{
  "id": 237012,
  "title": "Memory issues while extracting data spectrograms",
  "url": "/competitions/birdclef-2021/discussion/237012",
  "author_name": "",
  "post_date": "2021-05-06T17:24:32.577643Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hello,<br>\nDoes anyone have ideas as to how to manage memory for a large number of spectrograms.<br>\nThe easy approach I can use is have only a few species in training data, or a few examples per species, but I would like to know if  anyone has found thier way around this problem.</p>",
  "messages": [
    {
      "id": "1295761",
      "postDate": "05/06/2021 17:24:32",
      "content": "<p>Hello,<br>\nDoes anyone have ideas as to how to manage memory for a large number of spectrograms.<br>\nThe easy approach I can use is have only a few species in training data, or a few examples per species, but I would like to know if  anyone has found thier way around this problem.</p>",
      "rawMarkdown": "Hello,\nDoes anyone have ideas as to how to manage memory for a large number of spectrograms.\nThe easy approach I can use is have only a few species in training data, or a few examples per species, but I would like to know if  anyone has found thier way around this problem.",
      "votes": null
    },
    {
      "id": "1297188",
      "postDate": "05/07/2021 19:23:25",
      "content": "<p>You can save images to a working folder and then use a generator to load images when training/testing!</p>",
      "rawMarkdown": "You can save images to a working folder and then use a generator to load images when training/testing!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1297188,
      "author_name": "aarons0",
      "author_url": "",
      "post_date": "05/07/2021 19:23:25",
      "content": "<p>You can save images to a working folder and then use a generator to load images when training/testing!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1295761": "Hello,\nDoes anyone have ideas as to how to manage memory for a large number of spectrograms.\nThe easy approach I can use is have only a few species in training data, or a few examples per species, but I would like to know if  anyone has found thier way around this problem.",
    "1297188": "You can save images to a working folder and then use a generator to load images when training/testing!"
  },
  "source": "meta"
}