{
  "id": 170343,
  "title": "How does pre-trained models work?",
  "url": "/competitions/birdsong-recognition/discussion/170343",
  "author_name": "",
  "post_date": "2020-07-27T10:16:43.244761800Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>It is my first code competition, and I would like to know how do external pre-trained models work. Can I train the complete model locally and then just upload it to the notebook (without any further post-training) ? </p>",
  "messages": [
    {
      "id": "947526",
      "postDate": "07/27/2020 10:16:43",
      "content": "<p>It is my first code competition, and I would like to know how do external pre-trained models work. Can I train the complete model locally and then just upload it to the notebook (without any further post-training) ? </p>",
      "rawMarkdown": "It is my first code competition, and I would like to know how do external pre-trained models work. Can I train the complete model locally and then just upload it to the notebook (without any further post-training) ?",
      "votes": null
    },
    {
      "id": "947559",
      "postDate": "07/27/2020 10:44:58",
      "content": "<p>Yes you can. Most inference notebooks will use pre-trained models added as datasets.</p>",
      "rawMarkdown": "Yes you can. Most inference notebooks will use pre-trained models added as datasets.",
      "votes": null
    },
    {
      "id": "947688",
      "postDate": "07/27/2020 12:32:03",
      "content": "<p>Yes you can. You need to save your trained model as a file (.pth, .pkl etc). Then, create your own dataset using\"Data\". Upload your model file to your created dataset. Then attach that dataset to your notebook. Your notebook will be able to read the model as a file</p>",
      "rawMarkdown": "Yes you can. You need to save your trained model as a file (.pth, .pkl etc). Then, create your own dataset using\"Data\". Upload your model file to your created dataset. Then attach that dataset to your notebook. Your notebook will be able to read the model as a file",
      "votes": null
    },
    {
      "id": "947738",
      "postDate": "07/27/2020 13:04:49",
      "content": "<p>That's great, thanks for your help!!!</p>",
      "rawMarkdown": "That's great, thanks for your help!!!",
      "votes": null
    },
    {
      "id": "947739",
      "postDate": "07/27/2020 13:05:47",
      "content": "<p>Okay, so the notebook is only used as inference. Thanks!</p>",
      "rawMarkdown": "Okay, so the notebook is only used as inference. Thanks!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 947559,
      "author_name": "rohanrao",
      "author_url": "",
      "post_date": "07/27/2020 10:44:58",
      "content": "<p>Yes you can. Most inference notebooks will use pre-trained models added as datasets.</p>",
      "votes": null,
      "replies": [
        {
          "id": 947739,
          "author_name": "polmonroig",
          "author_url": "",
          "post_date": "07/27/2020 13:05:47",
          "content": "<p>Okay, so the notebook is only used as inference. Thanks!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 947688,
      "author_name": "alanchn31",
      "author_url": "",
      "post_date": "07/27/2020 12:32:03",
      "content": "<p>Yes you can. You need to save your trained model as a file (.pth, .pkl etc). Then, create your own dataset using\"Data\". Upload your model file to your created dataset. Then attach that dataset to your notebook. Your notebook will be able to read the model as a file</p>",
      "votes": null,
      "replies": [
        {
          "id": 947738,
          "author_name": "polmonroig",
          "author_url": "",
          "post_date": "07/27/2020 13:04:49",
          "content": "<p>That's great, thanks for your help!!!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "947526": "It is my first code competition, and I would like to know how do external pre-trained models work. Can I train the complete model locally and then just upload it to the notebook (without any further post-training) ?",
    "947559": "Yes you can. Most inference notebooks will use pre-trained models added as datasets.",
    "947688": "Yes you can. You need to save your trained model as a file (.pth, .pkl etc). Then, create your own dataset using\"Data\". Upload your model file to your created dataset. Then attach that dataset to your notebook. Your notebook will be able to read the model as a file",
    "947738": "That's great, thanks for your help!!!",
    "947739": "Okay, so the notebook is only used as inference. Thanks!"
  },
  "source": "meta"
}