{
  "id": 167596,
  "title": "Shape of train and test data",
  "url": "/competitions/tpu-getting-started/discussion/167596",
  "author_name": "Bhagyesh Agresar",
  "post_date": "2020-07-17T08:04:25.826000",
  "votes": 0,
  "comment_count": 2,
  "views": null,
  "content": "<p>in the exercise notebook, when running the code for checking the shape of train data, the output is:\n(128, 512, 512, 3), (128, )</p>\n\n<p>what is 128?</p>",
  "messages": [
    {
      "id": 932740,
      "postDate": "2020-07-17T08:47:04.993Z",
      "content": "<p>It is the batch, i.e. 128 observations stacked into one tensor. (128, ) means 128 target values stacked together.</p>",
      "rawMarkdown": "It is the batch, i.e. 128 observations stacked into one tensor. (128, ) means 128 target values stacked together.",
      "replies": [
        {
          "id": 932771,
          "postDate": "2020-07-17T09:12:37.383Z",
          "content": "<p>But there are 12753 training images in total and we generally define batch_size while training the model, how come the dataset is already put into batches before setting the batch_size = 16 * 8 ?</p>",
          "rawMarkdown": "But there are 12753 training images in total and we generally define batch_size while training the model, how come the dataset is already put into batches before setting the batch_size = 16 * 8 ?"
        }
      ]
    },
    {
      "id": 932698,
      "postDate": "2020-07-17T08:04:25.827Z",
      "content": "<p>in the exercise notebook, when running the code for checking the shape of train data, the output is:\n(128, 512, 512, 3), (128, )</p>\n\n<p>what is 128?</p>",
      "rawMarkdown": "in the exercise notebook, when running the code for checking the shape of train data, the output is:\n(128, 512, 512, 3), (128, )\n\nwhat is 128?"
    }
  ],
  "comments": [
    {
      "id": 932740,
      "author_name": "Mike Leske",
      "author_url": "",
      "post_date": "2020-07-17T08:47:04.993000",
      "content": "<p>It is the batch, i.e. 128 observations stacked into one tensor. (128, ) means 128 target values stacked together.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 932771,
          "author_name": "Bhagyesh Agresar",
          "author_url": "",
          "post_date": "2020-07-17T09:12:37.383000",
          "content": "<p>But there are 12753 training images in total and we generally define batch_size while training the model, how come the dataset is already put into batches before setting the batch_size = 16 * 8 ?</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "932740": "It is the batch, i.e. 128 observations stacked into one tensor. (128, ) means 128 target values stacked together.",
    "932698": "in the exercise notebook, when running the code for checking the shape of train data, the output is:\n(128, 512, 512, 3), (128, )\n\nwhat is 128?"
  }
}