{
  "id": 443733,
  "title": "Pypi packages download ",
  "url": "/competitions/bengaliai-speech/discussion/443733",
  "author_name": "",
  "post_date": "2023-09-28T14:18:45.236610900Z",
  "votes": 3,
  "comment_count": 6,
  "views": 0,
  "content": "<p>This has been an issue that troubled me for a long time. To build an inference notebook, we sometimes need to use a lots of packages that kaggle doesn't include. However, the internet is not allowed in the submission, and thus the competitors have to search and download packages from pypi one by one, and even look for some interconnected packages if necessary. </p>\n<p>What are ways that can make our lives easier?</p>",
  "messages": [
    {
      "id": "2460004",
      "postDate": "09/28/2023 14:18:45",
      "content": "<p>This has been an issue that troubled me for a long time. To build an inference notebook, we sometimes need to use a lots of packages that kaggle doesn't include. However, the internet is not allowed in the submission, and thus the competitors have to search and download packages from pypi one by one, and even look for some interconnected packages if necessary. </p>\n<p>What are ways that can make our lives easier?</p>",
      "rawMarkdown": "This has been an issue that troubled me for a long time. To build an inference notebook, we sometimes need to use a lots of packages that kaggle doesn't include. However, the internet is not allowed in the submission, and thus the competitors have to search and download packages from pypi one by one, and even look for some interconnected packages if necessary. \n\nWhat are ways that can make our lives easier?",
      "votes": null
    },
    {
      "id": "2460063",
      "postDate": "09/28/2023 14:53:03",
      "content": "<p>An easier way to do it is by installing whatever packages you need in a notebook with a connection, then freeze and save the installed packages and load them from this notebook in whatever no-connection notebook you need. Check <a href=\"https://www.kaggle.com/code/samuelepino/pip-installing-packages-with-no-internet\" target=\"_blank\">this notebook</a> to get the general idea. </p>",
      "rawMarkdown": "An easier way to do it is by installing whatever packages you need in a notebook with a connection, then freeze and save the installed packages and load them from this notebook in whatever no-connection notebook you need. Check [this notebook](https://www.kaggle.com/code/samuelepino/pip-installing-packages-with-no-internet) to get the general idea.",
      "votes": null
    },
    {
      "id": "2460079",
      "postDate": "09/28/2023 15:05:22",
      "content": "<p>Are there any downsides to this approach, considering the notebook environment is pinned? <a href=\"https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful\" target=\"_blank\">https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful</a></p>",
      "rawMarkdown": "Are there any downsides to this approach, considering the notebook environment is pinned? https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful",
      "votes": null
    },
    {
      "id": "2460081",
      "postDate": "09/28/2023 15:05:47",
      "content": "<p>I think I can answer this question now after a while of surfing:</p>\n<p><code>!pip download &lt;package_name&gt;</code></p>\n<p>This command will download all the packages to <code>\\kaggle\\working</code> instead of installing the packages, alongs with all sorts of dependencies. You may then download it to the local and upload to the kaggle dataset.</p>\n<p>Hope newcomers may find this post helpful</p>",
      "rawMarkdown": "I think I can answer this question now after a while of surfing:\n\n`!pip download <package_name>`\n\nThis command will download all the packages to `\\kaggle\\working` instead of installing the packages, alongs with all sorts of dependencies. You may then download it to the local and upload to the kaggle dataset.\n\nHope newcomers may find this post helpful",
      "votes": null
    },
    {
      "id": "2460098",
      "postDate": "09/28/2023 15:18:13",
      "content": "<p>Oh, it requires too many manual steps.<br>\nUploading from a local machine with a poor internet connection can be slow.</p>",
      "rawMarkdown": "Oh, it requires too many manual steps.\nUploading from a local machine with a poor internet connection can be slow.",
      "votes": null
    },
    {
      "id": "2460114",
      "postDate": "09/28/2023 15:24:26",
      "content": "<p>Unfortunately, it collects ALL packages, not only the necessary ones. But in combination with <a href=\"https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful\" target=\"_blank\">https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful</a> it allows to collect wheel files 😀</p>\n<p>Maybe there is a simpler solution?</p>",
      "rawMarkdown": "Unfortunately, it collects ALL packages, not only the necessary ones. But in combination with https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful it allows to collect wheel files 😀\n\nMaybe there is a simpler solution?",
      "votes": null
    },
