{
  "id": 132937,
  "title": "Tensorflow version",
  "url": "/competitions/bengaliai-cv19/discussion/132937",
  "author_name": "",
  "post_date": "2020-02-28T20:51:19.031282900Z",
  "votes": 1,
  "comment_count": 8,
  "views": 0,
  "content": "<p>Guys! Would anybody explain me - in some kernels I got tf version 2.1, but when I build model for older kernels I got version 1. Why is that? I know in colab I can switch easily between versions, but how do this in kaggle?\nThank you. </p>",
  "messages": [
    {
      "id": "759280",
      "postDate": "02/28/2020 20:51:19",
      "content": "<p>Guys! Would anybody explain me - in some kernels I got tf version 2.1, but when I build model for older kernels I got version 1. Why is that? I know in colab I can switch easily between versions, but how do this in kaggle?\nThank you. </p>",
      "rawMarkdown": "Guys! Would anybody explain me - in some kernels I got tf version 2.1, but when I build model for older kernels I got version 1. Why is that? I know in colab I can switch easily between versions, but how do this in kaggle?\nThank you.",
      "votes": null
    },
    {
      "id": "759293",
      "postDate": "02/28/2020 21:24:07",
      "content": "<p>Is pip install tensorflow for other version in the older kernel/s working?</p>",
      "rawMarkdown": "Is pip install tensorflow for other version in the older kernel/s working?",
      "votes": null
    },
    {
      "id": "759305",
      "postDate": "02/28/2020 21:41:11",
      "content": "<p>pip install tensorflow==x.x.x, e.g. pip install tensorflow==2.2.0</p>",
      "rawMarkdown": "pip install tensorflow==x.x.x, e.g. pip install tensorflow==2.2.0",
      "votes": null
    },
    {
      "id": "760927",
      "postDate": "03/01/2020 23:07:07",
      "content": "<p>I don't know? I've tried reinstall for old competitions, but it doesn't work OK</p>",
      "rawMarkdown": "I don't know? I've tried reinstall for old competitions, but it doesn't work OK",
      "votes": null
    },
    {
      "id": "760928",
      "postDate": "03/01/2020 23:10:22",
      "content": "<p>I don't know. Just install doesn't work OK. I've read advice to reboot kernel after reinstallation. And sometime it works, smtms still return to old version. I don't understand how it suppose to work</p>",
      "rawMarkdown": "I don't know. Just install doesn't work OK. I've read advice to reboot kernel after reinstallation. And sometime it works, smtms still return to old version. I don't understand how it suppose to work",
      "votes": null
    },
    {
      "id": "760941",
      "postDate": "03/01/2020 23:50:37",
      "content": "<p>Did you turn on internet?  Works fine for me with internet on, but keep in mind it disappears when you power off, and the competition doesn't allow submissions with internet on.</p>",
      "rawMarkdown": "Did you turn on internet?  Works fine for me with internet on, but keep in mind it disappears when you power off, and the competition doesn't allow submissions with internet on.",
      "votes": null
    },
    {
      "id": "761026",
      "postDate": "03/02/2020 03:41:21",
      "content": "<p>Notebooks remember which TensorFlow you were using when you first created the notebook. Old notebook will use \"Original\" docker with <code>tf 1.14</code> or you can switch to \"Latest Available\" and get <code>tf 2.1</code>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1723677%2F518a2dd855d21bb189437d8c07ae3685%2FScreen%20Shot%202020-03-01%20at%207.39.39%20PM.png?generation=1583120467683142&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "Notebooks remember which TensorFlow you were using when you first created the notebook. Old notebook will use \"Original\" docker with `tf 1.14` or you can switch to \"Latest Available\" and get `tf 2.1`\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1723677%2F518a2dd855d21bb189437d8c07ae3685%2FScreen%20Shot%202020-03-01%20at%207.39.39%20PM.png?generation=1583120467683142&amp;alt=media)",
      "votes": null
    },
    {
      "id": "761489",
      "postDate": "03/02/2020 15:26:37",
      "content": "<p>Oh! Thank u - didn't know that!</p>",
      "rawMarkdown": "Oh! Thank u - didn't know that!",
      "votes": null
    },
    {
      "id": "821515",
      "postDate": "04/26/2020 08:14:49",
      "content": "<h2>IMPORT OLD VERSION OF TENSORFLOW</h2>\n\n<p>Here is the solution to import old version of Tensorflow.</p>\n\n<p><a href=\"https://www.kaggle.com/deshmane/tensorflow-old-version\">https://www.kaggle.com/deshmane/tensorflow-old-version</a></p>\n\n<h3>NOTE</h3>\n\n<p>Make sure you dont import tensorflow before installing the old version, otherwise it keeps the previous packages and point to previous version and you wont see the installed version.</p>",
      "rawMarkdown": "## IMPORT OLD VERSION OF TENSORFLOW\nHere is the solution to import old version of Tensorflow.\n\nhttps://www.kaggle.com/deshmane/tensorflow-old-version\n\n### NOTE\nMake sure you dont import tensorflow before installing the old version, otherwise it keeps the previous packages and point to previous version and you wont see the installed version.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 759293,
