{
  "id": 274123,
  "title": "torch.manual_seed(3407) is all you need ",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/274123",
  "author_name": "",
  "post_date": "2021-09-24T04:50:24.589807300Z",
  "votes": 14,
  "comment_count": 2,
  "views": 0,
  "content": "<p>In this competition, with the amount of data is too small, the 3D data also has a lot of problems, our models are difficult to learn and it is often unstable.</p>\n<p>Have a little fun with this paper: <a href=\"https://arxiv.org/pdf/2109.08203.pdf\" target=\"_blank\"><strong>torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer vision</strong></a></p>\n<p>Another discussion on the same topic: <a href=\"https://www.kaggle.com/c/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/271214\" target=\"_blank\">Changing Random Seeds can get you from 0.540 to 0.710 in no time</a></p>",
  "messages": [
    {
      "id": "1522312",
      "postDate": "09/24/2021 04:50:24",
      "content": "<p>In this competition, with the amount of data is too small, the 3D data also has a lot of problems, our models are difficult to learn and it is often unstable.</p>\n<p>Have a little fun with this paper: <a href=\"https://arxiv.org/pdf/2109.08203.pdf\" target=\"_blank\"><strong>torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer vision</strong></a></p>\n<p>Another discussion on the same topic: <a href=\"https://www.kaggle.com/c/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/271214\" target=\"_blank\">Changing Random Seeds can get you from 0.540 to 0.710 in no time</a></p>",
      "rawMarkdown": "In this competition, with the amount of data is too small, the 3D data also has a lot of problems, our models are difficult to learn and it is often unstable.\n\nHave a little fun with this paper: [**torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer vision**](https://arxiv.org/pdf/2109.08203.pdf)\n\nAnother discussion on the same topic: [Changing Random Seeds can get you from 0.540 to 0.710 in no time](https://www.kaggle.com/c/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/271214)",
      "votes": null
    },
    {
      "id": "1524361",
      "postDate": "09/26/2021 13:07:40",
      "content": "<p>The question is how to find a seed to train a model maximizing the score on the private LB? </p>",
      "rawMarkdown": "The question is how to find a seed to train a model maximizing the score on the private LB?",
      "votes": null
    },
    {
      "id": "1531427",
      "postDate": "10/02/2021 02:04:01",
      "content": "<p>Interesting. I think I’ll use that.😃</p>",
      "rawMarkdown": "Interesting. I think I’ll use that.😃",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1524361,
      "author_name": "mikecho",
      "author_url": "",
      "post_date": "09/26/2021 13:07:40",
      "content": "<p>The question is how to find a seed to train a model maximizing the score on the private LB? </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1531427,
      "author_name": "hiroshikameya",
      "author_url": "",
      "post_date": "10/02/2021 02:04:01",
      "content": "<p>Interesting. I think I’ll use that.😃</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1522312": "In this competition, with the amount of data is too small, the 3D data also has a lot of problems, our models are difficult to learn and it is often unstable.\n\nHave a little fun with this paper: [**torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer vision**](https://arxiv.org/pdf/2109.08203.pdf)\n\nAnother discussion on the same topic: [Changing Random Seeds can get you from 0.540 to 0.710 in no time](https://www.kaggle.com/c/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/271214)",
    "1524361": "The question is how to find a seed to train a model maximizing the score on the private LB?",
    "1531427": "Interesting. I think I’ll use that.😃"
  },
  "source": "meta"
}