    {
      "id": "2460277",
      "postDate": "09/28/2023 17:12:56",
      "content": "<p>There is no local machine involved in the process. You download in one kaggle notebook, and load in a second kaggle notebook.</p>",
      "rawMarkdown": "There is no local machine involved in the process. You download in one kaggle notebook, and load in a second kaggle notebook.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2460063,
      "author_name": "shlomoron",
      "author_url": "",
      "post_date": "09/28/2023 14:53:03",
      "content": "<p>An easier way to do it is by installing whatever packages you need in a notebook with a connection, then freeze and save the installed packages and load them from this notebook in whatever no-connection notebook you need. Check <a href=\"https://www.kaggle.com/code/samuelepino/pip-installing-packages-with-no-internet\" target=\"_blank\">this notebook</a> to get the general idea. </p>",
      "votes": null,
      "replies": [
        {
          "id": 2460098,
          "author_name": "tsobolev",
          "author_url": "",
          "post_date": "09/28/2023 15:18:13",
          "content": "<p>Oh, it requires too many manual steps.<br>\nUploading from a local machine with a poor internet connection can be slow.</p>",
          "votes": null,
          "replies": [
            {
              "id": 2460277,
              "author_name": "shlomoron",
              "author_url": "",
              "post_date": "09/28/2023 17:12:56",
              "content": "<p>There is no local machine involved in the process. You download in one kaggle notebook, and load in a second kaggle notebook.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 2460079,
      "author_name": "tsobolev",
      "author_url": "",
      "post_date": "09/28/2023 15:05:22",
      "content": "<p>Are there any downsides to this approach, considering the notebook environment is pinned? <a href=\"https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful\" target=\"_blank\">https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2460081,
      "author_name": "renyiwei",
      "author_url": "",
      "post_date": "09/28/2023 15:05:47",
      "content": "<p>I think I can answer this question now after a while of surfing:</p>\n<p><code>!pip download &lt;package_name&gt;</code></p>\n<p>This command will download all the packages to <code>\\kaggle\\working</code> instead of installing the packages, alongs with all sorts of dependencies. You may then download it to the local and upload to the kaggle dataset.</p>\n<p>Hope newcomers may find this post helpful</p>",
      "votes": null,
      "replies": [
        {
          "id": 2460114,
          "author_name": "tsobolev",
          "author_url": "",
          "post_date": "09/28/2023 15:24:26",
          "content": "<p>Unfortunately, it collects ALL packages, not only the necessary ones. But in combination with <a href=\"https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful\" target=\"_blank\">https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful</a> it allows to collect wheel files 😀</p>\n<p>Maybe there is a simpler solution?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2460004": "This has been an issue that troubled me for a long time. To build an inference notebook, we sometimes need to use a lots of packages that kaggle doesn't include. However, the internet is not allowed in the submission, and thus the competitors have to search and download packages from pypi one by one, and even look for some interconnected packages if necessary. \n\nWhat are ways that can make our lives easier?",
    "2460063": "An easier way to do it is by installing whatever packages you need in a notebook with a connection, then freeze and save the installed packages and load them from this notebook in whatever no-connection notebook you need. Check [this notebook](https://www.kaggle.com/code/samuelepino/pip-installing-packages-with-no-internet) to get the general idea.",
    "2460079": "Are there any downsides to this approach, considering the notebook environment is pinned? https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful",
    "2460081": "I think I can answer this question now after a while of surfing:\n\n`!pip download <package_name>`\n\nThis command will download all the packages to `\\kaggle\\working` instead of installing the packages, alongs with all sorts of dependencies. You may then download it to the local and upload to the kaggle dataset.\n\nHope newcomers may find this post helpful",
    "2460098": "Oh, it requires too many manual steps.\nUploading from a local machine with a poor internet connection can be slow.",
    "2460114": "Unfortunately, it collects ALL packages, not only the necessary ones. But in combination with https://www.kaggle.com/tsobolev/installing-offline-packages-is-painful it allows to collect wheel files 😀\n\nMaybe there is a simpler solution?",
    "2460277": "There is no local machine involved in the process. You download in one kaggle notebook, and load in a second kaggle notebook."
  },
  "source": "meta"
}