      "author_name": "binaicrai",
      "author_url": "",
      "post_date": "02/28/2020 21:24:07",
      "content": "<p>Is pip install tensorflow for other version in the older kernel/s working?</p>",
      "votes": null,
      "replies": [
        {
          "id": 760928,
          "author_name": "serg132003",
          "author_url": "",
          "post_date": "03/01/2020 23:10:22",
          "content": "<p>I don't know. Just install doesn't work OK. I've read advice to reboot kernel after reinstallation. And sometime it works, smtms still return to old version. I don't understand how it suppose to work</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 759305,
      "author_name": "ljschuster",
      "author_url": "",
      "post_date": "02/28/2020 21:41:11",
      "content": "<p>pip install tensorflow==x.x.x, e.g. pip install tensorflow==2.2.0</p>",
      "votes": null,
      "replies": [
        {
          "id": 760927,
          "author_name": "serg132003",
          "author_url": "",
          "post_date": "03/01/2020 23:07:07",
          "content": "<p>I don't know? I've tried reinstall for old competitions, but it doesn't work OK</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 760941,
          "author_name": "ljschuster",
          "author_url": "",
          "post_date": "03/01/2020 23:50:37",
          "content": "<p>Did you turn on internet?  Works fine for me with internet on, but keep in mind it disappears when you power off, and the competition doesn't allow submissions with internet on.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 761026,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "03/02/2020 03:41:21",
      "content": "<p>Notebooks remember which TensorFlow you were using when you first created the notebook. Old notebook will use \"Original\" docker with <code>tf 1.14</code> or you can switch to \"Latest Available\" and get <code>tf 2.1</code>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1723677%2F518a2dd855d21bb189437d8c07ae3685%2FScreen%20Shot%202020-03-01%20at%207.39.39%20PM.png?generation=1583120467683142&amp;alt=media\" alt=\"\"></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 761489,
      "author_name": "serg132003",
      "author_url": "",
      "post_date": "03/02/2020 15:26:37",
      "content": "<p>Oh! Thank u - didn't know that!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 821515,
      "author_name": "deshmane",
      "author_url": "",
      "post_date": "04/26/2020 08:14:49",
      "content": "<h2>IMPORT OLD VERSION OF TENSORFLOW</h2>\n\n<p>Here is the solution to import old version of Tensorflow.</p>\n\n<p><a href=\"https://www.kaggle.com/deshmane/tensorflow-old-version\">https://www.kaggle.com/deshmane/tensorflow-old-version</a></p>\n\n<h3>NOTE</h3>\n\n<p>Make sure you dont import tensorflow before installing the old version, otherwise it keeps the previous packages and point to previous version and you wont see the installed version.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "759280": "Guys! Would anybody explain me - in some kernels I got tf version 2.1, but when I build model for older kernels I got version 1. Why is that? I know in colab I can switch easily between versions, but how do this in kaggle?\nThank you.",
    "759293": "Is pip install tensorflow for other version in the older kernel/s working?",
    "759305": "pip install tensorflow==x.x.x, e.g. pip install tensorflow==2.2.0",
    "760927": "I don't know? I've tried reinstall for old competitions, but it doesn't work OK",
    "760928": "I don't know. Just install doesn't work OK. I've read advice to reboot kernel after reinstallation. And sometime it works, smtms still return to old version. I don't understand how it suppose to work",
    "760941": "Did you turn on internet?  Works fine for me with internet on, but keep in mind it disappears when you power off, and the competition doesn't allow submissions with internet on.",
    "761026": "Notebooks remember which TensorFlow you were using when you first created the notebook. Old notebook will use \"Original\" docker with `tf 1.14` or you can switch to \"Latest Available\" and get `tf 2.1`\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1723677%2F518a2dd855d21bb189437d8c07ae3685%2FScreen%20Shot%202020-03-01%20at%207.39.39%20PM.png?generation=1583120467683142&amp;alt=media)",
    "761489": "Oh! Thank u - didn't know that!",
    "821515": "## IMPORT OLD VERSION OF TENSORFLOW\nHere is the solution to import old version of Tensorflow.\n\nhttps://www.kaggle.com/deshmane/tensorflow-old-version\n\n### NOTE\nMake sure you dont import tensorflow before installing the old version, otherwise it keeps the previous packages and point to previous version and you wont see the installed version."
  },
  "source": "meta"
}