{
  "id": 135881,
  "title": "Regarding Submission",
  "url": "/competitions/deepfake-detection-challenge/discussion/135881",
  "author_name": "",
  "post_date": "2020-03-16T14:29:04.830191400Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>As this competition is a code submission competition, I have a couple of concerns regarding submission:\n- Can I upload my trained model? \n - if yes: \n        - will the trained model count for 1GB external data source limit?\n        - Is the 9hours time limit testing time only?\n - If no:\n        - that means I need to retrain the model and test within 9hours time limit?</p>\n\n<p><strong>I am actually confused about how the testing will work. Any resources and suggestions are most welcomed.</strong>\nPS. I am very novice in Kaggle platform, sorry for asking such trivial details.</p>",
  "messages": [
    {
      "id": "775308",
      "postDate": "03/16/2020 14:29:04",
      "content": "<p>As this competition is a code submission competition, I have a couple of concerns regarding submission:\n- Can I upload my trained model? \n - if yes: \n        - will the trained model count for 1GB external data source limit?\n        - Is the 9hours time limit testing time only?\n - If no:\n        - that means I need to retrain the model and test within 9hours time limit?</p>\n\n<p><strong>I am actually confused about how the testing will work. Any resources and suggestions are most welcomed.</strong>\nPS. I am very novice in Kaggle platform, sorry for asking such trivial details.</p>",
      "rawMarkdown": "As this competition is a code submission competition, I have a couple of concerns regarding submission:\n- Can I upload my trained model? \n - if yes: \n        - will the trained model count for 1GB external data source limit?\n        - Is the 9hours time limit testing time only?\n - If no:\n        - that means I need to retrain the model and test within 9hours time limit?\n\n**I am actually confused about how the testing will work. Any resources and suggestions are most welcomed.**\nPS. I am very novice in Kaggle platform, sorry for asking such trivial details.",
      "votes": null
    },
    {
      "id": "775364",
      "postDate": "03/16/2020 15:51:01",
      "content": "<p>You can upload your model as a dataset and use it for inference. Models count toward the 1GB limit. A suggestion for testing would be to look up guides on how to use libraries in offline mode for kaggle notebooks. The kaggle notebooks already come with some libraries installed such as tensorflow 2.1, pytorch, and opencv2. But typically you will need to install a face recognition library that works in offline mode. Another tip would be to use the notebook instead of a script when submitting for the competition because the scripts throw a lot errors when committing them (at least that is my experience with them).</p>",
      "rawMarkdown": "You can upload your model as a dataset and use it for inference. Models count toward the 1GB limit. A suggestion for testing would be to look up guides on how to use libraries in offline mode for kaggle notebooks. The kaggle notebooks already come with some libraries installed such as tensorflow 2.1, pytorch, and opencv2. But typically you will need to install a face recognition library that works in offline mode. Another tip would be to use the notebook instead of a script when submitting for the competition because the scripts throw a lot errors when committing them (at least that is my experience with them).",
      "votes": null
    },
    {
      "id": "775425",
      "postDate": "03/16/2020 17:00:51",
      "content": "<p>Thanks a lot for the detailed information </p>",
      "rawMarkdown": "Thanks a lot for the detailed information",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 775364,
      "author_name": "davidmilam",
      "author_url": "",
      "post_date": "03/16/2020 15:51:01",
      "content": "<p>You can upload your model as a dataset and use it for inference. Models count toward the 1GB limit. A suggestion for testing would be to look up guides on how to use libraries in offline mode for kaggle notebooks. The kaggle notebooks already come with some libraries installed such as tensorflow 2.1, pytorch, and opencv2. But typically you will need to install a face recognition library that works in offline mode. Another tip would be to use the notebook instead of a script when submitting for the competition because the scripts throw a lot errors when committing them (at least that is my experience with them).</p>",
      "votes": null,
      "replies": [
        {
          "id": 775425,
          "author_name": "beingmiakashs",
          "author_url": "",
          "post_date": "03/16/2020 17:00:51",
          "content": "<p>Thanks a lot for the detailed information </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "775308": "As this competition is a code submission competition, I have a couple of concerns regarding submission:\n- Can I upload my trained model? \n - if yes: \n        - will the trained model count for 1GB external data source limit?\n        - Is the 9hours time limit testing time only?\n - If no:\n        - that means I need to retrain the model and test within 9hours time limit?\n\n**I am actually confused about how the testing will work. Any resources and suggestions are most welcomed.**\nPS. I am very novice in Kaggle platform, sorry for asking such trivial details.",
    "775364": "You can upload your model as a dataset and use it for inference. Models count toward the 1GB limit. A suggestion for testing would be to look up guides on how to use libraries in offline mode for kaggle notebooks. The kaggle notebooks already come with some libraries installed such as tensorflow 2.1, pytorch, and opencv2. But typically you will need to install a face recognition library that works in offline mode. Another tip would be to use the notebook instead of a script when submitting for the competition because the scripts throw a lot errors when committing them (at least that is my experience with them).",
    "775425": "Thanks a lot for the detailed information"
  },
  "source": "meta"
}