{
  "id": 125775,
  "title": "Are we supposed to create nq_test_tfrecord offline ?",
  "url": "/competitions/tensorflow2-question-answering/discussion/125775",
  "author_name": "",
  "post_date": "2020-01-13T15:07:30.228060300Z",
  "votes": null,
  "comment_count": 9,
  "views": 0,
  "content": "<p>Hi,</p>\n\n<p>I have a small doubt. Are we supposed to create nq_test_tfrecord offline and upload as part of input data or are we supposed to create it inside the kernel within the time limit?</p>",
  "messages": [
    {
      "id": "717752",
      "postDate": "01/13/2020 15:07:30",
      "content": "<p>Hi,</p>\n\n<p>I have a small doubt. Are we supposed to create nq_test_tfrecord offline and upload as part of input data or are we supposed to create it inside the kernel within the time limit?</p>",
      "rawMarkdown": "Hi,\n\nI have a small doubt. Are we supposed to create nq_test_tfrecord offline and upload as part of input data or are we supposed to create it inside the kernel within the time limit?",
      "votes": null
    },
    {
      "id": "717757",
      "postDate": "01/13/2020 15:21:53",
      "content": "<p>The private test set is accessible only from the kernel (after submitting), so you'd have to create it inside the kernel within the time limit.</p>",
      "rawMarkdown": "The private test set is accessible only from the kernel (after submitting), so you'd have to create it inside the kernel within the time limit.",
      "votes": null
    },
    {
      "id": "717761",
      "postDate": "01/13/2020 15:35:34",
      "content": "<p>One more thing, so filename has to be <code>simplified-nq-test.jsonl</code>, which will be automatically replaced by private dataset right.</p>",
      "rawMarkdown": "One more thing, so filename has to be ```simplified-nq-test.jsonl```, which will be automatically replaced by private dataset right.",
      "votes": null
    },
    {
      "id": "717770",
      "postDate": "01/13/2020 15:47:58",
      "content": "<p>Yes.</p>",
      "rawMarkdown": "Yes.",
      "votes": null
    },
    {
      "id": "717773",
      "postDate": "01/13/2020 15:49:45",
      "content": "<p>Thank you. Much appreciated.</p>",
      "rawMarkdown": "Thank you. Much appreciated.",
      "votes": null
    },
    {
      "id": "718287",
      "postDate": "01/14/2020 08:51:36",
      "content": "<p>How about training offline models?</p>",
      "rawMarkdown": "How about training offline models?",
      "votes": null
    },
    {
      "id": "718385",
      "postDate": "01/14/2020 10:47:47",
      "content": "<p>You are free to train offline models and submit only inference code with weights uploaded.</p>",
      "rawMarkdown": "You are free to train offline models and submit only inference code with weights uploaded.",
      "votes": null
    },
    {
      "id": "718387",
      "postDate": "01/14/2020 10:51:31",
      "content": "<p>Are we supposed to create nq_test_tfrecord offline ?</p>",
      "rawMarkdown": "Are we supposed to create nq_test_tfrecord offline ?",
      "votes": null
    },
    {
      "id": "718392",
      "postDate": "01/14/2020 10:57:53",
      "content": "<p>It is already has been answered above - you have to create it in your kernel code dynamically. </p>",
      "rawMarkdown": "It is already has been answered above - you have to create it in your kernel code dynamically.",
      "votes": null
    },
    {
      "id": "718397",
      "postDate": "01/14/2020 11:03:44",
      "content": "<p>I have asked. For offline models, are we supposed to create our own tf records?</p>",
      "rawMarkdown": "I have asked. For offline models, are we supposed to create our own tf records?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 717757,
      "author_name": "rohanrao",
      "author_url": "",
      "post_date": "01/13/2020 15:21:53",
      "content": "<p>The private test set is accessible only from the kernel (after submitting), so you'd have to create it inside the kernel within the time limit.</p>",
      "votes": null,
      "replies": [
        {
          "id": 717761,
          "author_name": "s4sarath",
          "author_url": "",
          "post_date": "01/13/2020 15:35:34",
          "content": "<p>One more thing, so filename has to be <code>simplified-nq-test.jsonl</code>, which will be automatically replaced by private dataset right.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 717770,
          "author_name": "rohanrao",
          "author_url": "",
          "post_date": "01/13/2020 15:47:58",
          "content": "<p>Yes.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 717773,
          "author_name": "s4sarath",
          "author_url": "",
          "post_date": "01/13/2020 15:49:45",
          "content": "<p>Thank you. Much appreciated.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 718287,
          "author_name": "zzj040203",
          "author_url": "",
          "post_date": "01/14/2020 08:51:36",
          "content": "<p>How about training offline models?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 718385,
          "author_name": "yaroshevskiy",
          "author_url": "",
          "post_date": "01/14/2020 10:47:47",
          "content": "<p>You are free to train offline models and submit only inference code with weights uploaded.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 718387,
          "author_name": "zzj040203",
          "author_url": "",
          "post_date": "01/14/2020 10:51:31",
          "content": "<p>Are we supposed to create nq_test_tfrecord offline ?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 718392,
          "author_name": "yaroshevskiy",
          "author_url": "",
          "post_date": "01/14/2020 10:57:53",
          "content": "<p>It is already has been answered above - you have to create it in your kernel code dynamically. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 718397,
          "author_name": "zzj040203",
          "author_url": "",
          "post_date": "01/14/2020 11:03:44",
          "content": "<p>I have asked. For offline models, are we supposed to create our own tf records?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "717752": "Hi,\n\nI have a small doubt. Are we supposed to create nq_test_tfrecord offline and upload as part of input data or are we supposed to create it inside the kernel within the time limit?",
    "717757": "The private test set is accessible only from the kernel (after submitting), so you'd have to create it inside the kernel within the time limit.",
    "717761": "One more thing, so filename has to be ```simplified-nq-test.jsonl```, which will be automatically replaced by private dataset right.",
    "717770": "Yes.",
    "717773": "Thank you. Much appreciated.",
    "718287": "How about training offline models?",
    "718385": "You are free to train offline models and submit only inference code with weights uploaded.",
    "718387": "Are we supposed to create nq_test_tfrecord offline ?",
    "718392": "It is already has been answered above - you have to create it in your kernel code dynamically.",
    "718397": "I have asked. For offline models, are we supposed to create our own tf records?"
  },
  "source": "meta"
}