{
  "id": 188257,
  "title": "Anyone using tensorflow 2.0/Keras for training ? ",
  "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/188257",
  "author_name": "",
  "post_date": "2020-10-02T14:06:50.854250400Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>All public kernels are using pytorch on gpu/tpu. is anyone able to train on TPU using tensorflow, so easy to handle the data as tfrecords. </p>",
  "messages": [
    {
      "id": "1035163",
      "postDate": "10/02/2020 14:06:50",
      "content": "<p>All public kernels are using pytorch on gpu/tpu. is anyone able to train on TPU using tensorflow, so easy to handle the data as tfrecords. </p>",
      "rawMarkdown": "All public kernels are using pytorch on gpu/tpu. is anyone able to train on TPU using tensorflow, so easy to handle the data as tfrecords.",
      "votes": null
    },
    {
      "id": "1035382",
      "postDate": "10/02/2020 17:26:15",
      "content": "<p>I'm using Tensorflow and yes I agree with you. But certaşnly PyTorch has some advantages about distributed and TPU based model training.</p>",
      "rawMarkdown": "I'm using Tensorflow and yes I agree with you. But certaşnly PyTorch has some advantages about distributed and TPU based model training.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1035382,
      "author_name": "feronial",
      "author_url": "",
      "post_date": "10/02/2020 17:26:15",
      "content": "<p>I'm using Tensorflow and yes I agree with you. But certaşnly PyTorch has some advantages about distributed and TPU based model training.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1035163": "All public kernels are using pytorch on gpu/tpu. is anyone able to train on TPU using tensorflow, so easy to handle the data as tfrecords.",
    "1035382": "I'm using Tensorflow and yes I agree with you. But certaşnly PyTorch has some advantages about distributed and TPU based model training."
  },
  "source": "meta"
}