{
  "id": 382951,
  "title": "Code competion - pre-trained models",
  "url": "/competitions/rsna-breast-cancer-detection/discussion/382951",
  "author_name": "Alex Olar",
  "post_date": "2023-02-01T16:59:39.645000",
  "votes": 0,
  "comment_count": 1,
  "views": 0,
  "content": "<p>What exactly does 'code competition' mean here? Can we pre-train models on public data, make them public then upload here or that is against the rules? Can we do the same with this dataset on top of additional datasets and just do the inference in the submission?</p>\n<p>I tried to read the FAQ about this, but it isn't clear, that using external data &amp; pre-trained models here could mean what I just described. Thanks in advance.</p>",
  "messages": [
    {
      "id": 2127289,
      "postDate": "2023-02-02T19:37:31.100Z",
      "content": "<p>I think the following quote answers your question about the data:</p>\n<blockquote>\n  <p>You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and <strong>equally accessible to use by all participants of the Competition</strong> for purposes of the competition <strong>at no cost to the other participants</strong>.</p>\n</blockquote>\n<p>If anyone can get the data legally for free, it is permitted. An example would be the DDSM with suitable license.</p>\n<p>And for the open source scripts and models:</p>\n<blockquote>\n  <p>Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.</p>\n</blockquote>\n<p>MIT license, for example, allows commercial use.</p>\n<p>It is good to note that some licenses make commercial use difficult, if not entirely limit the use.</p>",
      "rawMarkdown": "I think the following quote answers your question about the data:\n> You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and **equally accessible to use by all participants of the Competition** for purposes of the competition **at no cost to the other participants**.\n\nIf anyone can get the data legally for free, it is permitted. An example would be the DDSM with suitable license.\n\nAnd for the open source scripts and models:\n> Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.\n\nMIT license, for example, allows commercial use.\n\nIt is good to note that some licenses make commercial use difficult, if not entirely limit the use."
    },
    {
      "id": 2125447,
      "postDate": "2023-02-01T16:59:39.647Z",
      "content": "<p>What exactly does 'code competition' mean here? Can we pre-train models on public data, make them public then upload here or that is against the rules? Can we do the same with this dataset on top of additional datasets and just do the inference in the submission?</p>\n<p>I tried to read the FAQ about this, but it isn't clear, that using external data &amp; pre-trained models here could mean what I just described. Thanks in advance.</p>",
      "rawMarkdown": "What exactly does 'code competition' mean here? Can we pre-train models on public data, make them public then upload here or that is against the rules? Can we do the same with this dataset on top of additional datasets and just do the inference in the submission?\n\nI tried to read the FAQ about this, but it isn't clear, that using external data & pre-trained models here could mean what I just described. Thanks in advance."
    }
  ],
  "comments": [
    {
      "id": 2127289,
      "author_name": "Antti Isosalo",
      "author_url": "",
      "post_date": "2023-02-02T19:37:31.100000",
      "content": "<p>I think the following quote answers your question about the data:</p>\n<blockquote>\n  <p>You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and <strong>equally accessible to use by all participants of the Competition</strong> for purposes of the competition <strong>at no cost to the other participants</strong>.</p>\n</blockquote>\n<p>If anyone can get the data legally for free, it is permitted. An example would be the DDSM with suitable license.</p>\n<p>And for the open source scripts and models:</p>\n<blockquote>\n  <p>Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.</p>\n</blockquote>\n<p>MIT license, for example, allows commercial use.</p>\n<p>It is good to note that some licenses make commercial use difficult, if not entirely limit the use.</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2127289": "I think the following quote answers your question about the data:\n> You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and **equally accessible to use by all participants of the Competition** for purposes of the competition **at no cost to the other participants**.\n\nIf anyone can get the data legally for free, it is permitted. An example would be the DDSM with suitable license.\n\nAnd for the open source scripts and models:\n> Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.\n\nMIT license, for example, allows commercial use.\n\nIt is good to note that some licenses make commercial use difficult, if not entirely limit the use.",
    "2125447": "What exactly does 'code competition' mean here? Can we pre-train models on public data, make them public then upload here or that is against the rules? Can we do the same with this dataset on top of additional datasets and just do the inference in the submission?\n\nI tried to read the FAQ about this, but it isn't clear, that using external data & pre-trained models here could mean what I just described. Thanks in advance."
  }
}