{
  "id": 438975,
  "title": "Questions about notebook submissions",
  "url": "/competitions/child-mind-institute-detect-sleep-states/discussion/438975",
  "author_name": "",
  "post_date": "2023-09-13T08:38:16.542814200Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi everyone! We are new on kaggle, and have a few questions about the submission procedure on kaggle. We saw that we have to submit a single notebook that is run on kaggle's server with some CPU and GPU time constraints. We have a few questions:</p>\n<ul>\n<li>Can we upload our own pre-trained models? This would be nice, since they are locally trained which then can be used on the notebook? Or can we only use pretrained models that kaggle provides in the notebook interface?</li>\n<li>In which directory should we output our submission.csv? in the /kaggle/working dir?</li>\n<li>Can we upload our own datasets?</li>\n<li>How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..</li>\n<li>Are we allowed to upload other config / files to the online notebook kaggle environment?</li>\n</ul>\n<p>Thanks in advance!</p>",
  "messages": [
    {
      "id": "2435899",
      "postDate": "09/13/2023 08:38:16",
      "content": "<p>Hi everyone! We are new on kaggle, and have a few questions about the submission procedure on kaggle. We saw that we have to submit a single notebook that is run on kaggle's server with some CPU and GPU time constraints. We have a few questions:</p>\n<ul>\n<li>Can we upload our own pre-trained models? This would be nice, since they are locally trained which then can be used on the notebook? Or can we only use pretrained models that kaggle provides in the notebook interface?</li>\n<li>In which directory should we output our submission.csv? in the /kaggle/working dir?</li>\n<li>Can we upload our own datasets?</li>\n<li>How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..</li>\n<li>Are we allowed to upload other config / files to the online notebook kaggle environment?</li>\n</ul>\n<p>Thanks in advance!</p>",
      "rawMarkdown": "Hi everyone! We are new on kaggle, and have a few questions about the submission procedure on kaggle. We saw that we have to submit a single notebook that is run on kaggle's server with some CPU and GPU time constraints. We have a few questions:\n\n- Can we upload our own pre-trained models? This would be nice, since they are locally trained which then can be used on the notebook? Or can we only use pretrained models that kaggle provides in the notebook interface?\n- In which directory should we output our submission.csv? in the /kaggle/working dir?\n- Can we upload our own datasets?\n- How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..\n- Are we allowed to upload other config / files to the online notebook kaggle environment?\n\nThanks in advance!",
      "votes": null
    },
    {
      "id": "2435909",
      "postDate": "09/13/2023 08:48:31",
      "content": "<p><a href=\"https://www.kaggle.com/hugodeheer\" target=\"_blank\">@hugodeheer</a> some responses from my end- </p>\n<ol>\n<li><strong>Can we upload our own pre-trained models?</strong>\nThis is the preferred way. You may train elsewhere and load the results as a dataset/ model in your inference kernel and submit it. It will save hardware and GPU quotas a lot\nYou can train your own models as you desire/ use existing models- this is your decision completely</li>\n<li><strong>In which directory should we output our submission.csv? in the /kaggle/working dir?</strong>\nYou need to save the notebook (commit it) and then submit to the competition for scoring. As a general caution, please ensure that you don't create any bugs in code as an <strong>erroneous submission is counted as a submission</strong>. Double checking for semantic errors is good to avoid misgivings and misadventures.</li>\n<li><strong>Can we upload our own datasets?</strong>\nThis is another preferred way. You may ensue pre-processing as a dataset and use it as a dataset off a kernel/ individual dataset in your model training and inferencing</li>\n<li><strong>How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..</strong>\nSorry, I think others may provide you a better rejoinder for this. </li>\n<li><strong>Are we allowed to upload other config / files to the online notebook kaggle environment?</strong>\nYou are allowed to do so if you stay complaint with the competition rules and associated licensing agreements. As a general practice, participants usually upload extra packages as datasets and import them using <strong>sys</strong>. </li>\n</ol>\n<p>Hope I was able to answer you to an extent. I am sure other regular participants could educate you better in this regard. \nAll the best and happy learning!</p>",
      "rawMarkdown": "hugodeheer some responses from my end- \n1. **Can we upload our own pre-trained models?**\nThis is the preferred way. You may train elsewhere and load the results as a dataset/ model in your inference kernel and submit it. It will save hardware and GPU quotas a lot\nYou can train your own models as you desire/ use existing models- this is your decision completely\n2. **In which directory should we output our submission.csv? in the /kaggle/working dir?**\nYou need to save the notebook (commit it) and then submit to the competition for scoring. As a general caution, please ensure that you don't create any bugs in code as an **erroneous submission is counted as a submission**. Double checking for semantic errors is good to avoid misgivings and misadventures.\n3. **Can we upload our own datasets?**\nThis is another preferred way. You may ensue pre-processing as a dataset and use it as a dataset off a kernel/ individual dataset in your model training and inferencing\n4. **How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..**\nSorry, I think others may provide you a better rejoinder for this. \n5. **Are we allowed to upload other config / files to the online notebook kaggle environment?**\nYou are allowed to do so if you stay complaint with the competition rules and associated licensing agreements. As a general practice, participants usually upload extra packages as datasets and import them using **sys**. \n\nHope I was able to answer you to an extent. I am sure other regular participants could educate you better in this regard. \nAll the best and happy learning!",
