{
  "id": 137110,
  "title": "A fastai sub-competition: (Focused on fastai2)",
  "url": "/competitions/flower-classification-with-tpus/discussion/137110",
  "author_name": "Zach Mueller",
  "post_date": "2020-03-19T06:58:33.136000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Let me know if this is not allowed but for those of us in the new fastai course, we want to use this competition as a way to show how flexible the API can be, and try to bring in new ideas into the new framework. A few ground rules for our participants include:</p>\n\n<ol>\n<li>You are <strong>not</strong> allowed to enter the actual competition with the code written</li>\n<li>Your solutions should be on the fastai2 forum megathread (and here if it is asked)</li>\n<li>For those who do not have access to those forums yet, please do not ask for it, they will be available after the competition is over. </li>\n</ol>\n\n<p>Here is a gist I've made introducing the v2 framework for this competition:</p>\n\n<p><a href=\"https://gist.github.com/muellerzr/4fceb89f4fcfeb374673a11bb305d263\">https://gist.github.com/muellerzr/4fceb89f4fcfeb374673a11bb305d263</a></p>\n\n<p><strong>NOTE</strong>: <code>fastai2</code> does <strong>not</strong> support TPU's at the moment, so we are utilizing the GPU and PyTorch (back-end) instead of TPU and TensorFlow so keep that in mind. Thanks!</p>\n\n<p>I'll update this with a link to the thread soon</p>",
  "messages": [
    {
      "id": 779279,
      "postDate": "2020-03-19T06:58:33.137Z",
      "content": "<p>Let me know if this is not allowed but for those of us in the new fastai course, we want to use this competition as a way to show how flexible the API can be, and try to bring in new ideas into the new framework. A few ground rules for our participants include:</p>\n\n<ol>\n<li>You are <strong>not</strong> allowed to enter the actual competition with the code written</li>\n<li>Your solutions should be on the fastai2 forum megathread (and here if it is asked)</li>\n<li>For those who do not have access to those forums yet, please do not ask for it, they will be available after the competition is over. </li>\n</ol>\n\n<p>Here is a gist I've made introducing the v2 framework for this competition:</p>\n\n<p><a href=\"https://gist.github.com/muellerzr/4fceb89f4fcfeb374673a11bb305d263\">https://gist.github.com/muellerzr/4fceb89f4fcfeb374673a11bb305d263</a></p>\n\n<p><strong>NOTE</strong>: <code>fastai2</code> does <strong>not</strong> support TPU's at the moment, so we are utilizing the GPU and PyTorch (back-end) instead of TPU and TensorFlow so keep that in mind. Thanks!</p>\n\n<p>I'll update this with a link to the thread soon</p>",
      "rawMarkdown": "Let me know if this is not allowed but for those of us in the new fastai course, we want to use this competition as a way to show how flexible the API can be, and try to bring in new ideas into the new framework. A few ground rules for our participants include:\n\n1. You are **not** allowed to enter the actual competition with the code written\n2. Your solutions should be on the fastai2 forum megathread (and here if it is asked)\n3. For those who do not have access to those forums yet, please do not ask for it, they will be available after the competition is over. \n\nHere is a gist I've made introducing the v2 framework for this competition:\n\nhttps://gist.github.com/muellerzr/4fceb89f4fcfeb374673a11bb305d263\n\n**NOTE**: `fastai2` does **not** support TPU's at the moment, so we are utilizing the GPU and PyTorch (back-end) instead of TPU and TensorFlow so keep that in mind. Thanks!\n\nI'll update this with a link to the thread soon",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "779279": "Let me know if this is not allowed but for those of us in the new fastai course, we want to use this competition as a way to show how flexible the API can be, and try to bring in new ideas into the new framework. A few ground rules for our participants include:\n\n1. You are **not** allowed to enter the actual competition with the code written\n2. Your solutions should be on the fastai2 forum megathread (and here if it is asked)\n3. For those who do not have access to those forums yet, please do not ask for it, they will be available after the competition is over. \n\nHere is a gist I've made introducing the v2 framework for this competition:\n\nhttps://gist.github.com/muellerzr/4fceb89f4fcfeb374673a11bb305d263\n\n**NOTE**: `fastai2` does **not** support TPU's at the moment, so we are utilizing the GPU and PyTorch (back-end) instead of TPU and TensorFlow so keep that in mind. Thanks!\n\nI'll update this with a link to the thread soon"
  }
}