{
  "fetched_at": "2026-10-24T12:57:28Z",
  "source_url": "https://www.kaggle.com/competitions/asugnn/data",
  "pages": [
    {
      "name": "data-description",
      "content": "**Download the testing data at :**https://huggingface.co/caobin/ASUGNN\n\n\n**Submission File:**\nYour submission file should be a CSV file with two columns: `ID` and `formation_energy`. For each id in the file, provide the predicted formation energy / atom (formation_energy). A template upload file, named `res.csv`, is provided in the data section. Simply replace the `formation_energy` column with your real predictions.\n\n**Example:**\n```\nID,formation_energy\n1,0\n2,0\n3,0\n4,0\n...\n```\n\nhere is the upload template",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "The HMdataset contains 1,100 crystal structures along with their corresponding formation energy per atom. This dataset is provided for testing the zero-shot capability of various models. We are open-sourcing it to encourage participants to explore and benchmark their models' predictive performance.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "\nSubmissions are evaluated on MAE\n\n\n**Submission File:**\nYour submission file should be a CSV file with two columns: `ID` and `formation_energy`. For each id in the file, provide the predicted formation energy / atom (formation_energy). A template upload file, named `res.csv`, is provided in the data section. Simply replace the `formation_energy` column with your real predictions.\n\n**Example:**\n```\nID,formation_energy\n1,0\n2,0\n3,0\n4,0\n...\n```\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "\n\n - Don't cheat!\n - Apply yourself!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "abstract",
      "content": "\n**Welcome to the Competition!**\n\nThe **HMdataset** contains 1,100 crystal structures along with their corresponding formation energy per atom. This dataset is provided for testing the zero-shot capability of various models. We are open-sourcing it to encourage participants to explore and benchmark their models' predictive performance.\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}