{
  "id": 93054,
  "title": "Winning teams releasing the code for reproducibility",
  "url": "/competitions/freesound-audio-tagging-2019/discussion/93054",
  "author_name": "Eduardo Fonseca",
  "post_date": "2019-05-22T19:51:42.177000",
  "votes": 6,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Dear all, </p>\n\n<p>as per the <a href=\"https://www.kaggle.com/c/freesound-audio-tagging-2019/discussion/88064\">Rules</a>, winning teams are required to publish their systems under an open-source license in order to be considered winners. In this Challenge, there are <a href=\"https://www.kaggle.com/c/freesound-audio-tagging-2019/overview/prizes\">four winners</a>: three according to the private Leaderboard, and one additional winner of the Judges' Award.</p>\n\n<p>The goal of open-sourcing code is to generate open knowledge and that the community is able to reproduce the results and better understand the systems. Therefore, ideally, winners must share the full framework used for the winning submission. This includes, for example, feature extraction, training and prediction (not just a pre-trained model + inference code), and also installation and running instructions.</p>\n\n<p>Some examples of this are:\n<a href=\"https://github.com/DCASE-REPO/dcase2019_task2_baseline\">https://github.com/DCASE-REPO/dcase2019_task2_baseline</a>\n<a href=\"https://github.com/edufonseca/icassp19\">https://github.com/edufonseca/icassp19</a></p>\n\n<p>However, we understand that this is not always easy due to dependencies on earlier models or setups, etc. Therefore, we trust participants' good will in this respect and we hope that all winners will share the best they can.</p>\n\n<p>In addition, DCASE organizers encourage not only the winners but all participants to open-source their code after the challenge.</p>\n\n<p>Eduardo\n(on behalf of all the challenge organizers)</p>",
  "messages": [
    {
      "id": 535382,
      "postDate": "2019-05-22T19:51:42.177Z",
      "content": "<p>Dear all, </p>\n\n<p>as per the <a href=\"https://www.kaggle.com/c/freesound-audio-tagging-2019/discussion/88064\">Rules</a>, winning teams are required to publish their systems under an open-source license in order to be considered winners. In this Challenge, there are <a href=\"https://www.kaggle.com/c/freesound-audio-tagging-2019/overview/prizes\">four winners</a>: three according to the private Leaderboard, and one additional winner of the Judges' Award.</p>\n\n<p>The goal of open-sourcing code is to generate open knowledge and that the community is able to reproduce the results and better understand the systems. Therefore, ideally, winners must share the full framework used for the winning submission. This includes, for example, feature extraction, training and prediction (not just a pre-trained model + inference code), and also installation and running instructions.</p>\n\n<p>Some examples of this are:\n<a href=\"https://github.com/DCASE-REPO/dcase2019_task2_baseline\">https://github.com/DCASE-REPO/dcase2019_task2_baseline</a>\n<a href=\"https://github.com/edufonseca/icassp19\">https://github.com/edufonseca/icassp19</a></p>\n\n<p>However, we understand that this is not always easy due to dependencies on earlier models or setups, etc. Therefore, we trust participants' good will in this respect and we hope that all winners will share the best they can.</p>\n\n<p>In addition, DCASE organizers encourage not only the winners but all participants to open-source their code after the challenge.</p>\n\n<p>Eduardo\n(on behalf of all the challenge organizers)</p>",
      "rawMarkdown": "Dear all, \n\nas per the [Rules](https://www.kaggle.com/c/freesound-audio-tagging-2019/discussion/88064), winning teams are required to publish their systems under an open-source license in order to be considered winners. In this Challenge, there are [four winners](https://www.kaggle.com/c/freesound-audio-tagging-2019/overview/prizes): three according to the private Leaderboard, and one additional winner of the Judges' Award.\n\nThe goal of open-sourcing code is to generate open knowledge and that the community is able to reproduce the results and better understand the systems. Therefore, ideally, winners must share the full framework used for the winning submission. This includes, for example, feature extraction, training and prediction (not just a pre-trained model + inference code), and also installation and running instructions.\n\nSome examples of this are:\n[https://github.com/DCASE-REPO/dcase2019_task2_baseline](https://github.com/DCASE-REPO/dcase2019_task2_baseline)\n[https://github.com/edufonseca/icassp19](https://github.com/edufonseca/icassp19)\n\nHowever, we understand that this is not always easy due to dependencies on earlier models or setups, etc. Therefore, we trust participants' good will in this respect and we hope that all winners will share the best they can.\n\nIn addition, DCASE organizers encourage not only the winners but all participants to open-source their code after the challenge.\n\nEduardo\n(on behalf of all the challenge organizers)\n\n\n",
      "votes": 6
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "535382": "Dear all, \n\nas per the [Rules](https://www.kaggle.com/c/freesound-audio-tagging-2019/discussion/88064), winning teams are required to publish their systems under an open-source license in order to be considered winners. In this Challenge, there are [four winners](https://www.kaggle.com/c/freesound-audio-tagging-2019/overview/prizes): three according to the private Leaderboard, and one additional winner of the Judges' Award.\n\nThe goal of open-sourcing code is to generate open knowledge and that the community is able to reproduce the results and better understand the systems. Therefore, ideally, winners must share the full framework used for the winning submission. This includes, for example, feature extraction, training and prediction (not just a pre-trained model + inference code), and also installation and running instructions.\n\nSome examples of this are:\n[https://github.com/DCASE-REPO/dcase2019_task2_baseline](https://github.com/DCASE-REPO/dcase2019_task2_baseline)\n[https://github.com/edufonseca/icassp19](https://github.com/edufonseca/icassp19)\n\nHowever, we understand that this is not always easy due to dependencies on earlier models or setups, etc. Therefore, we trust participants' good will in this respect and we hope that all winners will share the best they can.\n\nIn addition, DCASE organizers encourage not only the winners but all participants to open-source their code after the challenge.\n\nEduardo\n(on behalf of all the challenge organizers)\n\n\n"
  }
}