyone</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "612205": "as the topic title says - looking for someone (preferably top 100) to form a team\n\nmy current score is:\n- 2x Effnet-B4 (classification) + TTA; no Kappa optimization\n- commit runtime around 20-30 minutes\n\nI'm using PyTorch, have 2x RTX2080ti so mostly I only submit when I see good CV\n\nlooking for:\n- either different models (Densenet, Resnext etc.) or different Effnet\n- not too many submits (so I can assume you're not trying to overfit to the LB :)\n- runtime (for commit) also at most 30 minutes\n- probably PyTorch also? not sure if it will be easy to combine predictions from PyTorch + Keras / TF in one notebook\n\nP.S. I assume that submit time should be 7x commit runtime so I would like to stay under 70 minutes combined commit time",
    "612319": "Don't fit all of the criteria but want to give it a shot. \n- My best score is 0.816 with Eff B0(Regression). Current LB score is an ensemble of couple other Efficientnet models. \n- Around 150 submits (Hope you are okay with it.)\n- Didn't keep track of time but I guess they are less than 30 minutes.\n- Using Fastai. But I believe it won't be an issue.",
    "612496": "Are you interested?  The deadline is in a few hours and it's 2.30 in the morning in my country :/",
    "612552": "Hello... I am interested in forming a team... I know I am on the low end, but I am highly motivated to do well. My current model is just a ResNet50 5fold CV so it might be something useful to add to your ensemble. I also use fastai",
    "612593": "Are you still interested?",
    "612965": "sorry guys, somehow wasn't getting email notifications on the replies\n\ngood luck to everyone"
  },
  "source": "meta"
}