{
  "id": 316859,
  "title": "Which backbone network works better, ConvNext or EffNet？",
  "url": "/competitions/happy-whale-and-dolphin/discussion/316859",
  "author_name": "",
  "post_date": "2022-04-04T09:40:48.581628300Z",
  "votes": 9,
  "comment_count": 5,
  "views": 0,
  "content": "<p>What kind of backbone network are you using?🤔</p>",
  "messages": [
    {
      "id": "1744768",
      "postDate": "04/04/2022 09:40:48",
      "content": "<p>What kind of backbone network are you using?🤔</p>",
      "rawMarkdown": "What kind of backbone network are you using?🤔",
      "votes": null
    },
    {
      "id": "1745040",
      "postDate": "04/04/2022 14:38:57",
      "content": "<p>For me EffNet better than ConvNext here</p>",
      "rawMarkdown": "For me EffNet better than ConvNext here",
      "votes": null
    },
    {
      "id": "1746722",
      "postDate": "04/06/2022 03:19:18",
      "content": "<p>is Batch size or learning scheduler more important?  I tuned the  s m  of arcface , it seems  related to val loss/acc, but not much with LB.  </p>",
      "rawMarkdown": "is Batch size or learning scheduler more important?  I tuned the  s m  of arcface , it seems  related to val loss/acc, but not much with LB.",
      "votes": null
    },
    {
      "id": "1746771",
      "postDate": "04/06/2022 04:54:51",
      "content": "<p>I am watching on your contribution <a href=\"https://www.kaggle.com/kwentar\" target=\"_blank\">@kwentar</a> in this competition and it is really great. Even though you are on TOP5 you share. Great! </p>",
      "rawMarkdown": "I am watching on your contribution @kwentar in this competition and it is really great. Even though you are on TOP5 you share. Great!",
      "votes": null
    },
    {
      "id": "1746905",
      "postDate": "04/06/2022 07:45:54",
      "content": "<p><a href=\"https://www.kaggle.com/dragonzhang\" target=\"_blank\">@dragonzhang</a> lr/batch and s/m are orthogonal, I cannot say which one is more important, both are worthy<br>\n<a href=\"https://www.kaggle.com/remekkinas\" target=\"_blank\">@remekkinas</a> thank you!</p>",
      "rawMarkdown": "dragonzhang lr/batch and s/m are orthogonal, I cannot say which one is more important, both are worthy\n@remekkinas thank you!",
      "votes": null
    },
    {
      "id": "1746933",
      "postDate": "04/06/2022 08:09:04",
      "content": "<p>I get 0.758 LB using single convnext_base model.I'd like to try effnet for last few days.😄</p>",
      "rawMarkdown": "I get 0.758 LB using single convnext_base model.I'd like to try effnet for last few days.😄",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1745040,
      "author_name": "kwentar",
      "author_url": "",
      "post_date": "04/04/2022 14:38:57",
      "content": "<p>For me EffNet better than ConvNext here</p>",
      "votes": null,
      "replies": [
        {
          "id": 1746722,
          "author_name": "dragonzhang",
          "author_url": "",
          "post_date": "04/06/2022 03:19:18",
          "content": "<p>is Batch size or learning scheduler more important?  I tuned the  s m  of arcface , it seems  related to val loss/acc, but not much with LB.  </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1746771,
          "author_name": "remekkinas",
          "author_url": "",
          "post_date": "04/06/2022 04:54:51",
          "content": "<p>I am watching on your contribution <a href=\"https://www.kaggle.com/kwentar\" target=\"_blank\">@kwentar</a> in this competition and it is really great. Even though you are on TOP5 you share. Great! </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1746905,
          "author_name": "kwentar",
          "author_url": "",
          "post_date": "04/06/2022 07:45:54",
          "content": "<p><a href=\"https://www.kaggle.com/dragonzhang\" target=\"_blank\">@dragonzhang</a> lr/batch and s/m are orthogonal, I cannot say which one is more important, both are worthy<br>\n<a href=\"https://www.kaggle.com/remekkinas\" target=\"_blank\">@remekkinas</a> thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1746933,
      "author_name": "jimmysmith1009",
      "author_url": "",
      "post_date": "04/06/2022 08:09:04",
      "content": "<p>I get 0.758 LB using single convnext_base model.I'd like to try effnet for last few days.😄</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1744768": "What kind of backbone network are you using?🤔",
    "1745040": "For me EffNet better than ConvNext here",
    "1746722": "is Batch size or learning scheduler more important?  I tuned the  s m  of arcface , it seems  related to val loss/acc, but not much with LB.",
    "1746771": "I am watching on your contribution @kwentar in this competition and it is really great. Even though you are on TOP5 you share. Great!",
    "1746905": "dragonzhang lr/batch and s/m are orthogonal, I cannot say which one is more important, both are worthy\n@remekkinas thank you!",
    "1746933": "I get 0.758 LB using single convnext_base model.I'd like to try effnet for last few days.😄"
  },
  "source": "meta"
}