{
  "id": 383258,
  "title": "Format test set",
  "url": "/competitions/icecube-neutrinos-in-deep-ice/discussion/383258",
  "author_name": "",
  "post_date": "2023-02-02T21:56:47.191651600Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Can we assume each test .parquet file contains 200K events (ie rows) in the test meta file ? <br>\nOr should we take into consideration batches of different sizes ? (the proper manner to code, but I am lazy ahah)<br>\nThanks ! </p>",
  "messages": [
    {
      "id": "2127431",
      "postDate": "02/02/2023 21:56:47",
      "content": "<p>Can we assume each test .parquet file contains 200K events (ie rows) in the test meta file ? <br>\nOr should we take into consideration batches of different sizes ? (the proper manner to code, but I am lazy ahah)<br>\nThanks ! </p>",
      "rawMarkdown": "Can we assume each test .parquet file contains 200K events (ie rows) in the test meta file ? \nOr should we take into consideration batches of different sizes ? (the proper manner to code, but I am lazy ahah)\nThanks !",
      "votes": null
    },
    {
      "id": "2128554",
      "postDate": "02/03/2023 19:54:09",
      "content": "<p>I believe it is around 1 million rows. But your code should probably be resilient. i.e. if there are 1,350,050 entities or 999,999 entities, it should still work. You can then use this, the batch ids, and the associated events and meta to generate inferences for submission.</p>\n<p>Hope this helps.</p>",
      "rawMarkdown": "I believe it is around 1 million rows. But your code should probably be resilient. i.e. if there are 1,350,050 entities or 999,999 entities, it should still work. You can then use this, the batch ids, and the associated events and meta to generate inferences for submission.\n\nHope this helps.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2128554,
      "author_name": "dschettler8845",
      "author_url": "",
      "post_date": "02/03/2023 19:54:09",
      "content": "<p>I believe it is around 1 million rows. But your code should probably be resilient. i.e. if there are 1,350,050 entities or 999,999 entities, it should still work. You can then use this, the batch ids, and the associated events and meta to generate inferences for submission.</p>\n<p>Hope this helps.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2127431": "Can we assume each test .parquet file contains 200K events (ie rows) in the test meta file ? \nOr should we take into consideration batches of different sizes ? (the proper manner to code, but I am lazy ahah)\nThanks !",
    "2128554": "I believe it is around 1 million rows. But your code should probably be resilient. i.e. if there are 1,350,050 entities or 999,999 entities, it should still work. You can then use this, the batch ids, and the associated events and meta to generate inferences for submission.\n\nHope this helps."
  },
  "source": "meta"
}