{
  "id": 163841,
  "title": "which is better training on TPU or GPU , Images or TfRecords ? What does the GrandMasters Suggest On this ?",
  "url": "/competitions/alaska2-image-steganalysis/discussion/163841",
  "author_name": "",
  "post_date": "2020-07-03T16:52:08.872864600Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "",
  "messages": [
    {
      "id": "914161",
      "postDate": "07/03/2020 16:52:08",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "914523",
      "postDate": "07/04/2020 02:13:04",
      "content": "<p>if you want to optimize the potentials of TPU, you should use TFRecords as input in model training for any TF/Keras version. Not sure how TFRecords behave on Pytorch</p>",
      "rawMarkdown": "if you want to optimize the potentials of TPU, you should use TFRecords as input in model training for any TF/Keras version. Not sure how TFRecords behave on Pytorch",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 914523,
      "author_name": "projdev",
      "author_url": "",
      "post_date": "07/04/2020 02:13:04",
      "content": "<p>if you want to optimize the potentials of TPU, you should use TFRecords as input in model training for any TF/Keras version. Not sure how TFRecords behave on Pytorch</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "914161": "",
    "914523": "if you want to optimize the potentials of TPU, you should use TFRecords as input in model training for any TF/Keras version. Not sure how TFRecords behave on Pytorch"
  },
  "source": "meta"
}