{
  "id": 328065,
  "title": "Say, we nailed the prediction. What's next?",
  "url": "/competitions/AI4Code/discussion/328065",
  "author_name": "",
  "post_date": "2022-05-30T18:03:26.824004Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>A brainstorming thread on what a really well trained model could help with down the line and how we would structure that. Looking for more ideas / interpretations</p>\n<blockquote>\n  <p>An understanding of the relationships between code and markdown could lend to fresh improvements across many aspects of AI-assisted development, such as<br>\n  the construction of better data filtering and preprocessing pipelines for model training</p>\n</blockquote>\n<p>Reduce dimensionality, clusterize &amp; attempt interpreting the cluster / membership:<br>\n– We observe cluster X (which looks like logging) in top 5 by size and enjoys 50% smaller intra-cluster distance --&gt; we could look into providing a logging primitive capturing a popular use case in our next version of the API</p>\n<blockquote>\n  <p>or automatic assessments of a notebook's readability.</p>\n</blockquote>\n<p>Cluster/classify blocks + regression/classification on top:<br>\n– \"These 2 screens of code usually enjoy 50-200 words of comments but your notebook has only 0\"</p>\n<hr>\n<p>Comment generation:<br>\n– Classification / ranking / seq generation: \"Comment of '# TODO: I have no idea why this works' usually accompanies blocks of code similar to this. Would you like to add? (Y/N)\"</p>",
  "messages": [
    {
      "id": "1806033",
      "postDate": "05/30/2022 18:03:26",
      "content": "<p>A brainstorming thread on what a really well trained model could help with down the line and how we would structure that. Looking for more ideas / interpretations</p>\n<blockquote>\n  <p>An understanding of the relationships between code and markdown could lend to fresh improvements across many aspects of AI-assisted development, such as<br>\n  the construction of better data filtering and preprocessing pipelines for model training</p>\n</blockquote>\n<p>Reduce dimensionality, clusterize &amp; attempt interpreting the cluster / membership:<br>\n– We observe cluster X (which looks like logging) in top 5 by size and enjoys 50% smaller intra-cluster distance --&gt; we could look into providing a logging primitive capturing a popular use case in our next version of the API</p>\n<blockquote>\n  <p>or automatic assessments of a notebook's readability.</p>\n</blockquote>\n<p>Cluster/classify blocks + regression/classification on top:<br>\n– \"These 2 screens of code usually enjoy 50-200 words of comments but your notebook has only 0\"</p>\n<hr>\n<p>Comment generation:<br>\n– Classification / ranking / seq generation: \"Comment of '# TODO: I have no idea why this works' usually accompanies blocks of code similar to this. Would you like to add? (Y/N)\"</p>",
      "rawMarkdown": "A brainstorming thread on what a really well trained model could help with down the line and how we would structure that. Looking for more ideas / interpretations\n\n> An understanding of the relationships between code and markdown could lend to fresh improvements across many aspects of AI-assisted development, such as\n> the construction of better data filtering and preprocessing pipelines for model training\n\nReduce dimensionality, clusterize & attempt interpreting the cluster / membership:\n– We observe cluster X (which looks like logging) in top 5 by size and enjoys 50% smaller intra-cluster distance --> we could look into providing a logging primitive capturing a popular use case in our next version of the API\n\n> or automatic assessments of a notebook's readability.\n\nCluster/classify blocks + regression/classification on top:\n– \"These 2 screens of code usually enjoy 50-200 words of comments but your notebook has only 0\"\n\n---\n\nComment generation:\n– Classification / ranking / seq generation: \"Comment of '# TODO: I have no idea why this works' usually accompanies blocks of code similar to this. Would you like to add? (Y/N)\"",
      "votes": null
    },
    {
      "id": "1809484",
      "postDate": "06/02/2022 17:48:17",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/soboleiv\" target=\"_blank\">@soboleiv</a> </p>\n<p>There are some interesting ideas discussed in the below forum post around the same topic. Please check it out.<br>\n<a href=\"https://www.kaggle.com/competitions/AI4Code/discussion/325205\" target=\"_blank\">https://www.kaggle.com/competitions/AI4Code/discussion/325205</a></p>\n<p>Thank you!</p>",
      "rawMarkdown": "Hi @soboleiv \n\nThere are some interesting ideas discussed in the below forum post around the same topic. Please check it out.\nhttps://www.kaggle.com/competitions/AI4Code/discussion/325205\n\nThank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1809484,
      "author_name": "sudalairajkumar",
      "author_url": "",
      "post_date": "06/02/2022 17:48:17",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/soboleiv\" target=\"_blank\">@soboleiv</a> </p>\n<p>There are some interesting ideas discussed in the below forum post around the same topic. Please check it out.<br>\n<a href=\"https://www.kaggle.com/competitions/AI4Code/discussion/325205\" target=\"_blank\">https://www.kaggle.com/competitions/AI4Code/discussion/325205</a></p>\n<p>Thank you!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1806033": "A brainstorming thread on what a really well trained model could help with down the line and how we would structure that. Looking for more ideas / interpretations\n\n> An understanding of the relationships between code and markdown could lend to fresh improvements across many aspects of AI-assisted development, such as\n> the construction of better data filtering and preprocessing pipelines for model training\n\nReduce dimensionality, clusterize & attempt interpreting the cluster / membership:\n– We observe cluster X (which looks like logging) in top 5 by size and enjoys 50% smaller intra-cluster distance --> we could look into providing a logging primitive capturing a popular use case in our next version of the API\n\n> or automatic assessments of a notebook's readability.\n\nCluster/classify blocks + regression/classification on top:\n– \"These 2 screens of code usually enjoy 50-200 words of comments but your notebook has only 0\"\n\n---\n\nComment generation:\n– Classification / ranking / seq generation: \"Comment of '# TODO: I have no idea why this works' usually accompanies blocks of code similar to this. Would you like to add? (Y/N)\"",
    "1809484": "Hi @soboleiv \n\nThere are some interesting ideas discussed in the below forum post around the same topic. Please check it out.\nhttps://www.kaggle.com/competitions/AI4Code/discussion/325205\n\nThank you!"
  },
  "source": "meta"
}