{
  "id": 433837,
  "title": "Colab Pro Memory Issue",
  "url": "/competitions/asl-fingerspelling/discussion/433837",
  "author_name": "C R Suthikshn Kumar",
  "post_date": "2023-08-23T04:07:13.642000",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>As I ran out of Kaggle GPU limit, I tried moving my Competition Notebook  and data to Colab Pro.<br>\nWhile uploading the training and test data, I observe it is not supporting 189 GB. Half way through,<br>\nit has run out of memory. Any advise on using Colab Pro for  ASLFR competition.</p>",
  "messages": [
    {
      "id": 2404108,
      "postDate": "2023-08-23T04:07:13.643Z",
      "content": "<p>As I ran out of Kaggle GPU limit, I tried moving my Competition Notebook  and data to Colab Pro.<br>\nWhile uploading the training and test data, I observe it is not supporting 189 GB. Half way through,<br>\nit has run out of memory. Any advise on using Colab Pro for  ASLFR competition.</p>",
      "rawMarkdown": "As I ran out of Kaggle GPU limit, I tried moving my Competition Notebook  and data to Colab Pro.\nWhile uploading the training and test data, I observe it is not supporting 189 GB. Half way through,\nit has run out of memory. Any advise on using Colab Pro for  ASLFR competition.",
      "votes": 1
    },
    {
      "id": 2410536,
      "postDate": "2023-08-27T04:11:10.030Z",
      "content": "<p>Did you try copying the files to Google Drive, mounting the folder, and using it? Below is an example - just set your paths.</p>\n<p>from google.colab import drive<br>\ndrive.mount('/content/drive') # Mount folder<br>\npath = '/content/drive/Folder/file' # Get path to necessary folder</p>",
      "rawMarkdown": "Did you try copying the files to Google Drive, mounting the folder, and using it? Below is an example - just set your paths.\n\nfrom google.colab import drive\ndrive.mount('/content/drive') # Mount folder\npath = '/content/drive/Folder/file' # Get path to necessary folder\n"
    },
    {
      "id": 2404128,
      "postDate": "2023-08-23T04:33:52.550Z",
      "content": "<p>i think you are not training with all the face landmark. you are using using a subset of the landmark. most people maybe are using less than 100 landmark points, and this is what you need to upload for training</p>",
      "rawMarkdown": "i think you are not training with all the face landmark. you are using using a subset of the landmark. most people maybe are using less than 100 landmark points, and this is what you need to upload for training",
      "replies": [
        {
          "id": 2404301,
          "postDate": "2023-08-23T07:09:42.410Z",
          "content": "<p>Thanks for your response. <br>\nThrough some kaggle postings which provided steps for migration to colab Pro, I got the following code to run in <br>\nmy colab notebook:</p>\n<p>! kaggle competitions download -c asl-fingerspelling</p>\n<p>! mkdir kaggle<br>\n!unzip asl-fingerspelling + \".zip\" -d kaggle</p>\n<h1>Unmount your Google Drive</h1>\n<p>drive.flush_and_unmount()<br>\nCan you advise, update to this code to download specific files. </p>",
          "rawMarkdown": "Thanks for your response. \nThrough some kaggle postings which provided steps for migration to colab Pro, I got the following code to run in \nmy colab notebook:\n\n! kaggle competitions download -c asl-fingerspelling\n\n! mkdir kaggle\n!unzip asl-fingerspelling + \".zip\" -d kaggle\n\n# Unmount your Google Drive\ndrive.flush_and_unmount()\nCan you advise, update to this code to download specific files. "
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2410536,
      "author_name": "Dmitriy Gakh",
      "author_url": "",
      "post_date": "2023-08-27T04:11:10.030000",
      "content": "<p>Did you try copying the files to Google Drive, mounting the folder, and using it? Below is an example - just set your paths.</p>\n<p>from google.colab import drive<br>\ndrive.mount('/content/drive') # Mount folder<br>\npath = '/content/drive/Folder/file' # Get path to necessary folder</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 2404128,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "2023-08-23T04:33:52.550000",
      "content": "<p>i think you are not training with all the face landmark. you are using using a subset of the landmark. most people maybe are using less than 100 landmark points, and this is what you need to upload for training</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2404301,
          "author_name": "C R Suthikshn Kumar",
          "author_url": "",
          "post_date": "2023-08-23T07:09:42.410000",
          "content": "<p>Thanks for your response. <br>\nThrough some kaggle postings which provided steps for migration to colab Pro, I got the following code to run in <br>\nmy colab notebook:</p>\n<p>! kaggle competitions download -c asl-fingerspelling</p>\n<p>! mkdir kaggle<br>\n!unzip asl-fingerspelling + \".zip\" -d kaggle</p>\n<h1>Unmount your Google Drive</h1>\n<p>drive.flush_and_unmount()<br>\nCan you advise, update to this code to download specific files. </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2404108": "As I ran out of Kaggle GPU limit, I tried moving my Competition Notebook  and data to Colab Pro.\nWhile uploading the training and test data, I observe it is not supporting 189 GB. Half way through,\nit has run out of memory. Any advise on using Colab Pro for  ASLFR competition.",
    "2410536": "Did you try copying the files to Google Drive, mounting the folder, and using it? Below is an example - just set your paths.\n\nfrom google.colab import drive\ndrive.mount('/content/drive') # Mount folder\npath = '/content/drive/Folder/file' # Get path to necessary folder\n",
    "2404128": "i think you are not training with all the face landmark. you are using using a subset of the landmark. most people maybe are using less than 100 landmark points, and this is what you need to upload for training"
  }
}