{
  "id": 442987,
  "title": "Read data from kaggle/working in submission process",
  "url": "/competitions/bengaliai-speech/discussion/442987",
  "author_name": "",
  "post_date": "2023-09-25T04:55:37.006328700Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>In Submission Process, I convert test dataset from mp3 files to wav file and save to kaggle/working folders (default). So can I read those files from kaggle/working to create TestDataloader ???</p>\n<p>Notebook run successfully in example test dataset but \"Notebook Threw Exception\" when run on real test dataset.</p>",
  "messages": [
    {
      "id": "2454740",
      "postDate": "09/25/2023 04:55:37",
      "content": "<p>In Submission Process, I convert test dataset from mp3 files to wav file and save to kaggle/working folders (default). So can I read those files from kaggle/working to create TestDataloader ???</p>\n<p>Notebook run successfully in example test dataset but \"Notebook Threw Exception\" when run on real test dataset.</p>",
      "rawMarkdown": "In Submission Process, I convert test dataset from mp3 files to wav file and save to kaggle/working folders (default). So can I read those files from kaggle/working to create TestDataloader ???\n\nNotebook run successfully in example test dataset but \"Notebook Threw Exception\" when run on real test dataset.",
      "votes": null
    },
    {
      "id": "2455700",
      "postDate": "09/25/2023 16:47:23",
      "content": "<p>Hi. There could be two reasons.</p>\n<ul>\n<li>Kaggle/working has around 20GB of disk memory. If your files take up more space than that, your notebook will crash.</li>\n<li>The total inference time allowed is 9 hours, which means that your notebook should be able to predict the 8000 test files within 9 hours time. Converting to WAV and loading takes significant time, leading to crossing the limit. Make sure your notebook can finish inference within the 9-hour limit.<br>\nGood luck!</li>\n</ul>",
      "rawMarkdown": "Hi. There could be two reasons.\n- Kaggle/working has around 20GB of disk memory. If your files take up more space than that, your notebook will crash.\n- The total inference time allowed is 9 hours, which means that your notebook should be able to predict the 8000 test files within 9 hours time. Converting to WAV and loading takes significant time, leading to crossing the limit. Make sure your notebook can finish inference within the 9-hour limit.\nGood luck!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2455700,
      "author_name": "mbmmurad",
      "author_url": "",
      "post_date": "09/25/2023 16:47:23",
      "content": "<p>Hi. There could be two reasons.</p>\n<ul>\n<li>Kaggle/working has around 20GB of disk memory. If your files take up more space than that, your notebook will crash.</li>\n<li>The total inference time allowed is 9 hours, which means that your notebook should be able to predict the 8000 test files within 9 hours time. Converting to WAV and loading takes significant time, leading to crossing the limit. Make sure your notebook can finish inference within the 9-hour limit.<br>\nGood luck!</li>\n</ul>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2454740": "In Submission Process, I convert test dataset from mp3 files to wav file and save to kaggle/working folders (default). So can I read those files from kaggle/working to create TestDataloader ???\n\nNotebook run successfully in example test dataset but \"Notebook Threw Exception\" when run on real test dataset.",
    "2455700": "Hi. There could be two reasons.\n- Kaggle/working has around 20GB of disk memory. If your files take up more space than that, your notebook will crash.\n- The total inference time allowed is 9 hours, which means that your notebook should be able to predict the 8000 test files within 9 hours time. Converting to WAV and loading takes significant time, leading to crossing the limit. Make sure your notebook can finish inference within the 9-hour limit.\nGood luck!"
  },
  "source": "meta"
}