{
  "id": 131061,
  "title": "CPU inference kernel commit time 63 seconds, but submission timed out ",
  "url": "/competitions/bengaliai-cv19/discussion/131061",
  "author_name": "",
  "post_date": "2020-02-18T02:41:15.104657600Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I committed an inference Kernel using CPU and it finished successfully in 63 seconds with output submission.csv I can see. When I tried submitting it, the submission timed out after 9 hours. I don't understand the big time difference. Isn't private testset only slightly larger than the public testset (46% vs 54%)?  Thank you for any explanation that helps me understand.</p>",
  "messages": [
    {
      "id": "748812",
      "postDate": "02/18/2020 02:41:15",
      "content": "<p>I committed an inference Kernel using CPU and it finished successfully in 63 seconds with output submission.csv I can see. When I tried submitting it, the submission timed out after 9 hours. I don't understand the big time difference. Isn't private testset only slightly larger than the public testset (46% vs 54%)?  Thank you for any explanation that helps me understand.</p>",
      "rawMarkdown": "I committed an inference Kernel using CPU and it finished successfully in 63 seconds with output submission.csv I can see. When I tried submitting it, the submission timed out after 9 hours. I don't understand the big time difference. Isn't private testset only slightly larger than the public testset (46% vs 54%)?  Thank you for any explanation that helps me understand.",
      "votes": null
    },
    {
      "id": "748815",
      "postDate": "02/18/2020 02:49:33",
      "content": "<p>The <code>test.csv</code> and <code>sample_submission.csv</code> that we can download and see is only 36 rows. It is neither the public test nor the private test. Both the public and private are hidden from us. The approximate size of the test data is 200,000 rows, so you should time how long it takes your model to predict the entire training data of 200,000 rows.</p>",
      "rawMarkdown": "The `test.csv` and `sample_submission.csv` that we can download and see is only 36 rows. It is neither the public test nor the private test. Both the public and private are hidden from us. The approximate size of the test data is 200,000 rows, so you should time how long it takes your model to predict the entire training data of 200,000 rows.",
      "votes": null
    },
    {
      "id": "748830",
      "postDate": "02/18/2020 03:10:59",
      "content": "<p>Thank you Chris. This explains. And thanks for answering my other question <a href=\"https://www.kaggle.com/c/bengaliai-cv19/discussion/131057\">https://www.kaggle.com/c/bengaliai-cv19/discussion/131057</a> caused by the same misunderstanding. </p>",
      "rawMarkdown": "Thank you Chris. This explains. And thanks for answering my other question https://www.kaggle.com/c/bengaliai-cv19/discussion/131057 caused by the same misunderstanding.",
      "votes": null
    },
    {
      "id": "749842",
      "postDate": "02/18/2020 23:51:01",
      "content": "<p>I've had similar problems before. Please see <a href=\"https://www.kaggle.com/c/bengaliai-cv19/discussion/123135\">https://www.kaggle.com/c/bengaliai-cv19/discussion/123135</a></p>",
      "rawMarkdown": "I've had similar problems before. Please see [https://www.kaggle.com/c/bengaliai-cv19/discussion/123135](https://www.kaggle.com/c/bengaliai-cv19/discussion/123135)",
      "votes": null
    },
    {
      "id": "749914",
      "postDate": "02/19/2020 02:10:28",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 748815,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "02/18/2020 02:49:33",
      "content": "<p>The <code>test.csv</code> and <code>sample_submission.csv</code> that we can download and see is only 36 rows. It is neither the public test nor the private test. Both the public and private are hidden from us. The approximate size of the test data is 200,000 rows, so you should time how long it takes your model to predict the entire training data of 200,000 rows.</p>",
      "votes": null,
      "replies": [
        {
          "id": 748830,
          "author_name": "zhangyang",
          "author_url": "",
          "post_date": "02/18/2020 03:10:59",
          "content": "<p>Thank you Chris. This explains. And thanks for answering my other question <a href=\"https://www.kaggle.com/c/bengaliai-cv19/discussion/131057\">https://www.kaggle.com/c/bengaliai-cv19/discussion/131057</a> caused by the same misunderstanding. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 749842,
      "author_name": "zhangeng",
      "author_url": "",
      "post_date": "02/18/2020 23:51:01",
      "content": "<p>I've had similar problems before. Please see <a href=\"https://www.kaggle.com/c/bengaliai-cv19/discussion/123135\">https://www.kaggle.com/c/bengaliai-cv19/discussion/123135</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 749914,
          "author_name": "zhangyang",
          "author_url": "",
          "post_date": "02/19/2020 02:10:28",
          "content": "<p>Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "748812": "I committed an inference Kernel using CPU and it finished successfully in 63 seconds with output submission.csv I can see. When I tried submitting it, the submission timed out after 9 hours. I don't understand the big time difference. Isn't private testset only slightly larger than the public testset (46% vs 54%)?  Thank you for any explanation that helps me understand.",
    "748815": "The `test.csv` and `sample_submission.csv` that we can download and see is only 36 rows. It is neither the public test nor the private test. Both the public and private are hidden from us. The approximate size of the test data is 200,000 rows, so you should time how long it takes your model to predict the entire training data of 200,000 rows.",
    "748830": "Thank you Chris. This explains. And thanks for answering my other question https://www.kaggle.com/c/bengaliai-cv19/discussion/131057 caused by the same misunderstanding.",
    "749842": "I've had similar problems before. Please see [https://www.kaggle.com/c/bengaliai-cv19/discussion/123135](https://www.kaggle.com/c/bengaliai-cv19/discussion/123135)",
    "749914": "Thank you!"
  },
  "source": "meta"
}