{
  "id": 339544,
  "title": "Time out while inference",
  "url": "/competitions/hubmap-organ-segmentation/discussion/339544",
  "author_name": "",
  "post_date": "2022-07-25T11:21:48.685841400Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>It seems Unpublished data is vast. (from data description).<br>\nWhat can I do to reduce time cost?<br>\nI built my baseline via mmlab's mmseg and mmdet.</p>",
  "messages": [
    {
      "id": "1870201",
      "postDate": "07/25/2022 11:21:48",
      "content": "<p>It seems Unpublished data is vast. (from data description).<br>\nWhat can I do to reduce time cost?<br>\nI built my baseline via mmlab's mmseg and mmdet.</p>",
      "rawMarkdown": "It seems Unpublished data is vast. (from data description).\nWhat can I do to reduce time cost?\nI built my baseline via mmlab's mmseg and mmdet.",
      "votes": null
    },
    {
      "id": "1871315",
      "postDate": "07/26/2022 07:58:38",
      "content": "<p>In the past competition, I have the same problem with you because of out memory problem. The problem came from appending the predicted mask to the result.</p>",
      "rawMarkdown": "In the past competition, I have the same problem with you because of out memory problem. The problem came from appending the predicted mask to the result.",
      "votes": null
    },
    {
      "id": "1871529",
      "postDate": "07/26/2022 10:14:20",
      "content": "<p>If you're interested in reducing the inference time, you might want to try out larger batch sizes (if they fit in the memory).</p>\n<p>The Devastator.</p>",
      "rawMarkdown": "If you're interested in reducing the inference time, you might want to try out larger batch sizes (if they fit in the memory).\n\nThe Devastator.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1871315,
      "author_name": "thancaocuong",
      "author_url": "",
      "post_date": "07/26/2022 07:58:38",
      "content": "<p>In the past competition, I have the same problem with you because of out memory problem. The problem came from appending the predicted mask to the result.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1871529,
      "author_name": "thedevastator",
      "author_url": "",
      "post_date": "07/26/2022 10:14:20",
      "content": "<p>If you're interested in reducing the inference time, you might want to try out larger batch sizes (if they fit in the memory).</p>\n<p>The Devastator.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1870201": "It seems Unpublished data is vast. (from data description).\nWhat can I do to reduce time cost?\nI built my baseline via mmlab's mmseg and mmdet.",
    "1871315": "In the past competition, I have the same problem with you because of out memory problem. The problem came from appending the predicted mask to the result.",
    "1871529": "If you're interested in reducing the inference time, you might want to try out larger batch sizes (if they fit in the memory).\n\nThe Devastator."
  },
  "source": "meta"
}