{
  "id": 140728,
  "title": "notebook requirement for this challenge",
  "url": "/competitions/flower-classification-with-tpus/discussion/140728",
  "author_name": "",
  "post_date": "2020-04-03T03:39:19.129982200Z",
  "votes": 7,
  "comment_count": 5,
  "views": 0,
  "content": "<p>It is mentioned that:</p>\n\n<p>\"Submissions to this competition must be made by an output from a Kaggle Notebook/Script.\nNotebook run-time per session capped at 3 hours.\"</p>\n\n<p>\"To be eligible for these prizes, your submission must generate predictions from machine learning code fully run on TPUs for both training and inference. That code must be made public following the end of the competition.\"</p>\n\n<hr>\n\n<p>to be eligible for prize, our solution can consist of one only one notebook run (i.e. everything strictly within 3 hr)?</p>\n\n<p>how about multiple notebook runs? e.g. training multiple models, each under 3 hr (train sequentially).Then we import the trained models as external data and ensemble, etc</p>",
  "messages": [
    {
      "id": "795784",
      "postDate": "04/03/2020 03:39:19",
      "content": "<p>It is mentioned that:</p>\n\n<p>\"Submissions to this competition must be made by an output from a Kaggle Notebook/Script.\nNotebook run-time per session capped at 3 hours.\"</p>\n\n<p>\"To be eligible for these prizes, your submission must generate predictions from machine learning code fully run on TPUs for both training and inference. That code must be made public following the end of the competition.\"</p>\n\n<hr>\n\n<p>to be eligible for prize, our solution can consist of one only one notebook run (i.e. everything strictly within 3 hr)?</p>\n\n<p>how about multiple notebook runs? e.g. training multiple models, each under 3 hr (train sequentially).Then we import the trained models as external data and ensemble, etc</p>",
      "rawMarkdown": "It is mentioned that:\n\n\"Submissions to this competition must be made by an output from a Kaggle Notebook/Script.\nNotebook run-time per session capped at 3 hours.\"\n\n\"To be eligible for these prizes, your submission must generate predictions from machine learning code fully run on TPUs for both training and inference. That code must be made public following the end of the competition.\"\n\n---\n\nto be eligible for prize, our solution can consist of one only one notebook run (i.e. everything strictly within 3 hr)?\n\nhow about multiple notebook runs? e.g. training multiple models, each under 3 hr (train sequentially).Then we import the trained models as external data and ensemble, etc",
      "votes": null
    },
    {
      "id": "796288",
      "postDate": "04/03/2020 13:19:00",
      "content": "<p>I didn't notice \"both training and inference\" before! \nSo does this mean if we want to ensemble the models, we need to train all models in one notebook and the time can't exceed 3 hours.</p>",
      "rawMarkdown": "I didn't notice \"both training and inference\" before! \nSo does this mean if we want to ensemble the models, we need to train all models in one notebook and the time can't exceed 3 hours.",
      "votes": null
    },
    {
      "id": "796711",
      "postDate": "04/03/2020 20:45:19",
      "content": "<p><a href=\"/hengck23\">@hengck23</a> Yes, you may combine/ensemble/string multiple notebooks, as long as all of the code, end-to-end, runs on TPUs (if you want to be prize-eligible). </p>",
      "rawMarkdown": "hengck23 Yes, you may combine/ensemble/string multiple notebooks, as long as all of the code, end-to-end, runs on TPUs (if you want to be prize-eligible).",
      "votes": null
    },
    {
      "id": "803248",
      "postDate": "04/10/2020 09:32:15",
      "content": "<p>And stays within the 3 hours when running end-to-end I assume?</p>",
      "rawMarkdown": "And stays within the 3 hours when running end-to-end I assume?",
      "votes": null
    },
    {
      "id": "803771",
      "postDate": "04/10/2020 20:25:17",
      "content": "<p>Clarification on the maximum total runtime (combined notebooks) would be much appreciated <a href=\"/juliaelliott\">@juliaelliott</a> </p>\n\n<p>Edit: I saw your update above. Thank you!</p>",
