{
  "id": 243178,
  "title": "Cuda out of memory??",
  "url": "/competitions/coleridgeinitiative-show-us-the-data/discussion/243178",
  "author_name": "",
  "post_date": "2021-06-01T13:10:26.123862900Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>what are the reasons of getting cuda out of memory?? and What are the ways to solve it</p>",
  "messages": [
    {
      "id": "1331424",
      "postDate": "06/01/2021 13:10:26",
      "content": "<p>what are the reasons of getting cuda out of memory?? and What are the ways to solve it</p>",
      "rawMarkdown": "what are the reasons of getting cuda out of memory?? and What are the ways to solve it",
      "votes": null
    },
    {
      "id": "1331436",
      "postDate": "06/01/2021 13:17:02",
      "content": "<p>Previous related reply：</p>\n<p>Notebook Exceeded Allowed Compute means that the memory requirements are too high for the system to meet. The largest memory allocation in the GPU environment is 13GB. In the past few weeks, I was also affected by memory allocation problems and kept submitting failed. There are a few ideas that may help you:</p>\n<p>Reduce the length of the input text. Choosing an appropriate text length can reduce the number of input samples to be constructed.<br>\nOptimize the preprocessing process and reduce the creation of variables containing a large number of data.<br>\nReduce the construction of intermediate variables. Consider creating tensor variables such as Input id when running the GPU, which can reduce memory usage by 80%.<br>\ngood luck!</p>",
      "rawMarkdown": "Previous related reply：\n\nNotebook Exceeded Allowed Compute means that the memory requirements are too high for the system to meet. The largest memory allocation in the GPU environment is 13GB. In the past few weeks, I was also affected by memory allocation problems and kept submitting failed. There are a few ideas that may help you:\n\nReduce the length of the input text. Choosing an appropriate text length can reduce the number of input samples to be constructed.\nOptimize the preprocessing process and reduce the creation of variables containing a large number of data.\nReduce the construction of intermediate variables. Consider creating tensor variables such as Input id when running the GPU, which can reduce memory usage by 80%.\ngood luck!",
      "votes": null
    },
    {
      "id": "1331886",
      "postDate": "06/01/2021 19:33:20",
      "content": "<p>Reducing your batch size, sequence length, or optimizing the data types can help decrease the memory usage. </p>",
      "rawMarkdown": "Reducing your batch size, sequence length, or optimizing the data types can help decrease the memory usage.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1331436,
      "author_name": "vanle73",
      "author_url": "",
      "post_date": "06/01/2021 13:17:02",
      "content": "<p>Previous related reply：</p>\n<p>Notebook Exceeded Allowed Compute means that the memory requirements are too high for the system to meet. The largest memory allocation in the GPU environment is 13GB. In the past few weeks, I was also affected by memory allocation problems and kept submitting failed. There are a few ideas that may help you:</p>\n<p>Reduce the length of the input text. Choosing an appropriate text length can reduce the number of input samples to be constructed.<br>\nOptimize the preprocessing process and reduce the creation of variables containing a large number of data.<br>\nReduce the construction of intermediate variables. Consider creating tensor variables such as Input id when running the GPU, which can reduce memory usage by 80%.<br>\ngood luck!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1331886,
      "author_name": "pinterbagz",
      "author_url": "",
      "post_date": "06/01/2021 19:33:20",
      "content": "<p>Reducing your batch size, sequence length, or optimizing the data types can help decrease the memory usage. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1331424": "what are the reasons of getting cuda out of memory?? and What are the ways to solve it",
    "1331436": "Previous related reply：\n\nNotebook Exceeded Allowed Compute means that the memory requirements are too high for the system to meet. The largest memory allocation in the GPU environment is 13GB. In the past few weeks, I was also affected by memory allocation problems and kept submitting failed. There are a few ideas that may help you:\n\nReduce the length of the input text. Choosing an appropriate text length can reduce the number of input samples to be constructed.\nOptimize the preprocessing process and reduce the creation of variables containing a large number of data.\nReduce the construction of intermediate variables. Consider creating tensor variables such as Input id when running the GPU, which can reduce memory usage by 80%.\ngood luck!",
    "1331886": "Reducing your batch size, sequence length, or optimizing the data types can help decrease the memory usage."
  },
  "source": "meta"
}