{
  "id": 546920,
  "title": "How many date_ids to use for training?",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/546920",
  "author_name": "",
  "post_date": "2024-11-18T20:18:05.779110Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Since there are about 1700 date_ids in dataset, may I ask you how many date_ids you used for training? And what is the performance of CV and LB?</p>",
  "messages": [
    {
      "id": "3049197",
      "postDate": "11/18/2024 20:18:05",
      "content": "<p>Since there are about 1700 date_ids in dataset, may I ask you how many date_ids you used for training? And what is the performance of CV and LB?</p>",
      "rawMarkdown": "Since there are about 1700 date_ids in dataset, may I ask you how many date_ids you used for training? And what is the performance of CV and LB?",
      "votes": null
    },
    {
      "id": "3049389",
      "postDate": "11/19/2024 04:01:15",
      "content": "<p>All of them.</p>",
      "rawMarkdown": "All of them.",
      "votes": null
    },
    {
      "id": "3049821",
      "postDate": "11/19/2024 14:44:14",
      "content": "<p>Most people are training starting after day 500 or so because of the high miss rate before that. I'm personally using partitions 5-9 but mostly due to memory issues, I intend to use other partitions in the future</p>",
      "rawMarkdown": "Most people are training starting after day 500 or so because of the high miss rate before that. I'm personally using partitions 5-9 but mostly due to memory issues, I intend to use other partitions in the future",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3049389,
      "author_name": "gunesevitan",
      "author_url": "",
      "post_date": "11/19/2024 04:01:15",
      "content": "<p>All of them.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3049821,
      "author_name": "natanlabarrere",
      "author_url": "",
      "post_date": "11/19/2024 14:44:14",
      "content": "<p>Most people are training starting after day 500 or so because of the high miss rate before that. I'm personally using partitions 5-9 but mostly due to memory issues, I intend to use other partitions in the future</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3049197": "Since there are about 1700 date_ids in dataset, may I ask you how many date_ids you used for training? And what is the performance of CV and LB?",
    "3049389": "All of them.",
    "3049821": "Most people are training starting after day 500 or so because of the high miss rate before that. I'm personally using partitions 5-9 but mostly due to memory issues, I intend to use other partitions in the future"
  },
  "source": "meta"
}