{
  "id": 293796,
  "title": "Mobile Data Usage While Training Models ?",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/293796",
  "author_name": "",
  "post_date": "2021-12-07T03:37:58.740463500Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Last time I trained a yolov4 model on colab. I am using mobile data for my work, which is limited. Here is the concern while training it, it consumed a lot of mobile data. Otherwise normally loading a kaggle notebook is also sometimes consuming a lot of data (If I run it). Can someone explain why is it so? Is it due to loading of weights ?</p>",
  "messages": [
    {
      "id": "1610218",
      "postDate": "12/07/2021 03:37:58",
      "content": "<p>Last time I trained a yolov4 model on colab. I am using mobile data for my work, which is limited. Here is the concern while training it, it consumed a lot of mobile data. Otherwise normally loading a kaggle notebook is also sometimes consuming a lot of data (If I run it). Can someone explain why is it so? Is it due to loading of weights ?</p>",
      "rawMarkdown": "Last time I trained a yolov4 model on colab. I am using mobile data for my work, which is limited. Here is the concern while training it, it consumed a lot of mobile data. Otherwise normally loading a kaggle notebook is also sometimes consuming a lot of data (If I run it). Can someone explain why is it so? Is it due to loading of weights ?",
      "votes": null
    },
    {
      "id": "1610252",
      "postDate": "12/07/2021 04:38:55",
      "content": "<p>Moreover If I can ask what does internet button on kaggle notebook does ? Thanks for replying.</p>",
      "rawMarkdown": "Moreover If I can ask what does internet button on kaggle notebook does ? Thanks for replying.",
      "votes": null
    },
    {
      "id": "1610591",
      "postDate": "12/07/2021 10:38:31",
      "content": "<p>Internet button in notebook editor enables/disables internet access from your kernel.<br>\nNote, that you cannot use the internet while submitting your work (it prevents cheating and downloading hidden test set.</p>",
      "rawMarkdown": "Internet button in notebook editor enables/disables internet access from your kernel.\nNote, that you cannot use the internet while submitting your work (it prevents cheating and downloading hidden test set.",
      "votes": null
    },
    {
      "id": "1610594",
      "postDate": "12/07/2021 10:41:50",
      "content": "<blockquote>\n  <p>loading a kaggle notebook is also sometimes consuming a lot of data (If I run it)</p>\n</blockquote>\n<p>I don't know what is going on while you running your notebook, but I guess that you use some pretrained model. For example, you use a pretrained model from TF hub. So, calling <code>hub.KerasLayer('./link/to/model')</code> will download model weights.<br>\nOne more implicit data usage example - <code>tf.keras.application.&lt;SomeModel&gt;</code>. By Default Keras application use <code>weights='imegenet'</code> argument. This will download weights as well</p>",
      "rawMarkdown": "> loading a kaggle notebook is also sometimes consuming a lot of data (If I run it)\n\nI don't know what is going on while you running your notebook, but I guess that you use some pretrained model. For example, you use a pretrained model from TF hub. So, calling `hub.KerasLayer('./link/to/model')` will download model weights.\nOne more implicit data usage example - `tf.keras.application.<SomeModel>`. By Default Keras application use `weights='imegenet'` argument. This will download weights as well",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1610252,
      "author_name": "mrfoomaster",
      "author_url": "",
      "post_date": "12/07/2021 04:38:55",
      "content": "<p>Moreover If I can ask what does internet button on kaggle notebook does ? Thanks for replying.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1610591,
          "author_name": "meowmeowmeowmeowmeow",
          "author_url": "",
          "post_date": "12/07/2021 10:38:31",
          "content": "<p>Internet button in notebook editor enables/disables internet access from your kernel.<br>\nNote, that you cannot use the internet while submitting your work (it prevents cheating and downloading hidden test set.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1610594,
      "author_name": "meowmeowmeowmeowmeow",
      "author_url": "",
      "post_date": "12/07/2021 10:41:50",
      "content": "<blockquote>\n  <p>loading a kaggle notebook is also sometimes consuming a lot of data (If I run it)</p>\n</blockquote>\n<p>I don't know what is going on while you running your notebook, but I guess that you use some pretrained model. For example, you use a pretrained model from TF hub. So, calling <code>hub.KerasLayer('./link/to/model')</code> will download model weights.<br>\nOne more implicit data usage example - <code>tf.keras.application.&lt;SomeModel&gt;</code>. By Default Keras application use <code>weights='imegenet'</code> argument. This will download weights as well</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1610218": "Last time I trained a yolov4 model on colab. I am using mobile data for my work, which is limited. Here is the concern while training it, it consumed a lot of mobile data. Otherwise normally loading a kaggle notebook is also sometimes consuming a lot of data (If I run it). Can someone explain why is it so? Is it due to loading of weights ?",
    "1610252": "Moreover If I can ask what does internet button on kaggle notebook does ? Thanks for replying.",
    "1610591": "Internet button in notebook editor enables/disables internet access from your kernel.\nNote, that you cannot use the internet while submitting your work (it prevents cheating and downloading hidden test set.",
    "1610594": "> loading a kaggle notebook is also sometimes consuming a lot of data (If I run it)\n\nI don't know what is going on while you running your notebook, but I guess that you use some pretrained model. For example, you use a pretrained model from TF hub. So, calling `hub.KerasLayer('./link/to/model')` will download model weights.\nOne more implicit data usage example - `tf.keras.application.<SomeModel>`. By Default Keras application use `weights='imegenet'` argument. This will download weights as well"
  },
  "source": "meta"
}