{
  "id": 373886,
  "title": "The training is really unstable",
  "url": "/competitions/rsna-breast-cancer-detection/discussion/373886",
  "author_name": "nicehzj",
  "post_date": "2022-12-23T23:16:55.289000",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>When I want to reproduce the code I found that the training progress is really unstable, even I use the same parameters.</p>",
  "messages": [
    {
      "id": 2074264,
      "postDate": "2022-12-23T23:16:55.290Z",
      "content": "<p>When I want to reproduce the code I found that the training progress is really unstable, even I use the same parameters.</p>",
      "rawMarkdown": "When I want to reproduce the code I found that the training progress is really unstable, even I use the same parameters.",
      "votes": 1
    },
    {
      "id": 2079533,
      "postDate": "2022-12-29T11:55:53.087Z",
      "content": "<p>Did you make sure that you also set all the seeds equally?</p>",
      "rawMarkdown": "Did you make sure that you also set all the seeds equally?"
    },
    {
      "id": 2074767,
      "postDate": "2022-12-24T15:14:44.447Z",
      "content": "<p>Yes, I ran into the same problem, and when I tried to reproduce some public notebook, it always didn't get the same results, even with the same parameters。 :(</p>",
      "rawMarkdown": "Yes, I ran into the same problem, and when I tried to reproduce some public notebook, it always didn't get the same results, even with the same parameters。 :(",
      "replies": [
        {
          "id": 2074790,
          "postDate": "2022-12-24T15:56:19.123Z",
          "content": "<p>Same here… I copied a public notebook and I found my eval loss is increasing…</p>",
          "rawMarkdown": "Same here... I copied a public notebook and I found my eval loss is increasing..."
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2079533,
      "author_name": "Ali Abdin",
      "author_url": "",
      "post_date": "2022-12-29T11:55:53.087000",
      "content": "<p>Did you make sure that you also set all the seeds equally?</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 2074767,
      "author_name": "Jijie Li",
      "author_url": "",
      "post_date": "2022-12-24T15:14:44.447000",
      "content": "<p>Yes, I ran into the same problem, and when I tried to reproduce some public notebook, it always didn't get the same results, even with the same parameters。 :(</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2074790,
          "author_name": "nicehzj",
          "author_url": "",
          "post_date": "2022-12-24T15:56:19.123000",
          "content": "<p>Same here… I copied a public notebook and I found my eval loss is increasing…</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2074264": "When I want to reproduce the code I found that the training progress is really unstable, even I use the same parameters.",
    "2079533": "Did you make sure that you also set all the seeds equally?",
    "2074767": "Yes, I ran into the same problem, and when I tried to reproduce some public notebook, it always didn't get the same results, even with the same parameters。 :("
  }
}