{
  "id": 445573,
  "title": "Building GNN from scratch",
  "url": "/competitions/predict-ai-model-runtime/discussion/445573",
  "author_name": "Kilaru Vasudeva",
  "post_date": "2023-10-07T16:33:13.293000",
  "votes": 6,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I have coded up a GNN from scratch using torch along with necessary preprocessing steps for Tile dataset. It may not be very efficient in competition/LB point of view but would be a very helpful educational resource. To understand GNN's theoretical concepts, refer to this <a href=\"https://youtu.be/8owQBFAHw7E?si=iEsfdUUIu-eWvTie\" target=\"_blank\">video</a> then follow the code in my notebook <a href=\"https://www.kaggle.com/code/vasudevak/building-gnn-from-scratch\" target=\"_blank\">here</a>.</p>\n<p>Hope it helps,<br>\nThank you, Vasudeva!</p>",
  "messages": [
    {
      "id": 2472841,
      "postDate": "2023-10-07T16:33:13.293Z",
      "content": "<p>I have coded up a GNN from scratch using torch along with necessary preprocessing steps for Tile dataset. It may not be very efficient in competition/LB point of view but would be a very helpful educational resource. To understand GNN's theoretical concepts, refer to this <a href=\"https://youtu.be/8owQBFAHw7E?si=iEsfdUUIu-eWvTie\" target=\"_blank\">video</a> then follow the code in my notebook <a href=\"https://www.kaggle.com/code/vasudevak/building-gnn-from-scratch\" target=\"_blank\">here</a>.</p>\n<p>Hope it helps,<br>\nThank you, Vasudeva!</p>",
      "rawMarkdown": "I have coded up a GNN from scratch using torch along with necessary preprocessing steps for Tile dataset. It may not be very efficient in competition/LB point of view but would be a very helpful educational resource. To understand GNN's theoretical concepts, refer to this [video](https://youtu.be/8owQBFAHw7E?si=iEsfdUUIu-eWvTie) then follow the code in my notebook [here](https://www.kaggle.com/code/vasudevak/building-gnn-from-scratch).\n\nHope it helps,\nThank you, Vasudeva!",
      "votes": 5
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2472841": "I have coded up a GNN from scratch using torch along with necessary preprocessing steps for Tile dataset. It may not be very efficient in competition/LB point of view but would be a very helpful educational resource. To understand GNN's theoretical concepts, refer to this [video](https://youtu.be/8owQBFAHw7E?si=iEsfdUUIu-eWvTie) then follow the code in my notebook [here](https://www.kaggle.com/code/vasudevak/building-gnn-from-scratch).\n\nHope it helps,\nThank you, Vasudeva!"
  }
}