{
  "id": 59791,
  "title": "Move dataset directly into GoogleCloud or AWS S3 buckets?",
  "url": "/competitions/imagenet-object-localization-challenge/discussion/59791",
  "author_name": "",
  "post_date": "2018-06-27T06:15:18.784230300Z",
  "votes": 21,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Is the Imagenet data already available in a readable Google Cloud or S3 bucket? It would be a lot easier than trying to download this to a personal computer via bandwidth limited internet. </p>",
  "messages": [
    {
      "id": "348700",
      "postDate": "06/27/2018 06:15:18",
      "content": "<p>Is the Imagenet data already available in a readable Google Cloud or S3 bucket? It would be a lot easier than trying to download this to a personal computer via bandwidth limited internet. </p>",
      "rawMarkdown": "Is the Imagenet data already available in a readable Google Cloud or S3 bucket? It would be a lot easier than trying to download this to a personal computer via bandwidth limited internet.",
      "votes": null
    },
    {
      "id": "446252",
      "postDate": "12/27/2018 19:02:12",
      "content": "<p>Bump</p>",
      "rawMarkdown": "Bump",
      "votes": null
    },
    {
      "id": "463967",
      "postDate": "01/31/2019 01:09:41",
      "content": "<p>I work for Google Cloud, documenting Cloud TPU for ML. We have a set of tutorials that all use the Imagenet dataset. Since it takes over 20 hours do download the dataset from Imagenet, we have a major barrier in enabling users to run the tutorials against the full Imagenet dataset. We would be very, very, happy to host a non-username/passwd protected mirror on the Google Cloud for this dataset. Please let us know if this is something we can work with Kaggle on. Thank you.</p>",
      "rawMarkdown": "I work for Google Cloud, documenting Cloud TPU for ML. We have a set of tutorials that all use the Imagenet dataset. Since it takes over 20 hours do download the dataset from Imagenet, we have a major barrier in enabling users to run the tutorials against the full Imagenet dataset. We would be very, very, happy to host a non-username/passwd protected mirror on the Google Cloud for this dataset. Please let us know if this is something we can work with Kaggle on. Thank you.",
      "votes": null
    },
    {
      "id": "481170",
      "postDate": "03/01/2019 05:34:01",
      "content": "<p>I think this would be really useful. Then I could try. Is there a way we could do this?</p>",
      "rawMarkdown": "I think this would be really useful. Then I could try. Is there a way we could do this?",
      "votes": null
    },
    {
      "id": "486699",
      "postDate": "03/09/2019 08:50:17",
      "content": "<p>Upvoted for better visibility.</p>",
      "rawMarkdown": "Upvoted for better visibility.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 446252,
      "author_name": "bflyth",
      "author_url": "",
      "post_date": "12/27/2018 19:02:12",
      "content": "<p>Bump</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 463967,
      "author_name": "gmadrone",
      "author_url": "",
      "post_date": "01/31/2019 01:09:41",
      "content": "<p>I work for Google Cloud, documenting Cloud TPU for ML. We have a set of tutorials that all use the Imagenet dataset. Since it takes over 20 hours do download the dataset from Imagenet, we have a major barrier in enabling users to run the tutorials against the full Imagenet dataset. We would be very, very, happy to host a non-username/passwd protected mirror on the Google Cloud for this dataset. Please let us know if this is something we can work with Kaggle on. Thank you.</p>",
      "votes": null,
      "replies": [
        {
          "id": 486699,
          "author_name": "jarmos",
          "author_url": "",
          "post_date": "03/09/2019 08:50:17",
          "content": "<p>Upvoted for better visibility.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 481170,
      "author_name": "chazie9",
      "author_url": "",
      "post_date": "03/01/2019 05:34:01",
      "content": "<p>I think this would be really useful. Then I could try. Is there a way we could do this?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "348700": "Is the Imagenet data already available in a readable Google Cloud or S3 bucket? It would be a lot easier than trying to download this to a personal computer via bandwidth limited internet.",
    "446252": "Bump",
    "463967": "I work for Google Cloud, documenting Cloud TPU for ML. We have a set of tutorials that all use the Imagenet dataset. Since it takes over 20 hours do download the dataset from Imagenet, we have a major barrier in enabling users to run the tutorials against the full Imagenet dataset. We would be very, very, happy to host a non-username/passwd protected mirror on the Google Cloud for this dataset. Please let us know if this is something we can work with Kaggle on. Thank you.",
    "481170": "I think this would be really useful. Then I could try. Is there a way we could do this?",
    "486699": "Upvoted for better visibility."
  },
  "source": "meta"
}