{
  "id": 378188,
  "title": "Scoring my submission is taking so much time. ",
  "url": "/competitions/rsna-breast-cancer-detection/discussion/378188",
  "author_name": "Nizar Haytham ",
  "post_date": "2023-01-14T18:17:49.030000",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I have trained a model in another notebook and I loaded the model to the notebook I am submittimg, It loads fast and every thing works very well on the small test dataset that is provided, I submitted it and it since 7 hours and it still didn't give a result, is this normal?<br>\n'knowing that I am using dicom library'. </p>",
  "messages": [
    {
      "id": 2100689,
      "postDate": "2023-01-15T11:06:38.937Z",
      "content": "<p>Yes, this is normal - look on these notebooks to speed up process:</p>\n<ul>\n<li><a href=\"https://www.kaggle.com/code/hengck23/3hr-tensorrt-nextvit-example\" target=\"_blank\">https://www.kaggle.com/code/hengck23/3hr-tensorrt-nextvit-example</a></li>\n<li><a href=\"https://www.kaggle.com/code/theoviel/rsna-breast-baseline-faster-inference-with-dali\" target=\"_blank\">https://www.kaggle.com/code/theoviel/rsna-breast-baseline-faster-inference-with-dali</a></li>\n</ul>",
      "rawMarkdown": "Yes, this is normal - look on these notebooks to speed up process:\n- https://www.kaggle.com/code/hengck23/3hr-tensorrt-nextvit-example\n- https://www.kaggle.com/code/theoviel/rsna-breast-baseline-faster-inference-with-dali\n",
      "votes": 1,
      "replies": [
        {
          "id": 2117856,
          "postDate": "2023-01-27T15:33:30.643Z",
          "content": "<p>Thanks for the links.</p>",
          "rawMarkdown": "Thanks for the links."
        }
      ]
    },
    {
      "id": 2100621,
      "postDate": "2023-01-15T09:57:34.813Z",
      "content": "<p>It took me also 9 hours. But I guess I got why it takes so much time for me. </p>\n<ul>\n<li>First, as there are lots of test images that we can't see, so it needs to process all the test images, I have heard dcmdl is faster than pydicom.</li>\n<li>Second could be while using inference, try to do some use of GPU and see whether it is faster. I will try them by myself in com coming weeks. I will try to give them feedback based on my try. </li>\n</ul>",
      "rawMarkdown": "It took me also 9 hours. But I guess I got why it takes so much time for me. \n- First, as there are lots of test images that we can't see, so it needs to process all the test images, I have heard dcmdl is faster than pydicom.\n- Second could be while using inference, try to do some use of GPU and see whether it is faster. I will try them by myself in com coming weeks. I will try to give them feedback based on my try. ",
      "votes": 1
    },
    {
      "id": 2099864,
      "postDate": "2023-01-14T18:17:49.030Z",
      "content": "<p>I have trained a model in another notebook and I loaded the model to the notebook I am submittimg, It loads fast and every thing works very well on the small test dataset that is provided, I submitted it and it since 7 hours and it still didn't give a result, is this normal?<br>\n'knowing that I am using dicom library'. </p>",
      "rawMarkdown": "I have trained a model in another notebook and I loaded the model to the notebook I am submittimg, It loads fast and every thing works very well on the small test dataset that is provided, I submitted it and it since 7 hours and it still didn't give a result, is this normal?\n'knowing that I am using dicom library'. ",
      "votes": 1
    },
    {
      "id": 2120396,
      "postDate": "2023-01-29T15:07:25.053Z",
      "content": "<p>You can use multithreading to accelerate.(If you do not add this code to your submissions)</p>",
      "rawMarkdown": "You can use multithreading to accelerate.(If you do not add this code to your submissions)"
    },
    {
      "id": 2099909,
      "postDate": "2023-01-14T19:03:51.340Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 2100689,
      "author_name": "Remek Kinas",
      "author_url": "",
      "post_date": "2023-01-15T11:06:38.937000",
      "content": "<p>Yes, this is normal - look on these notebooks to speed up process:</p>\n<ul>\n<li><a href=\"https://www.kaggle.com/code/hengck23/3hr-tensorrt-nextvit-example\" target=\"_blank\">https://www.kaggle.com/code/hengck23/3hr-tensorrt-nextvit-example</a></li>\n<li><a href=\"https://www.kaggle.com/code/theoviel/rsna-breast-baseline-faster-inference-with-dali\" target=\"_blank\">https://www.kaggle.com/code/theoviel/rsna-breast-baseline-faster-inference-with-dali</a></li>\n</ul>",
      "votes": 1,
      "replies": [
        {
          "id": 2117856,
          "author_name": "Mohammed Hasan goni",
          "author_url": "",
          "post_date": "2023-01-27T15:33:30.643000",
          "content": "<p>Thanks for the links.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2100621,
      "author_name": "Mohammed Hasan goni",
      "author_url": "",
      "post_date": "2023-01-15T09:57:34.813000",
      "content": "<p>It took me also 9 hours. But I guess I got why it takes so much time for me. </p>\n<ul>\n<li>First, as there are lots of test images that we can't see, so it needs to process all the test images, I have heard dcmdl is faster than pydicom.</li>\n<li>Second could be while using inference, try to do some use of GPU and see whether it is faster. I will try them by myself in com coming weeks. I will try to give them feedback based on my try. </li>\n</ul>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 2120396,
      "author_name": "yoyobar",
      "author_url": "",
      "post_date": "2023-01-29T15:07:25.053000",
      "content": "<p>You can use multithreading to accelerate.(If you do not add this code to your submissions)</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 2099909,
      "author_name": "",
      "author_url": "",
      "post_date": "2023-01-14T19:03:51.340000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2100689": "Yes, this is normal - look on these notebooks to speed up process:\n- https://www.kaggle.com/code/hengck23/3hr-tensorrt-nextvit-example\n- https://www.kaggle.com/code/theoviel/rsna-breast-baseline-faster-inference-with-dali\n",
    "2100621": "It took me also 9 hours. But I guess I got why it takes so much time for me. \n- First, as there are lots of test images that we can't see, so it needs to process all the test images, I have heard dcmdl is faster than pydicom.\n- Second could be while using inference, try to do some use of GPU and see whether it is faster. I will try them by myself in com coming weeks. I will try to give them feedback based on my try. ",
    "2099864": "I have trained a model in another notebook and I loaded the model to the notebook I am submittimg, It loads fast and every thing works very well on the small test dataset that is provided, I submitted it and it since 7 hours and it still didn't give a result, is this normal?\n'knowing that I am using dicom library'. ",
    "2120396": "You can use multithreading to accelerate.(If you do not add this code to your submissions)",
    "2099909": ""
  }
}