{
  "id": 178691,
  "title": "Memory allocation error while testing",
  "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/178691",
  "author_name": "",
  "post_date": "2020-08-31T03:43:05.372932800Z",
  "votes": -1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Is anyone else facing \"your notebook tried to allocate more memory than available\" error. I am not facing is while training but while testing on the dataset only. And it does not happen before at least 50% of the testing is complete.</p>",
  "messages": [
    {
      "id": "992203",
      "postDate": "08/31/2020 03:43:05",
      "content": "<p>Is anyone else facing \"your notebook tried to allocate more memory than available\" error. I am not facing is while training but while testing on the dataset only. And it does not happen before at least 50% of the testing is complete.</p>",
      "rawMarkdown": "Is anyone else facing \"your notebook tried to allocate more memory than available\" error. I am not facing is while training but while testing on the dataset only. And it does not happen before at least 50% of the testing is complete.",
      "votes": null
    },
    {
      "id": "992236",
      "postDate": "08/31/2020 04:20:14",
      "content": "<p>Check out this thread where I have tried to answer the same query.<br>\n<a href=\"https://www.kaggle.com/general/177871\" target=\"_blank\">https://www.kaggle.com/general/177871</a></p>",
      "rawMarkdown": "Check out this thread where I have tried to answer the same query.\nhttps://www.kaggle.com/general/177871",
      "votes": null
    },
    {
      "id": "992255",
      "postDate": "08/31/2020 04:40:53",
      "content": "<p>Thank you. Will take a look.</p>",
      "rawMarkdown": "Thank you. Will take a look.",
      "votes": null
    },
    {
      "id": "993322",
      "postDate": "08/31/2020 20:50:45",
      "content": "<p>Yes, I was facing the same issue. I think it's a memory leak issue. See here: <a href=\"url\" target=\"_blank\">https://github.com/pytorch/pytorch/issues/13246</a></p>\n<p>I solved it by setting num_workers in the dataloader to 0.</p>",
      "rawMarkdown": "Yes, I was facing the same issue. I think it's a memory leak issue. See here: [https://github.com/pytorch/pytorch/issues/13246](url)\n\nI solved it by setting num_workers in the dataloader to 0.",
      "votes": null
    },
    {
      "id": "993528",
      "postDate": "09/01/2020 02:10:02",
      "content": "<p>Thanks. Although I found a workaround, still I can try my original script now for testing.</p>",
      "rawMarkdown": "Thanks. Although I found a workaround, still I can try my original script now for testing.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 992236,
      "author_name": "oneplustricks",
      "author_url": "",
      "post_date": "08/31/2020 04:20:14",
      "content": "<p>Check out this thread where I have tried to answer the same query.<br>\n<a href=\"https://www.kaggle.com/general/177871\" target=\"_blank\">https://www.kaggle.com/general/177871</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 992255,
          "author_name": "sovitrath",
          "author_url": "",
          "post_date": "08/31/2020 04:40:53",
          "content": "<p>Thank you. Will take a look.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 993322,
      "author_name": "khushalbr",
      "author_url": "",
      "post_date": "08/31/2020 20:50:45",
      "content": "<p>Yes, I was facing the same issue. I think it's a memory leak issue. See here: <a href=\"url\" target=\"_blank\">https://github.com/pytorch/pytorch/issues/13246</a></p>\n<p>I solved it by setting num_workers in the dataloader to 0.</p>",
      "votes": null,
      "replies": [
        {
          "id": 993528,
          "author_name": "sovitrath",
          "author_url": "",
          "post_date": "09/01/2020 02:10:02",
          "content": "<p>Thanks. Although I found a workaround, still I can try my original script now for testing.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "992203": "Is anyone else facing \"your notebook tried to allocate more memory than available\" error. I am not facing is while training but while testing on the dataset only. And it does not happen before at least 50% of the testing is complete.",
    "992236": "Check out this thread where I have tried to answer the same query.\nhttps://www.kaggle.com/general/177871",
    "992255": "Thank you. Will take a look.",
    "993322": "Yes, I was facing the same issue. I think it's a memory leak issue. See here: [https://github.com/pytorch/pytorch/issues/13246](url)\n\nI solved it by setting num_workers in the dataloader to 0.",
    "993528": "Thanks. Although I found a workaround, still I can try my original script now for testing."
  },
  "source": "meta"
}