{
  "id": 468776,
  "title": "Unable to fix scoring error",
  "url": "/competitions/blood-vessel-segmentation/discussion/468776",
  "author_name": "",
  "post_date": "2024-01-17T22:52:47.835811800Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I have been trying quite a bit to solve my scoring error. Here are some observations i noticed in my notebook's submission: <br>\n1) My submission seems to have correct number of rows and proper column name. Tested this by submitting dummy rle of \"1 0\" for every image in test set(this worked)<br>\n2) getting submission scoring error after submitting rle of predicted labels. Also if predicted mask is empty then rle is \"1 0\"<br>\n3) I do not think model is causing OOM cause while submitting the dummy rle, I also ran the model and got prediction for each image</p>\n<p>Could anyone give me suggestions based on your experiences on why one might be getting the scoring error?</p>",
  "messages": [
    {
      "id": "2606866",
      "postDate": "01/17/2024 22:52:47",
      "content": "<p>I have been trying quite a bit to solve my scoring error. Here are some observations i noticed in my notebook's submission: <br>\n1) My submission seems to have correct number of rows and proper column name. Tested this by submitting dummy rle of \"1 0\" for every image in test set(this worked)<br>\n2) getting submission scoring error after submitting rle of predicted labels. Also if predicted mask is empty then rle is \"1 0\"<br>\n3) I do not think model is causing OOM cause while submitting the dummy rle, I also ran the model and got prediction for each image</p>\n<p>Could anyone give me suggestions based on your experiences on why one might be getting the scoring error?</p>",
      "rawMarkdown": "I have been trying quite a bit to solve my scoring error. Here are some observations i noticed in my notebook's submission: \n1) My submission seems to have correct number of rows and proper column name. Tested this by submitting dummy rle of \"1 0\" for every image in test set(this worked)\n2) getting submission scoring error after submitting rle of predicted labels. Also if predicted mask is empty then rle is \"1 0\"\n3) I do not think model is causing OOM cause while submitting the dummy rle, I also ran the model and got prediction for each image\n\nCould anyone give me suggestions based on your experiences on why one might be getting the scoring error?",
      "votes": null
    },
    {
      "id": "2607478",
      "postDate": "01/18/2024 08:57:34",
      "content": "<p>If you are preprocessing inputs to feed your model you shoud restore the outputs too. If you are already doing it may be there is a small bug that makes output sizes differ from original ones.</p>",
      "rawMarkdown": "If you are preprocessing inputs to feed your model you shoud restore the outputs too. If you are already doing it may be there is a small bug that makes output sizes differ from original ones.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2607478,
      "author_name": "sacuscreed",
      "author_url": "",
      "post_date": "01/18/2024 08:57:34",
      "content": "<p>If you are preprocessing inputs to feed your model you shoud restore the outputs too. If you are already doing it may be there is a small bug that makes output sizes differ from original ones.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2606866": "I have been trying quite a bit to solve my scoring error. Here are some observations i noticed in my notebook's submission: \n1) My submission seems to have correct number of rows and proper column name. Tested this by submitting dummy rle of \"1 0\" for every image in test set(this worked)\n2) getting submission scoring error after submitting rle of predicted labels. Also if predicted mask is empty then rle is \"1 0\"\n3) I do not think model is causing OOM cause while submitting the dummy rle, I also ran the model and got prediction for each image\n\nCould anyone give me suggestions based on your experiences on why one might be getting the scoring error?",
    "2607478": "If you are preprocessing inputs to feed your model you shoud restore the outputs too. If you are already doing it may be there is a small bug that makes output sizes differ from original ones."
  },
  "source": "meta"
}