{
  "id": 205282,
  "title": "What should be the inference time ?",
  "url": "/competitions/riiid-test-answer-prediction/discussion/205282",
  "author_name": "",
  "post_date": "2020-12-19T10:30:02.339859300Z",
  "votes": 1,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hello all, </p>\n<p>What should be the maximum inference time to successfully submit within 9 hours limit ?</p>\n<p>Mine is 58 seconds . Will it be completed in time alloted ?<br>\nHere is the screenshot.<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4947407%2F40ae3f63fb9a827cd4186d6297363e9f%2FScreenshot%20(1).png?generation=1608373764943380&amp;alt=media\" alt=\"\"></p>\n<p>Thanks </p>",
  "messages": [
    {
      "id": "1118717",
      "postDate": "12/19/2020 10:30:02",
      "content": "<p>Hello all, </p>\n<p>What should be the maximum inference time to successfully submit within 9 hours limit ?</p>\n<p>Mine is 58 seconds . Will it be completed in time alloted ?<br>\nHere is the screenshot.<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4947407%2F40ae3f63fb9a827cd4186d6297363e9f%2FScreenshot%20(1).png?generation=1608373764943380&amp;alt=media\" alt=\"\"></p>\n<p>Thanks </p>",
      "rawMarkdown": "Hello all, \n\nWhat should be the maximum inference time to successfully submit within 9 hours limit ?\n\nMine is 58 seconds . Will it be completed in time alloted ?\nHere is the screenshot.\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4947407%2F40ae3f63fb9a827cd4186d6297363e9f%2FScreenshot%20(1).png?generation=1608373764943380&alt=media)\n\n Thanks",
      "votes": null
    },
    {
      "id": "1118832",
      "postDate": "12/19/2020 12:41:02",
      "content": "<p>Example test has 108 rows while the real test has 2.5m. There are certain overheads for small dataset but in general I found that <code>t * 7000</code> is a pretty good ballpark estimate. </p>\n<p>For the final model ensembles, using tito's <code>iter_env</code> simulation is the way to go.</p>",
      "rawMarkdown": "Example test has 108 rows while the real test has 2.5m. There are certain overheads for small dataset but in general I found that `t * 7000` is a pretty good ballpark estimate. \n\nFor the final model ensembles, using tito's `iter_env` simulation is the way to go.",
      "votes": null
    },
    {
      "id": "1118845",
      "postDate": "12/19/2020 12:55:40",
      "content": "<p><a href=\"https://www.kaggle.com/scaomath\" target=\"_blank\">@scaomath</a> what is t * 7000 ?</p>",
      "rawMarkdown": "scaomath what is t * 7000 ?",
      "votes": null
    },
    {
      "id": "1118901",
      "postDate": "12/19/2020 14:07:11",
      "content": "<p><code>t</code> stands for the inference time taken for the example test file.</p>",
      "rawMarkdown": "`t` stands for the inference time taken for the example test file.",
      "votes": null
    },
    {
      "id": "1119833",
      "postDate": "12/20/2020 11:46:55",
      "content": "<p>I use looping with dict to generate predictor and it takes me around 2-3 hours for submission. My runtime for the submission example is around 0.15-0.25 seconds</p>",
      "rawMarkdown": "I use looping with dict to generate predictor and it takes me around 2-3 hours for submission. My runtime for the submission example is around 0.15-0.25 seconds",
      "votes": null
    },
    {
      "id": "1120000",
      "postDate": "12/20/2020 14:38:31",
      "content": "<p>thanks, any way mine was not successfully submitted, so i did little changes in my inference and it works now. </p>",
      "rawMarkdown": "thanks, any way mine was not successfully submitted, so i did little changes in my inference and it works now.",
      "votes": null
    },
    {
      "id": "1120001",
      "postDate": "12/20/2020 14:39:09",
      "content": "<p>yes now mine is also .5 seconds which takes around 3.5 hrs</p>",
      "rawMarkdown": "yes now mine is also .5 seconds which takes around 3.5 hrs",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1118832,
      "author_name": "scaomath",
      "author_url": "",
      "post_date": "12/19/2020 12:41:02",
      "content": "<p>Example test has 108 rows while the real test has 2.5m. There are certain overheads for small dataset but in general I found that <code>t * 7000</code> is a pretty good ballpark estimate. </p>\n<p>For the final model ensembles, using tito's <code>iter_env</code> simulation is the way to go.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1118845,
          "author_name": "ptrikp",
          "author_url": "",
          "post_date": "12/19/2020 12:55:40",
          "content": "<p><a href=\"https://www.kaggle.com/scaomath\" target=\"_blank\">@scaomath</a> what is t * 7000 ?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1118901,
          "author_name": "scaomath",
          "author_url": "",
          "post_date": "12/19/2020 14:07:11",
          "content": "<p><code>t</code> stands for the inference time taken for the example test file.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1120000,
          "author_name": "ptrikp",
          "author_url": "",
          "post_date": "12/20/2020 14:38:31",
          "content": "<p>thanks, any way mine was not successfully submitted, so i did little changes in my inference and it works now. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1119833,
      "author_name": "marcellosusanto",
      "author_url": "",
      "post_date": "12/20/2020 11:46:55",
      "content": "<p>I use looping with dict to generate predictor and it takes me around 2-3 hours for submission. My runtime for the submission example is around 0.15-0.25 seconds</p>",
      "votes": null,
      "replies": [
        {
          "id": 1120001,
          "author_name": "ptrikp",
          "author_url": "",
          "post_date": "12/20/2020 14:39:09",
          "content": "<p>yes now mine is also .5 seconds which takes around 3.5 hrs</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1118717": "Hello all, \n\nWhat should be the maximum inference time to successfully submit within 9 hours limit ?\n\nMine is 58 seconds . Will it be completed in time alloted ?\nHere is the screenshot.\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4947407%2F40ae3f63fb9a827cd4186d6297363e9f%2FScreenshot%20(1).png?generation=1608373764943380&alt=media)\n\n Thanks",
    "1118832": "Example test has 108 rows while the real test has 2.5m. There are certain overheads for small dataset but in general I found that `t * 7000` is a pretty good ballpark estimate. \n\nFor the final model ensembles, using tito's `iter_env` simulation is the way to go.",
    "1118845": "scaomath what is t * 7000 ?",
    "1118901": "`t` stands for the inference time taken for the example test file.",
    "1119833": "I use looping with dict to generate predictor and it takes me around 2-3 hours for submission. My runtime for the submission example is around 0.15-0.25 seconds",
    "1120000": "thanks, any way mine was not successfully submitted, so i did little changes in my inference and it works now.",
    "1120001": "yes now mine is also .5 seconds which takes around 3.5 hrs"
  },
  "source": "meta"
}