      "votes": null
    },
    {
      "id": "2435924",
      "postDate": "09/13/2023 08:57:07",
      "content": "<p>Thanks for your quick response! Very helpful :)</p>",
      "rawMarkdown": "Thanks for your quick response! Very helpful :)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2435909,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "09/13/2023 08:48:31",
      "content": "<p><a href=\"https://www.kaggle.com/hugodeheer\" target=\"_blank\">@hugodeheer</a> some responses from my end- </p>\n<ol>\n<li><strong>Can we upload our own pre-trained models?</strong>\nThis is the preferred way. You may train elsewhere and load the results as a dataset/ model in your inference kernel and submit it. It will save hardware and GPU quotas a lot\nYou can train your own models as you desire/ use existing models- this is your decision completely</li>\n<li><strong>In which directory should we output our submission.csv? in the /kaggle/working dir?</strong>\nYou need to save the notebook (commit it) and then submit to the competition for scoring. As a general caution, please ensure that you don't create any bugs in code as an <strong>erroneous submission is counted as a submission</strong>. Double checking for semantic errors is good to avoid misgivings and misadventures.</li>\n<li><strong>Can we upload our own datasets?</strong>\nThis is another preferred way. You may ensue pre-processing as a dataset and use it as a dataset off a kernel/ individual dataset in your model training and inferencing</li>\n<li><strong>How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..</strong>\nSorry, I think others may provide you a better rejoinder for this. </li>\n<li><strong>Are we allowed to upload other config / files to the online notebook kaggle environment?</strong>\nYou are allowed to do so if you stay complaint with the competition rules and associated licensing agreements. As a general practice, participants usually upload extra packages as datasets and import them using <strong>sys</strong>. </li>\n</ol>\n<p>Hope I was able to answer you to an extent. I am sure other regular participants could educate you better in this regard. \nAll the best and happy learning!</p>",
      "votes": null,
      "replies": [
        {
          "id": 2435924,
          "author_name": "hugodeheer",
          "author_url": "",
          "post_date": "09/13/2023 08:57:07",
          "content": "<p>Thanks for your quick response! Very helpful :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2435899": "Hi everyone! We are new on kaggle, and have a few questions about the submission procedure on kaggle. We saw that we have to submit a single notebook that is run on kaggle's server with some CPU and GPU time constraints. We have a few questions:\n\n- Can we upload our own pre-trained models? This would be nice, since they are locally trained which then can be used on the notebook? Or can we only use pretrained models that kaggle provides in the notebook interface?\n- In which directory should we output our submission.csv? in the /kaggle/working dir?\n- Can we upload our own datasets?\n- How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..\n- Are we allowed to upload other config / files to the online notebook kaggle environment?\n\nThanks in advance!",
    "2435909": "hugodeheer some responses from my end- \n1. **Can we upload our own pre-trained models?**\nThis is the preferred way. You may train elsewhere and load the results as a dataset/ model in your inference kernel and submit it. It will save hardware and GPU quotas a lot\nYou can train your own models as you desire/ use existing models- this is your decision completely\n2. **In which directory should we output our submission.csv? in the /kaggle/working dir?**\nYou need to save the notebook (commit it) and then submit to the competition for scoring. As a general caution, please ensure that you don't create any bugs in code as an **erroneous submission is counted as a submission**. Double checking for semantic errors is good to avoid misgivings and misadventures.\n3. **Can we upload our own datasets?**\nThis is another preferred way. You may ensue pre-processing as a dataset and use it as a dataset off a kernel/ individual dataset in your model training and inferencing\n4. **How do other teams collaborate using Git-repositories? Since working together in a notebook is not ideal with git..**\nSorry, I think others may provide you a better rejoinder for this. \n5. **Are we allowed to upload other config / files to the online notebook kaggle environment?**\nYou are allowed to do so if you stay complaint with the competition rules and associated licensing agreements. As a general practice, participants usually upload extra packages as datasets and import them using **sys**. \n\nHope I was able to answer you to an extent. I am sure other regular participants could educate you better in this regard. \nAll the best and happy learning!",
    "2435924": "Thanks for your quick response! Very helpful :)"
  },
  "source": "meta"
}