{
  "id": 308217,
  "title": "Better usage of TPU resources?",
  "url": "/competitions/happy-whale-and-dolphin/discussion/308217",
  "author_name": "Paul",
  "post_date": "2022-02-17T17:00:13.977000",
  "votes": 4,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi I'm a newbie here, first competition. I was wondering if it was possible to parallelize tensor flow learning on a single TPU, or to better optimize TPU usage.</p>\n<p>I copied the public ArcFace + EfficientNetv6 notebook and ran it on Kaggle only to realize that it's barely using the whole TPU when training.</p>\n<p>I was wondering if it was possible to get a faster runtime and use the whole TPU at the same time.</p>\n<p>Feel free to point me in the right direction or tell me it's impossible, I'm open to both.</p>\n<p>Cheers</p>",
  "messages": [
    {
      "id": 1694722,
      "postDate": "2022-02-17T17:00:13.977Z",
      "content": "<p>Hi I'm a newbie here, first competition. I was wondering if it was possible to parallelize tensor flow learning on a single TPU, or to better optimize TPU usage.</p>\n<p>I copied the public ArcFace + EfficientNetv6 notebook and ran it on Kaggle only to realize that it's barely using the whole TPU when training.</p>\n<p>I was wondering if it was possible to get a faster runtime and use the whole TPU at the same time.</p>\n<p>Feel free to point me in the right direction or tell me it's impossible, I'm open to both.</p>\n<p>Cheers</p>",
      "rawMarkdown": "Hi I'm a newbie here, first competition. I was wondering if it was possible to parallelize tensor flow learning on a single TPU, or to better optimize TPU usage.\n\nI copied the public ArcFace + EfficientNetv6 notebook and ran it on Kaggle only to realize that it's barely using the whole TPU when training.\n\nI was wondering if it was possible to get a faster runtime and use the whole TPU at the same time.\n\nFeel free to point me in the right direction or tell me it's impossible, I'm open to both.\n\nCheers",
      "votes": 4
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1694722": "Hi I'm a newbie here, first competition. I was wondering if it was possible to parallelize tensor flow learning on a single TPU, or to better optimize TPU usage.\n\nI copied the public ArcFace + EfficientNetv6 notebook and ran it on Kaggle only to realize that it's barely using the whole TPU when training.\n\nI was wondering if it was possible to get a faster runtime and use the whole TPU at the same time.\n\nFeel free to point me in the right direction or tell me it's impossible, I'm open to both.\n\nCheers"
  }
}