{
  "id": 139052,
  "title": "Google Colab SIGSEV Error",
  "url": "/competitions/jigsaw-multilingual-toxic-comment-classification/discussion/139052",
  "author_name": "",
  "post_date": "2020-03-27T08:17:32.449084700Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I tried to run abhishek's kernel on google colab. It is giving SIGSEV error at the end of one epoch. Anyone facing the same problem?</p>",
  "messages": [
    {
      "id": "787927",
      "postDate": "03/27/2020 08:17:32",
      "content": "<p>I tried to run abhishek's kernel on google colab. It is giving SIGSEV error at the end of one epoch. Anyone facing the same problem?</p>",
      "rawMarkdown": "I tried to run abhishek's kernel on google colab. It is giving SIGSEV error at the end of one epoch. Anyone facing the same problem?",
      "votes": null
    },
    {
      "id": "788185",
      "postDate": "03/27/2020 13:10:40",
      "content": "<p>You should set nproc=1,I guess the memmory is not enough.You can refer https://github.com/PyTorchLightning/pytorch-lightning/issues/996</p>",
      "rawMarkdown": "You should set nproc=1,I guess the memmory is not enough.You can refer [https://github.com/PyTorchLightning/pytorch-lightning/issues/996](this )",
      "votes": null
    },
    {
      "id": "788229",
      "postDate": "03/27/2020 13:52:51",
      "content": "<p>Thanks. Your provided link is what I am looking for. \n<a href=\"https://github.com/PyTorchLightning/pytorch-lightning/issues/996\">https://github.com/PyTorchLightning/pytorch-lightning/issues/996</a></p>",
      "rawMarkdown": "Thanks. Your provided link is what I am looking for. \nhttps://github.com/PyTorchLightning/pytorch-lightning/issues/996",
      "votes": null
    },
    {
      "id": "788432",
      "postDate": "03/27/2020 17:01:17",
      "content": "<p>It's also not the same TPU on Colab. TPUv2 on Colab, TPUv3 on Kaggle with twice the memory and processing power. A model that runs on the bigger one can OOM on the smaller. You can try to reduce the batch size if it happens.</p>",
      "rawMarkdown": "It's also not the same TPU on Colab. TPUv2 on Colab, TPUv3 on Kaggle with twice the memory and processing power. A model that runs on the bigger one can OOM on the smaller. You can try to reduce the batch size if it happens.",
      "votes": null
    },
    {
      "id": "788709",
      "postDate": "03/27/2020 23:03:37",
      "content": "<p>Thanks, Kaggle Team.</p>",
      "rawMarkdown": "Thanks, Kaggle Team.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 788185,
      "author_name": "townmeng",
      "author_url": "",
      "post_date": "03/27/2020 13:10:40",
      "content": "<p>You should set nproc=1,I guess the memmory is not enough.You can refer https://github.com/PyTorchLightning/pytorch-lightning/issues/996</p>",
      "votes": null,
      "replies": [
        {
          "id": 788229,
          "author_name": "shayekh",
          "author_url": "",
          "post_date": "03/27/2020 13:52:51",
          "content": "<p>Thanks. Your provided link is what I am looking for. \n<a href=\"https://github.com/PyTorchLightning/pytorch-lightning/issues/996\">https://github.com/PyTorchLightning/pytorch-lightning/issues/996</a></p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 788432,
      "author_name": "mgorner",
      "author_url": "",
      "post_date": "03/27/2020 17:01:17",
      "content": "<p>It's also not the same TPU on Colab. TPUv2 on Colab, TPUv3 on Kaggle with twice the memory and processing power. A model that runs on the bigger one can OOM on the smaller. You can try to reduce the batch size if it happens.</p>",
      "votes": null,
      "replies": [
        {
          "id": 788709,
          "author_name": "shayekh",
          "author_url": "",
          "post_date": "03/27/2020 23:03:37",
          "content": "<p>Thanks, Kaggle Team.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "787927": "I tried to run abhishek's kernel on google colab. It is giving SIGSEV error at the end of one epoch. Anyone facing the same problem?",
    "788185": "You should set nproc=1,I guess the memmory is not enough.You can refer [https://github.com/PyTorchLightning/pytorch-lightning/issues/996](this )",
    "788229": "Thanks. Your provided link is what I am looking for. \nhttps://github.com/PyTorchLightning/pytorch-lightning/issues/996",
    "788432": "It's also not the same TPU on Colab. TPUv2 on Colab, TPUv3 on Kaggle with twice the memory and processing power. A model that runs on the bigger one can OOM on the smaller. You can try to reduce the batch size if it happens.",
    "788709": "Thanks, Kaggle Team."
  },
  "source": "meta"
}