{
  "id": 306647,
  "title": "Need Help with my first competition",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/306647",
  "author_name": "",
  "post_date": "2022-02-10T09:56:34.117810300Z",
  "votes": 5,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi, this is my first competition and I don't fully understand how the submission process works.<br>\nThe rules say it has to be trhow notebooks that have less than 9 hours of running time. Does that mean that I have to train the model inside the kaggle notebook and it sohuld run for less than 9 hours? Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?</p>\n<p>Thanks.</p>",
  "messages": [
    {
      "id": "1684145",
      "postDate": "02/10/2022 09:56:34",
      "content": "<p>Hi, this is my first competition and I don't fully understand how the submission process works.<br>\nThe rules say it has to be trhow notebooks that have less than 9 hours of running time. Does that mean that I have to train the model inside the kaggle notebook and it sohuld run for less than 9 hours? Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?</p>\n<p>Thanks.</p>",
      "rawMarkdown": "Hi, this is my first competition and I don't fully understand how the submission process works.\nThe rules say it has to be trhow notebooks that have less than 9 hours of running time. Does that mean that I have to train the model inside the kaggle notebook and it sohuld run for less than 9 hours? Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?\n\nThanks.",
      "votes": null
    },
    {
      "id": "1684161",
      "postDate": "02/10/2022 10:09:21",
      "content": "<p>Create 2 separate notebooks for train and test,  just save the weight file from train and load in the inference notebook, you can check this notebook for reference <a href=\"https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-train\" target=\"_blank\">train notebook</a> and <a href=\"https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-infer\" target=\"_blank\">inference notebook</a></p>",
      "rawMarkdown": "Create 2 separate notebooks for train and test,  just save the weight file from train and load in the inference notebook, you can check this notebook for reference [train notebook](https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-train) and [inference notebook](https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-infer)",
      "votes": null
    },
    {
      "id": "1684168",
      "postDate": "02/10/2022 10:12:00",
      "content": "<p>Thanks, but should the training be done in kaggle as well? I was planning to train the model in my own pc.</p>",
      "rawMarkdown": "Thanks, but should the training be done in kaggle as well? I was planning to train the model in my own pc.",
      "votes": null
    },
    {
      "id": "1684170",
      "postDate": "02/10/2022 10:14:35",
      "content": "<blockquote>\n  <p>Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?</p>\n</blockquote>\n<p>You can't directly submit a submission CSV file in this competition, You can train a model anywhere, create a dataset, add your model weight file and use it in the inference notebook.</p>",
      "rawMarkdown": "> Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?\n\nYou can't directly submit a submission CSV file in this competition, You can train a model anywhere, create a dataset, add your model weight file and use it in the inference notebook.",
      "votes": null
    },
    {
      "id": "1684201",
      "postDate": "02/10/2022 10:37:33",
      "content": "<p>Given the competition ends in 5 days, you may consider switching to another competition with a longer deadline …</p>",
      "rawMarkdown": "Given the competition ends in 5 days, you may consider switching to another competition with a longer deadline ...",
      "votes": null
    },
    {
      "id": "1684229",
      "postDate": "02/10/2022 10:54:26",
      "content": "<p>I have the training coded, I only need to make the inference and submission.</p>",
      "rawMarkdown": "I have the training coded, I only need to make the inference and submission.",
      "votes": null
    },
    {
      "id": "1684816",
      "postDate": "02/10/2022 18:28:09",
      "content": "<p>You are allowed to train the model from anywhere (Colab, local machine, AWS, GCP, etc) and get the trained weights .pt. It is not required to train on Kaggle cloud, and there is no limitation for the training time. The 9 hours limitation is for running inference and scoring of your submission.</p>\n<p>Once you have the weight, you should create a complete inference notebook on Kaggle, use the API they provided for running inference on the test set and make submissions. The API is designed to hide the test set, and only 3 images will be available for you to see.</p>\n<p>Once you run the notebook, click \"Submit\" in the \"Competitions\" tab on the right, and wait for a while for the scoring. It will not be instant like some competitions, and it may take anywhere from 1 to 9 hours depending on your model.</p>",
      "rawMarkdown": "You are allowed to train the model from anywhere (Colab, local machine, AWS, GCP, etc) and get the trained weights .pt. It is not required to train on Kaggle cloud, and there is no limitation for the training time. The 9 hours limitation is for running inference and scoring of your submission.\n\nOnce you have the weight, you should create a complete inference notebook on Kaggle, use the API they provided for running inference on the test set and make submissions. The API is designed to hide the test set, and only 3 images will be available for you to see.\n\nOnce you run the notebook, click \"Submit\" in the \"Competitions\" tab on the right, and wait for a while for the scoring. It will not be instant like some competitions, and it may take anywhere from 1 to 9 hours depending on your model.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1684161,
