{
  "id": 361942,
  "title": "google not sharing its wisdom with google colab",
  "url": "/competitions/rsna-2022-cervical-spine-fracture-detection/discussion/361942",
  "author_name": "GUNER",
  "post_date": "2022-10-24T19:18:52.319000",
  "votes": 7,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I am trying to train a model for this competition, as is, in V-100 in colab pro.<br>\nIt is insistently giving me A100 that charges more than twice, for my modest model.<br>\nI probably tried to open/close 10 sessions, and no… consistently and insistently it gives A-100.<br>\n(dont tell its a resource availabilty issue:-)</p>\n<p>This issue: 'that the user does not want that huge A100' could have been solved with some if/then/else statements from way back couldn't it?</p>",
  "messages": [
    {
      "id": 2002422,
      "postDate": "2022-10-24T19:18:52.320Z",
      "content": "<p>I am trying to train a model for this competition, as is, in V-100 in colab pro.<br>\nIt is insistently giving me A100 that charges more than twice, for my modest model.<br>\nI probably tried to open/close 10 sessions, and no… consistently and insistently it gives A-100.<br>\n(dont tell its a resource availabilty issue:-)</p>\n<p>This issue: 'that the user does not want that huge A100' could have been solved with some if/then/else statements from way back couldn't it?</p>",
      "rawMarkdown": "I am trying to train a model for this competition, as is, in V-100 in colab pro.\nIt is insistently giving me A100 that charges more than twice, for my modest model.\nI probably tried to open/close 10 sessions, and no... consistently and insistently it gives A-100.\n(dont tell its a resource availabilty issue:-)\n\nThis issue: 'that the user does not want that huge A100' could have been solved with some if/then/else statements from way back couldn't it?\n",
      "votes": 7
    },
    {
      "id": 2003062,
      "postDate": "2022-10-25T09:18:33.307Z",
      "content": "<p>I also noticed that. I used to train a lot of models on Colab for kaggling but..</p>\n<p>According to Colab's latest pricing - around 50$ / 32h for a single A100.<br>\nCompare this to when Colab Pro+ was first released, we were able to use three V100s at the same time for $50 / month.</p>\n<p>My guess is that Google probably intends to remove the majority of GPUs from Colab to deal with this bad economic situation..</p>",
      "rawMarkdown": "I also noticed that. I used to train a lot of models on Colab for kaggling but..\n\nAccording to Colab's latest pricing - around 50$ / 32h for a single A100.\nCompare this to when Colab Pro+ was first released, we were able to use three V100s at the same time for $50 / month.\n\nMy guess is that Google probably intends to remove the majority of GPUs from Colab to deal with this bad economic situation..",
      "votes": 1,
      "replies": [
        {
          "id": 2003640,
          "postDate": "2022-10-25T16:58:21.133Z",
          "content": "<p>Yes, probably. It might also be a gear change to more profit-making selection/allocation algorithms. <br>\nWhat also sucks is that, such changes are being advertised in 50s tv advertisement style:  'big exiting news and changes for our customers to make their life more easy',,or more like those ads in groomy science fiction movies where 'the boss is always right, kind and merciful, and we should all be more grateful then ever' :-)<br>\nHopefully, they don't mess with Kaggle though, it feels so good here being with all these talents and good people from around the world..</p>",
          "rawMarkdown": "Yes, probably. It might also be a gear change to more profit-making selection/allocation algorithms. \nWhat also sucks is that, such changes are being advertised in 50s tv advertisement style:  'big exiting news and changes for our customers to make their life more easy',,or more like those ads in groomy science fiction movies where 'the boss is always right, kind and merciful, and we should all be more grateful then ever' :-)\nHopefully, they don't mess with Kaggle though, it feels so good here being with all these talents and good people from around the world..",
          "votes": 1
        },
        {
          "id": 2004688,
          "postDate": "2022-10-26T13:18:54.543Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    },
    {
      "id": 2005112,
      "postDate": "2022-10-26T18:32:00.667Z",
      "content": "<p>So it is ~1.56$/h for A100 on colab now, is that official or own estimation? What a timing, they released colab in Sweden some days ago ;)<br>\nBut is glad for the Kaggle T4x2 with extra memory :) 👍🙏</p>",
      "rawMarkdown": "So it is ~1.56$/h for A100 on colab now, is that official or own estimation? What a timing, they released colab in Sweden some days ago ;)\nBut is glad for the Kaggle T4x2 with extra memory :) 👍🙏"
    }
  ],
  "comments": [
    {
      "id": 2003062,
      "author_name": "Qishen Ha",
      "author_url": "",
      "post_date": "2022-10-25T09:18:33.307000",
      "content": "<p>I also noticed that. I used to train a lot of models on Colab for kaggling but..</p>\n<p>According to Colab's latest pricing - around 50$ / 32h for a single A100.<br>\nCompare this to when Colab Pro+ was first released, we were able to use three V100s at the same time for $50 / month.</p>\n<p>My guess is that Google probably intends to remove the majority of GPUs from Colab to deal with this bad economic situation..</p>",
      "votes": 1,
      "replies": [
        {
          "id": 2003640,
          "author_name": "GUNER",
          "author_url": "",
          "post_date": "2022-10-25T16:58:21.133000",
          "content": "<p>Yes, probably. It might also be a gear change to more profit-making selection/allocation algorithms. <br>\nWhat also sucks is that, such changes are being advertised in 50s tv advertisement style:  'big exiting news and changes for our customers to make their life more easy',,or more like those ads in groomy science fiction movies where 'the boss is always right, kind and merciful, and we should all be more grateful then ever' :-)<br>\nHopefully, they don't mess with Kaggle though, it feels so good here being with all these talents and good people from around the world..</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 2004688,
          "author_name": "",
          "author_url": "",
          "post_date": "2022-10-26T13:18:54.543000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2005112,
      "author_name": "Kirderf",
      "author_url": "",
      "post_date": "2022-10-26T18:32:00.667000",
      "content": "<p>So it is ~1.56$/h for A100 on colab now, is that official or own estimation? What a timing, they released colab in Sweden some days ago ;)<br>\nBut is glad for the Kaggle T4x2 with extra memory :) 👍🙏</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2002422": "I am trying to train a model for this competition, as is, in V-100 in colab pro.\nIt is insistently giving me A100 that charges more than twice, for my modest model.\nI probably tried to open/close 10 sessions, and no... consistently and insistently it gives A-100.\n(dont tell its a resource availabilty issue:-)\n\nThis issue: 'that the user does not want that huge A100' could have been solved with some if/then/else statements from way back couldn't it?\n",
    "2003062": "I also noticed that. I used to train a lot of models on Colab for kaggling but..\n\nAccording to Colab's latest pricing - around 50$ / 32h for a single A100.\nCompare this to when Colab Pro+ was first released, we were able to use three V100s at the same time for $50 / month.\n\nMy guess is that Google probably intends to remove the majority of GPUs from Colab to deal with this bad economic situation..",
    "2005112": "So it is ~1.56$/h for A100 on colab now, is that official or own estimation? What a timing, they released colab in Sweden some days ago ;)\nBut is glad for the Kaggle T4x2 with extra memory :) 👍🙏"
  }
}