{
  "id": 181514,
  "title": "RuntimeError: CUDA error: device-side assert triggered",
  "url": "/competitions/birdsong-recognition/discussion/181514",
  "author_name": "",
  "post_date": "2020-09-09T07:03:21.945954500Z",
  "votes": 4,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hi. I ran my code for more epochs and after 34 epoch I got the above error. Any idea why is this happening so ?</p>",
  "messages": [
    {
      "id": "1003682",
      "postDate": "09/09/2020 07:03:21",
      "content": "<p>Hi. I ran my code for more epochs and after 34 epoch I got the above error. Any idea why is this happening so ?</p>",
      "rawMarkdown": "Hi. I ran my code for more epochs and after 34 epoch I got the above error. Any idea why is this happening so ?",
      "votes": null
    },
    {
      "id": "1003763",
      "postDate": "09/09/2020 08:20:23",
      "content": "<p>I had this error in the past with pytorch. It was often an error in tensor dimensions.  This discussion may help you diagnose your issue <a href=\"https://github.com/pytorch/pytorch/issues/4144\" target=\"_blank\">https://github.com/pytorch/pytorch/issues/4144</a></p>\n<p>reaidng other error reports, I see it is always an out of bound condition violation.  </p>\n<p>It could be a pytorch bug as well.</p>\n<p>Well, if you use pytorch.</p>",
      "rawMarkdown": "I had this error in the past with pytorch. It was often an error in tensor dimensions.  This discussion may help you diagnose your issue https://github.com/pytorch/pytorch/issues/4144\n\nreaidng other error reports, I see it is always an out of bound condition violation.  \n\nIt could be a pytorch bug as well.\n\nWell, if you use pytorch.",
      "votes": null
    },
    {
      "id": "1003798",
      "postDate": "09/09/2020 09:02:41",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a> . Pleasure to read your discussions as always. Thank you for the prompt response. :)</p>",
      "rawMarkdown": "Hi @cpmpml . Pleasure to read your discussions as always. Thank you for the prompt response. :)",
      "votes": null
    },
    {
      "id": "1003926",
      "postDate": "09/09/2020 11:10:59",
      "content": "<p><a href=\"https://www.kaggle.com/hidehisaarai1213/introduction-to-sound-event-detection/comments#975258\" target=\"_blank\">https://www.kaggle.com/hidehisaarai1213/introduction-to-sound-event-detection/comments#975258</a></p>",
      "rawMarkdown": "https://www.kaggle.com/hidehisaarai1213/introduction-to-sound-event-detection/comments#975258",
      "votes": null
    },
    {
      "id": "1004047",
      "postDate": "09/09/2020 12:26:23",
      "content": "<p>use torch.clamp. Error because the input of BCELoss cannot out of [0 ,1]</p>",
      "rawMarkdown": "use torch.clamp. Error because the input of BCELoss cannot out of [0 ,1]",
      "votes": null
    },
    {
      "id": "1004272",
      "postDate": "09/09/2020 15:16:51",
      "content": "<p>I faced the same issue while using crossentropy loss, check the target labels proper or not i.e target label value should not cross the no of classes we are predicting  at output layer</p>",
      "rawMarkdown": "I faced the same issue while using crossentropy loss, check the target labels proper or not i.e target label value should not cross the no of classes we are predicting  at output layer",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1003763,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "09/09/2020 08:20:23",
      "content": "<p>I had this error in the past with pytorch. It was often an error in tensor dimensions.  This discussion may help you diagnose your issue <a href=\"https://github.com/pytorch/pytorch/issues/4144\" target=\"_blank\">https://github.com/pytorch/pytorch/issues/4144</a></p>\n<p>reaidng other error reports, I see it is always an out of bound condition violation.  </p>\n<p>It could be a pytorch bug as well.</p>\n<p>Well, if you use pytorch.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1003798,
          "author_name": "jamshaidsohail5",
          "author_url": "",
          "post_date": "09/09/2020 09:02:41",
          "content": "<p>Hi <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a> . Pleasure to read your discussions as always. Thank you for the prompt response. :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1003926,
      "author_name": "atamazian",
      "author_url": "",
      "post_date": "09/09/2020 11:10:59",
      "content": "<p><a href=\"https://www.kaggle.com/hidehisaarai1213/introduction-to-sound-event-detection/comments#975258\" target=\"_blank\">https://www.kaggle.com/hidehisaarai1213/introduction-to-sound-event-detection/comments#975258</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1004047,
      "author_name": "doanquanvietnamca",
      "author_url": "",
      "post_date": "09/09/2020 12:26:23",
      "content": "<p>use torch.clamp. Error because the input of BCELoss cannot out of [0 ,1]</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1004272,
      "author_name": "aditya23071991",
      "author_url": "",
      "post_date": "09/09/2020 15:16:51",
      "content": "<p>I faced the same issue while using crossentropy loss, check the target labels proper or not i.e target label value should not cross the no of classes we are predicting  at output layer</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1003682": "Hi. I ran my code for more epochs and after 34 epoch I got the above error. Any idea why is this happening so ?",
    "1003763": "I had this error in the past with pytorch. It was often an error in tensor dimensions.  This discussion may help you diagnose your issue https://github.com/pytorch/pytorch/issues/4144\n\nreaidng other error reports, I see it is always an out of bound condition violation.  \n\nIt could be a pytorch bug as well.\n\nWell, if you use pytorch.",
    "1003798": "Hi @cpmpml . Pleasure to read your discussions as always. Thank you for the prompt response. :)",
    "1003926": "https://www.kaggle.com/hidehisaarai1213/introduction-to-sound-event-detection/comments#975258",
    "1004047": "use torch.clamp. Error because the input of BCELoss cannot out of [0 ,1]",
    "1004272": "I faced the same issue while using crossentropy loss, check the target labels proper or not i.e target label value should not cross the no of classes we are predicting  at output layer"
  },
  "source": "meta"
}