{
  "id": 546795,
  "title": "idea to try: TabM: Advancing Tabular Deep Learning With Parameter-Efficient Ensembling",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/546795",
  "author_name": "",
  "post_date": "2024-11-18T05:28:14.963617500Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p><a href=\"https://github.com/yandex-research/tabm\" target=\"_blank\">https://github.com/yandex-research/tabm</a><br>\ndidn't try yet, but looks promising </p>",
  "messages": [
    {
      "id": "3048559",
      "postDate": "11/18/2024 05:28:14",
      "content": "<p><a href=\"https://github.com/yandex-research/tabm\" target=\"_blank\">https://github.com/yandex-research/tabm</a><br>\ndidn't try yet, but looks promising </p>",
      "rawMarkdown": "https://github.com/yandex-research/tabm\ndidn't try yet, but looks promising",
      "votes": null
    },
    {
      "id": "3048609",
      "postDate": "11/18/2024 06:48:01",
      "content": "<p>I have tried this model, but there is too many samples so it take dataloader long time to load data.I have to use an idiot method: loading data from parquet directly😶‍🌫️.There is my original code: <a href=\"https://www.kaggle.com/code/i2nfinit3y/jane-street-tabm-training\" target=\"_blank\">https://www.kaggle.com/code/i2nfinit3y/jane-street-tabm-training</a>, and I have not submitted my inference result, hoping you can fix my terrible code.</p>",
      "rawMarkdown": "I have tried this model, but there is too many samples so it take dataloader long time to load data.I have to use an idiot method: loading data from parquet directly😶‍🌫️.There is my original code: https://www.kaggle.com/code/i2nfinit3y/jane-street-tabm-training, and I have not submitted my inference result, hoping you can fix my terrible code.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3048609,
      "author_name": "i2nfinit3y",
      "author_url": "",
      "post_date": "11/18/2024 06:48:01",
      "content": "<p>I have tried this model, but there is too many samples so it take dataloader long time to load data.I have to use an idiot method: loading data from parquet directly😶‍🌫️.There is my original code: <a href=\"https://www.kaggle.com/code/i2nfinit3y/jane-street-tabm-training\" target=\"_blank\">https://www.kaggle.com/code/i2nfinit3y/jane-street-tabm-training</a>, and I have not submitted my inference result, hoping you can fix my terrible code.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3048559": "https://github.com/yandex-research/tabm\ndidn't try yet, but looks promising",
    "3048609": "I have tried this model, but there is too many samples so it take dataloader long time to load data.I have to use an idiot method: loading data from parquet directly😶‍🌫️.There is my original code: https://www.kaggle.com/code/i2nfinit3y/jane-street-tabm-training, and I have not submitted my inference result, hoping you can fix my terrible code."
  },
  "source": "meta"
}