{
  "id": 398996,
  "title": "ensembling Graphnet",
  "url": "/competitions/icecube-neutrinos-in-deep-ice/discussion/398996",
  "author_name": "",
  "post_date": "2023-04-01T21:39:19.510905800Z",
  "votes": 6,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I have found that putting 2 versions of graphnet together improves peoformance. It is not the best public notebook but I found it interesting. I used the baseline one we were provided plus another one trained on 5 batches with the AdamW optimiser</p>\n<p><a href=\"https://www.kaggle.com/code/domdejonge/graphnet-baseline-submission\" target=\"_blank\">https://www.kaggle.com/code/domdejonge/graphnet-baseline-submission</a></p>",
  "messages": [
    {
      "id": "2205763",
      "postDate": "04/01/2023 21:39:19",
      "content": "<p>I have found that putting 2 versions of graphnet together improves peoformance. It is not the best public notebook but I found it interesting. I used the baseline one we were provided plus another one trained on 5 batches with the AdamW optimiser</p>\n<p><a href=\"https://www.kaggle.com/code/domdejonge/graphnet-baseline-submission\" target=\"_blank\">https://www.kaggle.com/code/domdejonge/graphnet-baseline-submission</a></p>",
      "rawMarkdown": "I have found that putting 2 versions of graphnet together improves peoformance. It is not the best public notebook but I found it interesting. I used the baseline one we were provided plus another one trained on 5 batches with the AdamW optimiser\n\nhttps://www.kaggle.com/code/domdejonge/graphnet-baseline-submission",
      "votes": null
    },
    {
      "id": "2209936",
      "postDate": "04/05/2023 03:30:46",
      "content": "<p>May I ask how many data use to train and how long, maybe because my computing resource limit, I havn't train successfully yet😅</p>",
      "rawMarkdown": "May I ask how many data use to train and how long, maybe because my computing resource limit, I havn't train successfully yet😅",
      "votes": null
    },
    {
      "id": "2210654",
      "postDate": "04/05/2023 14:37:23",
      "content": "<p>Trained on kaggle GPU with 1 batch + Adamw optomiser</p>",
      "rawMarkdown": "Trained on kaggle GPU with 1 batch + Adamw optomiser",
      "votes": null
    },
    {
      "id": "2211763",
      "postDate": "04/06/2023 08:47:09",
      "content": "<p>interesting, let me try, however thanks <a href=\"https://www.kaggle.com/domdejonge\" target=\"_blank\">@domdejonge</a> for Info 👍</p>",
      "rawMarkdown": "interesting, let me try, however thanks @domdejonge for Info 👍",
      "votes": null
    },
    {
      "id": "2211821",
      "postDate": "04/06/2023 09:31:08",
      "content": "<p>Thx👀, I will keep trying</p>",
      "rawMarkdown": "Thx👀, I will keep trying",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2209936,
      "author_name": "roger92",
      "author_url": "",
      "post_date": "04/05/2023 03:30:46",
      "content": "<p>May I ask how many data use to train and how long, maybe because my computing resource limit, I havn't train successfully yet😅</p>",
      "votes": null,
      "replies": [
        {
          "id": 2210654,
          "author_name": "domdejonge",
          "author_url": "",
          "post_date": "04/05/2023 14:37:23",
          "content": "<p>Trained on kaggle GPU with 1 batch + Adamw optomiser</p>",
          "votes": null,
          "replies": [
            {
              "id": 2211821,
              "author_name": "roger92",
              "author_url": "",
              "post_date": "04/06/2023 09:31:08",
              "content": "<p>Thx👀, I will keep trying</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 2211763,
      "author_name": "mdbadrulislam",
      "author_url": "",
      "post_date": "04/06/2023 08:47:09",
      "content": "<p>interesting, let me try, however thanks <a href=\"https://www.kaggle.com/domdejonge\" target=\"_blank\">@domdejonge</a> for Info 👍</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2205763": "I have found that putting 2 versions of graphnet together improves peoformance. It is not the best public notebook but I found it interesting. I used the baseline one we were provided plus another one trained on 5 batches with the AdamW optimiser\n\nhttps://www.kaggle.com/code/domdejonge/graphnet-baseline-submission",
    "2209936": "May I ask how many data use to train and how long, maybe because my computing resource limit, I havn't train successfully yet😅",
    "2210654": "Trained on kaggle GPU with 1 batch + Adamw optomiser",
    "2211763": "interesting, let me try, however thanks @domdejonge for Info 👍",
    "2211821": "Thx👀, I will keep trying"
  },
  "source": "meta"
}