{
  "id": 543623,
  "title": "How can I avoid CPU crash while training the model?",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/543623",
  "author_name": "",
  "post_date": "2024-10-31T17:46:17.388793200Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I am facing the CPU crash as the training volume is reasonably big.<br>\nIs there any solution to avoid the CPU crash?</p>",
  "messages": [
    {
      "id": "3033041",
      "postDate": "10/31/2024 17:46:17",
      "content": "<p>I am facing the CPU crash as the training volume is reasonably big.<br>\nIs there any solution to avoid the CPU crash?</p>",
      "rawMarkdown": "I am facing the CPU crash as the training volume is reasonably big.\nIs there any solution to avoid the CPU crash?",
      "votes": null
    },
    {
      "id": "3033066",
      "postDate": "10/31/2024 18:09:56",
      "content": "<p>Don't train on CPU, it will crash for sure. I recommend you to switch to another GPU provider like <a href=\"runpod.io\" target=\"_blank\">runpod.io</a> <a href=\"https://www.kaggle.com/farhankardan\" target=\"_blank\">@farhankardan</a> </p>",
      "rawMarkdown": "Don't train on CPU, it will crash for sure. I recommend you to switch to another GPU provider like [runpod.io](runpod.io) @farhankardan",
      "votes": null
    },
    {
      "id": "3033208",
      "postDate": "10/31/2024 22:04:55",
      "content": "<p>1-try using portion of the data rather than all of it <br>\n2- write your code carefully to avoid exhausting the memory and git rid of any variables that you're done with </p>",
      "rawMarkdown": "1-try using portion of the data rather than all of it \n2- write your code carefully to avoid exhausting the memory and git rid of any variables that you're done with",
      "votes": null
    },
    {
      "id": "3033580",
      "postDate": "11/01/2024 10:23:54",
      "content": "<p>Thanks for your solutions ((: <br>\nThat is exactly where I have problem. How I can train the model properly if I train it only one partition? </p>",
      "rawMarkdown": "Thanks for your solutions ((: \nThat is exactly where I have problem. How I can train the model properly if I train it only one partition?",
      "votes": null
    },
    {
      "id": "3033581",
      "postDate": "11/01/2024 10:24:17",
      "content": "<p>Sure, I'll try it and see if it can become any better</p>",
      "rawMarkdown": "Sure, I'll try it and see if it can become any better",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3033066,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "10/31/2024 18:09:56",
      "content": "<p>Don't train on CPU, it will crash for sure. I recommend you to switch to another GPU provider like <a href=\"runpod.io\" target=\"_blank\">runpod.io</a> <a href=\"https://www.kaggle.com/farhankardan\" target=\"_blank\">@farhankardan</a> </p>",
      "votes": null,
      "replies": [
        {
          "id": 3033581,
          "author_name": "farhankardan",
          "author_url": "",
          "post_date": "11/01/2024 10:24:17",
          "content": "<p>Sure, I'll try it and see if it can become any better</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3033208,
      "author_name": "aymanallawi",
      "author_url": "",
      "post_date": "10/31/2024 22:04:55",
      "content": "<p>1-try using portion of the data rather than all of it <br>\n2- write your code carefully to avoid exhausting the memory and git rid of any variables that you're done with </p>",
      "votes": null,
      "replies": [
        {
          "id": 3033580,
          "author_name": "farhankardan",
          "author_url": "",
          "post_date": "11/01/2024 10:23:54",
          "content": "<p>Thanks for your solutions ((: <br>\nThat is exactly where I have problem. How I can train the model properly if I train it only one partition? </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3033041": "I am facing the CPU crash as the training volume is reasonably big.\nIs there any solution to avoid the CPU crash?",
    "3033066": "Don't train on CPU, it will crash for sure. I recommend you to switch to another GPU provider like [runpod.io](runpod.io) @farhankardan",
    "3033208": "1-try using portion of the data rather than all of it \n2- write your code carefully to avoid exhausting the memory and git rid of any variables that you're done with",
    "3033580": "Thanks for your solutions ((: \nThat is exactly where I have problem. How I can train the model properly if I train it only one partition?",
    "3033581": "Sure, I'll try it and see if it can become any better"
  },
  "source": "meta"
}