{
  "id": 147771,
  "title": "Class Balance with PyTorch/XLA",
  "url": "/competitions/jigsaw-multilingual-toxic-comment-classification/discussion/147771",
  "author_name": "Alex Shonenkov",
  "post_date": "2020-05-01T17:42:19.774000",
  "votes": 7,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi everyone!</p>\n\n<p>I have published new kernel about class balance and solution for distributed training with PyTorch/XLA using streaming balance samplers</p>\n\n<p>It can help you avoid OOM problems and use full set for training</p>\n\n<p><a href=\"https://www.kaggle.com/shonenkov/class-balance-with-pytorch-xla\">https://www.kaggle.com/shonenkov/class-balance-with-pytorch-xla</a></p>",
  "messages": [
    {
      "id": 829337,
      "postDate": "2020-05-01T17:42:19.773Z",
      "content": "<p>Hi everyone!</p>\n\n<p>I have published new kernel about class balance and solution for distributed training with PyTorch/XLA using streaming balance samplers</p>\n\n<p>It can help you avoid OOM problems and use full set for training</p>\n\n<p><a href=\"https://www.kaggle.com/shonenkov/class-balance-with-pytorch-xla\">https://www.kaggle.com/shonenkov/class-balance-with-pytorch-xla</a></p>",
      "rawMarkdown": "Hi everyone!\n\nI have published new kernel about class balance and solution for distributed training with PyTorch/XLA using streaming balance samplers\n\nIt can help you avoid OOM problems and use full set for training\n\n\nhttps://www.kaggle.com/shonenkov/class-balance-with-pytorch-xla",
      "votes": 7
    },
    {
      "id": 829809,
      "postDate": "2020-05-02T05:29:38.597Z",
      "rawMarkdown": "",
      "votes": -5,
      "isDeleted": true
    },
    {
      "id": 829384,
      "postDate": "2020-05-01T18:40:38.027Z",
      "content": "<p>Thanks for sharing :)</p>",
      "rawMarkdown": "Thanks for sharing :)"
    }
  ],
  "comments": [
    {
      "id": 829809,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-05-02T05:29:38.597000",
      "content": "",
      "votes": -5,
      "replies": []
    },
    {
      "id": 829384,
      "author_name": "Alberto Maria Falletta",
      "author_url": "",
      "post_date": "2020-05-01T18:40:38.027000",
      "content": "<p>Thanks for sharing :)</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "829337": "Hi everyone!\n\nI have published new kernel about class balance and solution for distributed training with PyTorch/XLA using streaming balance samplers\n\nIt can help you avoid OOM problems and use full set for training\n\n\nhttps://www.kaggle.com/shonenkov/class-balance-with-pytorch-xla",
    "829809": "",
    "829384": "Thanks for sharing :)"
  }
}