      "author_name": "mahipalsingh",
      "author_url": "",
      "post_date": "02/10/2022 10:09:21",
      "content": "<p>Create 2 separate notebooks for train and test,  just save the weight file from train and load in the inference notebook, you can check this notebook for reference <a href=\"https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-train\" target=\"_blank\">train notebook</a> and <a href=\"https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-infer\" target=\"_blank\">inference notebook</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 1684168,
          "author_name": "ricardfos",
          "author_url": "",
          "post_date": "02/10/2022 10:12:00",
          "content": "<p>Thanks, but should the training be done in kaggle as well? I was planning to train the model in my own pc.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1684170,
          "author_name": "mahipalsingh",
          "author_url": "",
          "post_date": "02/10/2022 10:14:35",
          "content": "<blockquote>\n  <p>Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?</p>\n</blockquote>\n<p>You can't directly submit a submission CSV file in this competition, You can train a model anywhere, create a dataset, add your model weight file and use it in the inference notebook.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1684201,
      "author_name": "fabiendaniel",
      "author_url": "",
      "post_date": "02/10/2022 10:37:33",
      "content": "<p>Given the competition ends in 5 days, you may consider switching to another competition with a longer deadline …</p>",
      "votes": null,
      "replies": [
        {
          "id": 1684229,
          "author_name": "ricardfos",
          "author_url": "",
          "post_date": "02/10/2022 10:54:26",
          "content": "<p>I have the training coded, I only need to make the inference and submission.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1684816,
      "author_name": "anhnhunhat",
      "author_url": "",
      "post_date": "02/10/2022 18:28:09",
      "content": "<p>You are allowed to train the model from anywhere (Colab, local machine, AWS, GCP, etc) and get the trained weights .pt. It is not required to train on Kaggle cloud, and there is no limitation for the training time. The 9 hours limitation is for running inference and scoring of your submission.</p>\n<p>Once you have the weight, you should create a complete inference notebook on Kaggle, use the API they provided for running inference on the test set and make submissions. The API is designed to hide the test set, and only 3 images will be available for you to see.</p>\n<p>Once you run the notebook, click \"Submit\" in the \"Competitions\" tab on the right, and wait for a while for the scoring. It will not be instant like some competitions, and it may take anywhere from 1 to 9 hours depending on your model.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1684145": "Hi, this is my first competition and I don't fully understand how the submission process works.\nThe rules say it has to be trhow notebooks that have less than 9 hours of running time. Does that mean that I have to train the model inside the kaggle notebook and it sohuld run for less than 9 hours? Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?\n\nThanks.",
    "1684161": "Create 2 separate notebooks for train and test,  just save the weight file from train and load in the inference notebook, you can check this notebook for reference [train notebook](https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-train) and [inference notebook](https://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-infer)",
    "1684168": "Thanks, but should the training be done in kaggle as well? I was planning to train the model in my own pc.",
    "1684170": "> Can't I train the model outside, get the csv output file and then make a kaggle notebook using that file from my pc to do the submission?\n\nYou can't directly submit a submission CSV file in this competition, You can train a model anywhere, create a dataset, add your model weight file and use it in the inference notebook.",
    "1684201": "Given the competition ends in 5 days, you may consider switching to another competition with a longer deadline ...",
    "1684229": "I have the training coded, I only need to make the inference and submission.",
    "1684816": "You are allowed to train the model from anywhere (Colab, local machine, AWS, GCP, etc) and get the trained weights .pt. It is not required to train on Kaggle cloud, and there is no limitation for the training time. The 9 hours limitation is for running inference and scoring of your submission.\n\nOnce you have the weight, you should create a complete inference notebook on Kaggle, use the API they provided for running inference on the test set and make submissions. The API is designed to hide the test set, and only 3 images will be available for you to see.\n\nOnce you run the notebook, click \"Submit\" in the \"Competitions\" tab on the right, and wait for a while for the scoring. It will not be instant like some competitions, and it may take anywhere from 1 to 9 hours depending on your model."
  },
  "source": "meta"
}