{
  "id": 419162,
  "title": "Training Log for phase 2",
  "url": "/competitions/budgeted-model-training-iccv-2023-rcv-workshop/discussion/419162",
  "author_name": "",
  "post_date": "2023-06-24T13:17:32.488110300Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hello,</p>\n<p>I wonder if we could get the training log after running on your side? We want to improve the model so that log might be beneficial for us to decide what to modify.</p>\n<p>Thanks!</p>",
  "messages": [
    {
      "id": "2315845",
      "postDate": "06/24/2023 13:17:32",
      "content": "<p>Hello,</p>\n<p>I wonder if we could get the training log after running on your side? We want to improve the model so that log might be beneficial for us to decide what to modify.</p>\n<p>Thanks!</p>",
      "rawMarkdown": "Hello,\n\nI wonder if we could get the training log after running on your side? We want to improve the model so that log might be beneficial for us to decide what to modify.\n\nThanks!",
      "votes": null
    },
    {
      "id": "2317067",
      "postDate": "06/25/2023 12:24:47",
      "content": "<p><a href=\"https://www.kaggle.com/rishabh2001\" target=\"_blank\">@rishabh2001</a> </p>",
      "rawMarkdown": "rishabh2001",
      "votes": null
    },
    {
      "id": "2317086",
      "postDate": "06/25/2023 12:38:43",
      "content": "<p>Hi Tu Vo, thanks for the question. I would like to clarify that we have deliberately kept the training on the phase II private, to evaluate the training methodologies independent from the training instances. This promotes developing more generalizable approaches and avoids overfitting on a particular dataset. All participants should use public training/val and test data to decide any hyperparameters or modifications needed.</p>",
      "rawMarkdown": "Hi Tu Vo, thanks for the question. I would like to clarify that we have deliberately kept the training on the phase II private, to evaluate the training methodologies independent from the training instances. This promotes developing more generalizable approaches and avoids overfitting on a particular dataset. All participants should use public training/val and test data to decide any hyperparameters or modifications needed.",
      "votes": null
    },
    {
      "id": "2317749",
      "postDate": "06/26/2023 00:56:33",
      "content": "<p>Thanks! That makes sense! </p>",
      "rawMarkdown": "Thanks! That makes sense!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2317067,
      "author_name": "tuvovan211",
      "author_url": "",
      "post_date": "06/25/2023 12:24:47",
      "content": "<p><a href=\"https://www.kaggle.com/rishabh2001\" target=\"_blank\">@rishabh2001</a> </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2317086,
      "author_name": "rishabh2001",
      "author_url": "",
      "post_date": "06/25/2023 12:38:43",
      "content": "<p>Hi Tu Vo, thanks for the question. I would like to clarify that we have deliberately kept the training on the phase II private, to evaluate the training methodologies independent from the training instances. This promotes developing more generalizable approaches and avoids overfitting on a particular dataset. All participants should use public training/val and test data to decide any hyperparameters or modifications needed.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2317749,
          "author_name": "tuvovan211",
          "author_url": "",
          "post_date": "06/26/2023 00:56:33",
          "content": "<p>Thanks! That makes sense! </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2315845": "Hello,\n\nI wonder if we could get the training log after running on your side? We want to improve the model so that log might be beneficial for us to decide what to modify.\n\nThanks!",
    "2317067": "rishabh2001",
    "2317086": "Hi Tu Vo, thanks for the question. I would like to clarify that we have deliberately kept the training on the phase II private, to evaluate the training methodologies independent from the training instances. This promotes developing more generalizable approaches and avoids overfitting on a particular dataset. All participants should use public training/val and test data to decide any hyperparameters or modifications needed.",
    "2317749": "Thanks! That makes sense!"
  },
  "source": "meta"
}