{
  "id": 172380,
  "title": "How can we use TPU/GPU to run our codes, using tensorflow/keras in Kaggle?",
  "url": "/competitions/global-wheat-detection/discussion/172380",
  "author_name": "",
  "post_date": "2020-08-04T20:50:28.971804Z",
  "votes": -2,
  "comment_count": 2,
  "views": 0,
  "content": "",
  "messages": [
    {
      "id": "958202",
      "postDate": "08/04/2020 20:50:28",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "958206",
      "postDate": "08/04/2020 20:57:10",
      "content": "<p>You use them by selecting either GPU or TPU from the Accelerator dropdown on the right-hand side toolbar within the notebook editor. </p>\n\n<p>For training, you can use either (assuming your framework supports TPU). However for inference, you can only use GPU.</p>",
      "rawMarkdown": "You use them by selecting either GPU or TPU from the Accelerator dropdown on the right-hand side toolbar within the notebook editor. \n\nFor training, you can use either (assuming your framework supports TPU). However for inference, you can only use GPU.",
      "votes": null
    },
    {
      "id": "958217",
      "postDate": "08/04/2020 21:09:56",
      "content": "<p>It's also worth noting that to run your inference notebook using a GPU, you need to select GPU in the kernel you are submitting with before you click commit. </p>",
      "rawMarkdown": "It's also worth noting that to run your inference notebook using a GPU, you need to select GPU in the kernel you are submitting with before you click commit.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 958206,
      "author_name": "matthewmasters",
      "author_url": "",
      "post_date": "08/04/2020 20:57:10",
      "content": "<p>You use them by selecting either GPU or TPU from the Accelerator dropdown on the right-hand side toolbar within the notebook editor. </p>\n\n<p>For training, you can use either (assuming your framework supports TPU). However for inference, you can only use GPU.</p>",
      "votes": null,
      "replies": [
        {
          "id": 958217,
          "author_name": "stanleyjzheng",
          "author_url": "",
          "post_date": "08/04/2020 21:09:56",
          "content": "<p>It's also worth noting that to run your inference notebook using a GPU, you need to select GPU in the kernel you are submitting with before you click commit. </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "958202": "",
    "958206": "You use them by selecting either GPU or TPU from the Accelerator dropdown on the right-hand side toolbar within the notebook editor. \n\nFor training, you can use either (assuming your framework supports TPU). However for inference, you can only use GPU.",
    "958217": "It's also worth noting that to run your inference notebook using a GPU, you need to select GPU in the kernel you are submitting with before you click commit."
  },
  "source": "meta"
}