{
  "id": 44327,
  "title": "Can I use Custom Operation on LasberryPi for Special Prize?",
  "url": "/competitions/tensorflow-speech-recognition-challenge/discussion/44327",
  "author_name": "",
  "post_date": "2017-11-27T07:09:02.506231800Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>For special prize, It says 'Be runnable as frozen TensorFlow GraphDef files with no additional dependencies beyond TensorFlow 1.4.'.</p>\n\n<p>So.. Can i use a custom operation? It has one cc file and should be built before running.</p>",
  "messages": [
    {
      "id": "248881",
      "postDate": "11/27/2017 07:09:02",
      "content": "<p>For special prize, It says 'Be runnable as frozen TensorFlow GraphDef files with no additional dependencies beyond TensorFlow 1.4.'.</p>\n\n<p>So.. Can i use a custom operation? It has one cc file and should be built before running.</p>",
      "rawMarkdown": "For special prize, It says 'Be runnable as frozen TensorFlow GraphDef files with no additional dependencies beyond TensorFlow 1.4.'.\n\nSo.. Can i use a custom operation? It has one cc file and should be built before running.",
      "votes": null
    },
    {
      "id": "253940",
      "postDate": "12/05/2017 22:09:47",
      "content": "<blockquote>\n  <p>Can i use a custom operation?</p>\n</blockquote>\n\n<p>Unfortunately not, since that would make it hard to distribute as a TensorFlow example (which was my goal when drawing up the special prize rules). If people did <code>pip</code> installs, they wouldn't be able to run it for instance.</p>\n\n<p>I might be able to offer some advice on expressing the operation using standard TF operators though, if you are able to share what it's doing?</p>",
      "rawMarkdown": "&gt; Can i use a custom operation?\n\nUnfortunately not, since that would make it hard to distribute as a TensorFlow example (which was my goal when drawing up the special prize rules). If people did `pip` installs, they wouldn't be able to run it for instance.\n\nI might be able to offer some advice on expressing the operation using standard TF operators though, if you are able to share what it's doing?",
      "votes": null
    },
    {
      "id": "254003",
      "postDate": "12/06/2017 02:42:23",
      "content": "<p>Thanks.</p>\n\n<p>I was thinking that LCNN can be very helpful.</p>\n\n<p>There is a repository that I implemented. : <a href=\"https://github.com/ildoonet/tf-lcnn\">https://github.com/ildoonet/tf-lcnn</a></p>\n\n<p>I will look around other alternatives. Thanks again.</p>",
      "rawMarkdown": "Thanks.\n\nI was thinking that LCNN can be very helpful.\n\nThere is a repository that I implemented. : https://github.com/ildoonet/tf-lcnn\n\nI will look around other alternatives. Thanks again.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 253940,
      "author_name": "petewarden",
      "author_url": "",
      "post_date": "12/05/2017 22:09:47",
      "content": "<blockquote>\n  <p>Can i use a custom operation?</p>\n</blockquote>\n\n<p>Unfortunately not, since that would make it hard to distribute as a TensorFlow example (which was my goal when drawing up the special prize rules). If people did <code>pip</code> installs, they wouldn't be able to run it for instance.</p>\n\n<p>I might be able to offer some advice on expressing the operation using standard TF operators though, if you are able to share what it's doing?</p>",
      "votes": null,
      "replies": [
        {
          "id": 254003,
          "author_name": "ildoonet",
          "author_url": "",
          "post_date": "12/06/2017 02:42:23",
          "content": "<p>Thanks.</p>\n\n<p>I was thinking that LCNN can be very helpful.</p>\n\n<p>There is a repository that I implemented. : <a href=\"https://github.com/ildoonet/tf-lcnn\">https://github.com/ildoonet/tf-lcnn</a></p>\n\n<p>I will look around other alternatives. Thanks again.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "248881": "For special prize, It says 'Be runnable as frozen TensorFlow GraphDef files with no additional dependencies beyond TensorFlow 1.4.'.\n\nSo.. Can i use a custom operation? It has one cc file and should be built before running.",
    "253940": "&gt; Can i use a custom operation?\n\nUnfortunately not, since that would make it hard to distribute as a TensorFlow example (which was my goal when drawing up the special prize rules). If people did `pip` installs, they wouldn't be able to run it for instance.\n\nI might be able to offer some advice on expressing the operation using standard TF operators though, if you are able to share what it's doing?",
    "254003": "Thanks.\n\nI was thinking that LCNN can be very helpful.\n\nThere is a repository that I implemented. : https://github.com/ildoonet/tf-lcnn\n\nI will look around other alternatives. Thanks again."
  },
  "source": "meta"
}