{
  "id": 14755,
  "title": "Convolutional Neural Network in Python for GPU (code)",
  "url": "/competitions/diabetic-retinopathy-detection/discussion/14755",
  "author_name": "",
  "post_date": "2015-06-16T11:48:33.647Z",
  "votes": 5,
  "comment_count": 1,
  "views": 3535,
  "content": "<p>Hello,</p>\n\n<p>I prepared an image recognition script with an improved deep learning Theano code for a GPU. It is easier to use than the default code from the Theano CNN tutorial. I hope it will make it easier to use Theano with your own flow and datasets!</p>\n\n<p>Link:</p>\n<p>https://www.kaggle.com/lukasz8000/digit-recognizer/test1</p>\n\n<p>The code is the same as in the tutorial, but is extended:</p>\n<p>1. Has separate fit and predict methods</p>\n<p>2. Features export/import of the trained model to a file</p>\n\n<p>Best,</p>\n<p>Lukasz</p>",
  "messages": [
    {
      "id": "81982",
      "postDate": "06/16/2015 11:48:33",
      "content": "<p>Hello,</p>\n\n<p>I prepared an image recognition script with an improved deep learning Theano code for a GPU. It is easier to use than the default code from the Theano CNN tutorial. I hope it will make it easier to use Theano with your own flow and datasets!</p>\n\n<p>Link:</p>\n<p>https://www.kaggle.com/lukasz8000/digit-recognizer/test1</p>\n\n<p>The code is the same as in the tutorial, but is extended:</p>\n<p>1. Has separate fit and predict methods</p>\n<p>2. Features export/import of the trained model to a file</p>\n\n<p>Best,</p>\n<p>Lukasz</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "82333",
      "postDate": "06/18/2015 18:58:38",
      "content": "<p>Thanks!</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 82333,
      "author_name": "adubinsky",
      "author_url": "",
      "post_date": "06/18/2015 18:58:38",
      "content": "<p>Thanks!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "81982": "",
    "82333": ""
  },
  "source": "meta"
}