{
  "id": 94309,
  "title": "How could I install PyG in kernel?",
  "url": "/competitions/champs-scalar-coupling/discussion/94309",
  "author_name": "",
  "post_date": "2019-06-04T00:21:04.278418900Z",
  "votes": null,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I try the installation guide in the docs but failed</p>",
  "messages": [
    {
      "id": "542479",
      "postDate": "06/04/2019 00:21:04",
      "content": "<p>I try the installation guide in the docs but failed</p>",
      "rawMarkdown": "I try the installation guide in the docs but failed",
      "votes": null
    },
    {
      "id": "543539",
      "postDate": "06/04/2019 15:26:44",
      "content": "<p>Hi, I have also tried to install PyG in the kernel, it didn't work after several times. I used my own computer to install the PyG. </p>",
      "rawMarkdown": "Hi, I have also tried to install PyG in the kernel, it didn't work after several times. I used my own computer to install the PyG.",
      "votes": null
    },
    {
      "id": "546490",
      "postDate": "06/06/2019 16:09:17",
      "content": "<p>I have solved this problem! <a href=\"https://www.kaggle.com/docs/kernels#modifying-the-default-environment\">Here</a> are the instructions on how to install packages in Kernels. Make sure you have enabled \"Internet\" in the settings pane 👌 </p>",
      "rawMarkdown": "I have solved this problem! [Here](https://www.kaggle.com/docs/kernels#modifying-the-default-environment) are the instructions on how to install packages in Kernels. Make sure you have enabled \"Internet\" in the settings pane 👌",
      "votes": null
    },
    {
      "id": "546529",
      "postDate": "06/06/2019 16:38:41",
      "content": "<p>Thanks! I just tried the install and it worked well!\nBtw, I am working on building DataSet in PyG with these molecule data. Wondering if you have any thought about how to construct the data?</p>",
      "rawMarkdown": "Thanks! I just tried the install and it worked well!\nBtw, I am working on building DataSet in PyG with these molecule data. Wondering if you have any thought about how to construct the data?",
      "votes": null
    },
    {
      "id": "546882",
      "postDate": "06/07/2019 01:46:05",
      "content": "<p>Building Dataset in PyG is my main trouble now, I think there are two way to go:<a href=\"https://rusty1s.github.io/pytorch_geometric/build/html/notes/create_dataset.html\">Creating your own datasets</a> or \"simply pass a regular python list holding torch_geometric.data.Data objects and pass them to torch_geometric.data.DataLoader \"\"\n`from torch_geometric.data import Data, DataLoader</p>\n\n<p>data_list = [Data(...), ..., Data(...)]\nloader = DataLoader(data_list, batch_size=32)`</p>",
      "rawMarkdown": "Building Dataset in PyG is my main trouble now, I think there are two way to go:[Creating your own datasets](https://rusty1s.github.io/pytorch_geometric/build/html/notes/create_dataset.html) or \"simply pass a regular python list holding torch_geometric.data.Data objects and pass them to torch_geometric.data.DataLoader \"\"\n`from torch_geometric.data import Data, DataLoader\n\ndata_list = [Data(...), ..., Data(...)]\nloader = DataLoader(data_list, batch_size=32)`",
      "votes": null
    },
    {
      "id": "546907",
      "postDate": "06/07/2019 02:20:09",
      "content": "<p>Thanks for your suggestion!\nToday I found these two papers/github projects(<a href=\"https://github.com/Tony-Y/cgnn)(https://github.com/txie-93/cgcnn#define-a-customized-dataset)\">https://github.com/Tony-Y/cgnn)(https://github.com/txie-93/cgcnn#define-a-customized-dataset)</a>. I feel it may help for constructing the dataset, and I am still reading them. I will keep you posted if it works. </p>",
      "rawMarkdown": "Thanks for your suggestion!\nToday I found these two papers/github projects(https://github.com/Tony-Y/cgnn)(https://github.com/txie-93/cgcnn#define-a-customized-dataset). I feel it may help for constructing the dataset, and I am still reading them. I will keep you posted if it works.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 543539,
      "author_name": "hd2019",
      "author_url": "",
      "post_date": "06/04/2019 15:26:44",
      "content": "<p>Hi, I have also tried to install PyG in the kernel, it didn't work after several times. I used my own computer to install the PyG. </p>",
      "votes": null,
      "replies": [
        {
          "id": 546490,
          "author_name": "milkshakespeare",
          "author_url": "",
          "post_date": "06/06/2019 16:09:17",
          "content": "<p>I have solved this problem! <a href=\"https://www.kaggle.com/docs/kernels#modifying-the-default-environment\">Here</a> are the instructions on how to install packages in Kernels. Make sure you have enabled \"Internet\" in the settings pane 👌 </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 546529,
          "author_name": "hd2019",
          "author_url": "",
          "post_date": "06/06/2019 16:38:41",
          "content": "<p>Thanks! I just tried the install and it worked well!\nBtw, I am working on building DataSet in PyG with these molecule data. Wondering if you have any thought about how to construct the data?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 546882,
          "author_name": "milkshakespeare",
          "author_url": "",
          "post_date": "06/07/2019 01:46:05",
          "content": "<p>Building Dataset in PyG is my main trouble now, I think there are two way to go:<a href=\"https://rusty1s.github.io/pytorch_geometric/build/html/notes/create_dataset.html\">Creating your own datasets</a> or \"simply pass a regular python list holding torch_geometric.data.Data objects and pass them to torch_geometric.data.DataLoader \"\"\n`from torch_geometric.data import Data, DataLoader</p>\n\n<p>data_list = [Data(...), ..., Data(...)]\nloader = DataLoader(data_list, batch_size=32)`</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 546907,
          "author_name": "hd2019",
          "author_url": "",
          "post_date": "06/07/2019 02:20:09",
          "content": "<p>Thanks for your suggestion!\nToday I found these two papers/github projects(<a href=\"https://github.com/Tony-Y/cgnn)(https://github.com/txie-93/cgcnn#define-a-customized-dataset)\">https://github.com/Tony-Y/cgnn)(https://github.com/txie-93/cgcnn#define-a-customized-dataset)</a>. I feel it may help for constructing the dataset, and I am still reading them. I will keep you posted if it works. </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "542479": "I try the installation guide in the docs but failed",
    "543539": "Hi, I have also tried to install PyG in the kernel, it didn't work after several times. I used my own computer to install the PyG.",
    "546490": "I have solved this problem! [Here](https://www.kaggle.com/docs/kernels#modifying-the-default-environment) are the instructions on how to install packages in Kernels. Make sure you have enabled \"Internet\" in the settings pane 👌",
    "546529": "Thanks! I just tried the install and it worked well!\nBtw, I am working on building DataSet in PyG with these molecule data. Wondering if you have any thought about how to construct the data?",
    "546882": "Building Dataset in PyG is my main trouble now, I think there are two way to go:[Creating your own datasets](https://rusty1s.github.io/pytorch_geometric/build/html/notes/create_dataset.html) or \"simply pass a regular python list holding torch_geometric.data.Data objects and pass them to torch_geometric.data.DataLoader \"\"\n`from torch_geometric.data import Data, DataLoader\n\ndata_list = [Data(...), ..., Data(...)]\nloader = DataLoader(data_list, batch_size=32)`",
    "546907": "Thanks for your suggestion!\nToday I found these two papers/github projects(https://github.com/Tony-Y/cgnn)(https://github.com/txie-93/cgcnn#define-a-customized-dataset). I feel it may help for constructing the dataset, and I am still reading them. I will keep you posted if it works."
  },
  "source": "meta"
}