      "rawMarkdown": "Clarification on the maximum total runtime (combined notebooks) would be much appreciated @juliaelliott \n\nEdit: I saw your update above. Thank you!",
      "votes": null
    },
    {
      "id": "803794",
      "postDate": "04/10/2020 21:02:07",
      "content": "<p><a href=\"/muaddev\">@muaddev</a> No, The 3 hour limit is a systematically-imposed session and notebook limit. It does not mean that your entire end-to-end model must run in 3 hours to be prize-eligible.</p>",
      "rawMarkdown": "muaddev No, The 3 hour limit is a systematically-imposed session and notebook limit. It does not mean that your entire end-to-end model must run in 3 hours to be prize-eligible.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 796288,
      "author_name": "xiejialun",
      "author_url": "",
      "post_date": "04/03/2020 13:19:00",
      "content": "<p>I didn't notice \"both training and inference\" before! \nSo does this mean if we want to ensemble the models, we need to train all models in one notebook and the time can't exceed 3 hours.</p>",
      "votes": null,
      "replies": [
        {
          "id": 803771,
          "author_name": "ilu000",
          "author_url": "",
          "post_date": "04/10/2020 20:25:17",
          "content": "<p>Clarification on the maximum total runtime (combined notebooks) would be much appreciated <a href=\"/juliaelliott\">@juliaelliott</a> </p>\n\n<p>Edit: I saw your update above. Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 796711,
      "author_name": "juliaelliott",
      "author_url": "",
      "post_date": "04/03/2020 20:45:19",
      "content": "<p><a href=\"/hengck23\">@hengck23</a> Yes, you may combine/ensemble/string multiple notebooks, as long as all of the code, end-to-end, runs on TPUs (if you want to be prize-eligible). </p>",
      "votes": null,
      "replies": [
        {
          "id": 803248,
          "author_name": "muaddev",
          "author_url": "",
          "post_date": "04/10/2020 09:32:15",
          "content": "<p>And stays within the 3 hours when running end-to-end I assume?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 803794,
          "author_name": "juliaelliott",
          "author_url": "",
          "post_date": "04/10/2020 21:02:07",
          "content": "<p><a href=\"/muaddev\">@muaddev</a> No, The 3 hour limit is a systematically-imposed session and notebook limit. It does not mean that your entire end-to-end model must run in 3 hours to be prize-eligible.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "795784": "It is mentioned that:\n\n\"Submissions to this competition must be made by an output from a Kaggle Notebook/Script.\nNotebook run-time per session capped at 3 hours.\"\n\n\"To be eligible for these prizes, your submission must generate predictions from machine learning code fully run on TPUs for both training and inference. That code must be made public following the end of the competition.\"\n\n---\n\nto be eligible for prize, our solution can consist of one only one notebook run (i.e. everything strictly within 3 hr)?\n\nhow about multiple notebook runs? e.g. training multiple models, each under 3 hr (train sequentially).Then we import the trained models as external data and ensemble, etc",
    "796288": "I didn't notice \"both training and inference\" before! \nSo does this mean if we want to ensemble the models, we need to train all models in one notebook and the time can't exceed 3 hours.",
    "796711": "hengck23 Yes, you may combine/ensemble/string multiple notebooks, as long as all of the code, end-to-end, runs on TPUs (if you want to be prize-eligible).",
    "803248": "And stays within the 3 hours when running end-to-end I assume?",
    "803771": "Clarification on the maximum total runtime (combined notebooks) would be much appreciated @juliaelliott \n\nEdit: I saw your update above. Thank you!",
    "803794": "muaddev No, The 3 hour limit is a systematically-imposed session and notebook limit. It does not mean that your entire end-to-end model must run in 3 hours to be prize-eligible."
  },
  "source": "meta"
}