{
  "id": 123424,
  "title": "Usage of older versions of tensorflow (1.14) without turning on the internet",
  "url": "/competitions/deepfake-detection-challenge/discussion/123424",
  "author_name": "",
  "post_date": "2019-12-27T15:07:39.251128Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi all, </p>\n\n<p>I would like to know if there is a possibility of using an older version of TensorFlow (in my case tf=1.14). I have pre-trained models and codes written on TF 1.14. I need this because i am using  \"slim = tf.contrib.slim\". And tf.contrib is not available in TF 2.X. So, Is there a possible way of overcoming this issue since I need to use it in my submission code. </p>\n\n<p>I could simply install the older version, but due to competition rules, I cannot Turn on the internet. </p>\n\n<p>Thanks</p>",
  "messages": [
    {
      "id": "704513",
      "postDate": "12/27/2019 15:07:39",
      "content": "<p>Hi all, </p>\n\n<p>I would like to know if there is a possibility of using an older version of TensorFlow (in my case tf=1.14). I have pre-trained models and codes written on TF 1.14. I need this because i am using  \"slim = tf.contrib.slim\". And tf.contrib is not available in TF 2.X. So, Is there a possible way of overcoming this issue since I need to use it in my submission code. </p>\n\n<p>I could simply install the older version, but due to competition rules, I cannot Turn on the internet. </p>\n\n<p>Thanks</p>",
      "rawMarkdown": "Hi all, \n\nI would like to know if there is a possibility of using an older version of TensorFlow (in my case tf=1.14). I have pre-trained models and codes written on TF 1.14. I need this because i am using  \"slim = tf.contrib.slim\". And tf.contrib is not available in TF 2.X. So, Is there a possible way of overcoming this issue since I need to use it in my submission code. \n\nI could simply install the older version, but due to competition rules, I cannot Turn on the internet. \n\nThanks",
      "votes": null
    },
    {
      "id": "705348",
      "postDate": "12/28/2019 19:58:02",
      "content": "<p>You can install it from external dataset. Just replace the package name by the path to the wheel file. I uploaded these <a href=\"https://www.kaggle.com/unkownhihi/tensorflow-estimator\">tensorflow estimator</a>, <a href=\"https://www.kaggle.com/unkownhihi/tensorboard\">tensorboard</a>, and <a href=\"https://www.kaggle.com/unkownhihi/tensorflow\">tensorflow</a>. I know you just want tensorflow but pip will try to fetch the other packages while installing tensorflow. Be sure to install tensorboard and tensorflow estimator before the tensorflow.</p>",
      "rawMarkdown": "You can install it from external dataset. Just replace the package name by the path to the wheel file. I uploaded these [tensorflow estimator](https://www.kaggle.com/unkownhihi/tensorflow-estimator), [tensorboard](https://www.kaggle.com/unkownhihi/tensorboard), and [tensorflow](https://www.kaggle.com/unkownhihi/tensorflow). I know you just want tensorflow but pip will try to fetch the other packages while installing tensorflow. Be sure to install tensorboard and tensorflow estimator before the tensorflow.",
      "votes": null
    },
    {
      "id": "706341",
      "postDate": "12/30/2019 09:05:18",
      "content": "<p>I had a similar issue in another competition and this <a href=\"https://www.kaggle.com/c/tensorflow2-question-answering/discussion/119239\">discussion topic</a> and <a href=\"https://www.kaggle.com/kashnitsky/offline-tensorflow-1-13-1-and-bert-tensorflow\">notebook</a> solved it.</p>",
      "rawMarkdown": "I had a similar issue in another competition and this [discussion topic](https://www.kaggle.com/c/tensorflow2-question-answering/discussion/119239) and [notebook](https://www.kaggle.com/kashnitsky/offline-tensorflow-1-13-1-and-bert-tensorflow) solved it.",
      "votes": null
    },
    {
      "id": "706635",
      "postDate": "12/30/2019 16:37:49",
      "content": "<p>I have uploaded tensorflow estimator, tensorboard, and tensorflow in <a href=\"https://www.kaggle.com/chenshen03/packages\">packages</a>. You can install them by pip.</p>",
      "rawMarkdown": "I have uploaded tensorflow estimator, tensorboard, and tensorflow in [packages](https://www.kaggle.com/chenshen03/packages). You can install them by pip.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 705348,
      "author_name": "unkownhihi",
      "author_url": "",
      "post_date": "12/28/2019 19:58:02",
      "content": "<p>You can install it from external dataset. Just replace the package name by the path to the wheel file. I uploaded these <a href=\"https://www.kaggle.com/unkownhihi/tensorflow-estimator\">tensorflow estimator</a>, <a href=\"https://www.kaggle.com/unkownhihi/tensorboard\">tensorboard</a>, and <a href=\"https://www.kaggle.com/unkownhihi/tensorflow\">tensorflow</a>. I know you just want tensorflow but pip will try to fetch the other packages while installing tensorflow. Be sure to install tensorboard and tensorflow estimator before the tensorflow.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 706341,
      "author_name": "msheriey",
      "author_url": "",
      "post_date": "12/30/2019 09:05:18",
      "content": "<p>I had a similar issue in another competition and this <a href=\"https://www.kaggle.com/c/tensorflow2-question-answering/discussion/119239\">discussion topic</a> and <a href=\"https://www.kaggle.com/kashnitsky/offline-tensorflow-1-13-1-and-bert-tensorflow\">notebook</a> solved it.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 706635,
      "author_name": "chenshen03",
      "author_url": "",
      "post_date": "12/30/2019 16:37:49",
      "content": "<p>I have uploaded tensorflow estimator, tensorboard, and tensorflow in <a href=\"https://www.kaggle.com/chenshen03/packages\">packages</a>. You can install them by pip.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "704513": "Hi all, \n\nI would like to know if there is a possibility of using an older version of TensorFlow (in my case tf=1.14). I have pre-trained models and codes written on TF 1.14. I need this because i am using  \"slim = tf.contrib.slim\". And tf.contrib is not available in TF 2.X. So, Is there a possible way of overcoming this issue since I need to use it in my submission code. \n\nI could simply install the older version, but due to competition rules, I cannot Turn on the internet. \n\nThanks",
    "705348": "You can install it from external dataset. Just replace the package name by the path to the wheel file. I uploaded these [tensorflow estimator](https://www.kaggle.com/unkownhihi/tensorflow-estimator), [tensorboard](https://www.kaggle.com/unkownhihi/tensorboard), and [tensorflow](https://www.kaggle.com/unkownhihi/tensorflow). I know you just want tensorflow but pip will try to fetch the other packages while installing tensorflow. Be sure to install tensorboard and tensorflow estimator before the tensorflow.",
    "706341": "I had a similar issue in another competition and this [discussion topic](https://www.kaggle.com/c/tensorflow2-question-answering/discussion/119239) and [notebook](https://www.kaggle.com/kashnitsky/offline-tensorflow-1-13-1-and-bert-tensorflow) solved it.",
    "706635": "I have uploaded tensorflow estimator, tensorboard, and tensorflow in [packages](https://www.kaggle.com/chenshen03/packages). You can install them by pip."
  },
  "source": "meta"
}