{
  "id": 231906,
  "title": "Tips for making the kaggle notebook usable?",
  "url": "/competitions/birdclef-2021/discussion/231906",
  "author_name": "",
  "post_date": "2021-04-11T00:47:42.772307400Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I don't know if I'm alone in this, but I always have a lot of issues when using the kernel notebook. Mainly it is extremely slow and freezes. I initially thought it was the model size, but trimming it down to a single convolution and a fully connected layer trains at the same speed?</p>",
  "messages": [
    {
      "id": "1269832",
      "postDate": "04/11/2021 00:47:42",
      "content": "<p>I don't know if I'm alone in this, but I always have a lot of issues when using the kernel notebook. Mainly it is extremely slow and freezes. I initially thought it was the model size, but trimming it down to a single convolution and a fully connected layer trains at the same speed?</p>",
      "rawMarkdown": "I don't know if I'm alone in this, but I always have a lot of issues when using the kernel notebook. Mainly it is extremely slow and freezes. I initially thought it was the model size, but trimming it down to a single convolution and a fully connected layer trains at the same speed?",
      "votes": null
    },
    {
      "id": "1270287",
      "postDate": "04/11/2021 13:45:12",
      "content": "<p>You should tun GPU or TPU accelerator on depending on what code you use.  Look at the bottom right of your screen when editing notebooks for the toggle.  GPU maybe easier to start with (disclaimer: I work at NVIDIA hence I may be biased towards GPU).</p>",
      "rawMarkdown": "You should tun GPU or TPU accelerator on depending on what code you use.  Look at the bottom right of your screen when editing notebooks for the toggle.  GPU maybe easier to start with (disclaimer: I work at NVIDIA hence I may be biased towards GPU).",
      "votes": null
    },
    {
      "id": "1270344",
      "postDate": "04/11/2021 14:50:38",
      "content": "<p>yep. it's not running on GPU. thank you!</p>",
      "rawMarkdown": "yep. it's not running on GPU. thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1270287,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "04/11/2021 13:45:12",
      "content": "<p>You should tun GPU or TPU accelerator on depending on what code you use.  Look at the bottom right of your screen when editing notebooks for the toggle.  GPU maybe easier to start with (disclaimer: I work at NVIDIA hence I may be biased towards GPU).</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1270344,
      "author_name": "willrice",
      "author_url": "",
      "post_date": "04/11/2021 14:50:38",
      "content": "<p>yep. it's not running on GPU. thank you!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1269832": "I don't know if I'm alone in this, but I always have a lot of issues when using the kernel notebook. Mainly it is extremely slow and freezes. I initially thought it was the model size, but trimming it down to a single convolution and a fully connected layer trains at the same speed?",
    "1270287": "You should tun GPU or TPU accelerator on depending on what code you use.  Look at the bottom right of your screen when editing notebooks for the toggle.  GPU maybe easier to start with (disclaimer: I work at NVIDIA hence I may be biased towards GPU).",
    "1270344": "yep. it's not running on GPU. thank you!"
  },
  "source": "meta"
}