{
  "id": 68275,
  "title": "Did anyone try the new ML package released by NVIDIA yesterday in this competition?",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/68275",
  "author_name": "",
  "post_date": "2018-10-10T23:56:12.142208Z",
  "votes": 12,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Nvidia has just released a new machine learning package that contains faster replacements for pandas and xgboost (<a href=\"http://rapids.ai/\">http://rapids.ai/</a>). I couldn't try it because a Pascal GPU or better is required and mine is older, but it seems very promising. Has anyone tried it in this competition?</p>",
  "messages": [
    {
      "id": "401958",
      "postDate": "10/10/2018 23:56:12",
      "content": "<p>Nvidia has just released a new machine learning package that contains faster replacements for pandas and xgboost (<a href=\"http://rapids.ai/\">http://rapids.ai/</a>). I couldn't try it because a Pascal GPU or better is required and mine is older, but it seems very promising. Has anyone tried it in this competition?</p>",
      "rawMarkdown": "Nvidia has just released a new machine learning package that contains faster replacements for pandas and xgboost (http://rapids.ai/). I couldn't try it because a Pascal GPU or better is required and mine is older, but it seems very promising. Has anyone tried it in this competition?",
      "votes": null
    },
    {
      "id": "402045",
      "postDate": "10/11/2018 05:14:58",
      "content": "<p>Thanks for sharing, this looks very interesting. Also claims to have gpu support for sklearn. We might all need to be using this soon. I expect it will be available in kernels for users whose gpu does not support it.</p>\n\n<p>I haven't tried it on this competition as yet. There is a leak via the external data. We have a new leader with plb 0.58. Hopefully we will get some confirmation from Kaggle regarding the status of this competition and whether it will need to be abandoned. </p>",
      "rawMarkdown": "Thanks for sharing, this looks very interesting. Also claims to have gpu support for sklearn. We might all need to be using this soon. I expect it will be available in kernels for users whose gpu does not support it.\n\nI haven't tried it on this competition as yet. There is a leak via the external data. We have a new leader with plb 0.58. Hopefully we will get some confirmation from Kaggle regarding the status of this competition and whether it will need to be abandoned.",
      "votes": null
    },
    {
      "id": "402263",
      "postDate": "10/11/2018 12:36:10",
      "content": "<p>It's been about 6 months since I tried anything with NVIDIA CUDA. My experience is that it's very finicky about which environments it will work with. Maybe this new package will make it easier to work with.</p>",
      "rawMarkdown": "It's been about 6 months since I tried anything with NVIDIA CUDA. My experience is that it's very finicky about which environments it will work with. Maybe this new package will make it easier to work with.",
      "votes": null
    },
    {
      "id": "402469",
      "postDate": "10/11/2018 19:08:55",
      "content": "<p>Thanks for keeping us up-to-date <a href=\"/pandabambu\">@pandabambu</a></p>",
      "rawMarkdown": "Thanks for keeping us up-to-date @pandabambu",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 402045,
      "author_name": "datahobbit",
      "author_url": "",
      "post_date": "10/11/2018 05:14:58",
      "content": "<p>Thanks for sharing, this looks very interesting. Also claims to have gpu support for sklearn. We might all need to be using this soon. I expect it will be available in kernels for users whose gpu does not support it.</p>\n\n<p>I haven't tried it on this competition as yet. There is a leak via the external data. We have a new leader with plb 0.58. Hopefully we will get some confirmation from Kaggle regarding the status of this competition and whether it will need to be abandoned. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 402263,
      "author_name": "afgonczol",
      "author_url": "",
      "post_date": "10/11/2018 12:36:10",
      "content": "<p>It's been about 6 months since I tried anything with NVIDIA CUDA. My experience is that it's very finicky about which environments it will work with. Maybe this new package will make it easier to work with.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 402469,
      "author_name": "mbkinaci",
      "author_url": "",
      "post_date": "10/11/2018 19:08:55",
      "content": "<p>Thanks for keeping us up-to-date <a href=\"/pandabambu\">@pandabambu</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "401958": "Nvidia has just released a new machine learning package that contains faster replacements for pandas and xgboost (http://rapids.ai/). I couldn't try it because a Pascal GPU or better is required and mine is older, but it seems very promising. Has anyone tried it in this competition?",
    "402045": "Thanks for sharing, this looks very interesting. Also claims to have gpu support for sklearn. We might all need to be using this soon. I expect it will be available in kernels for users whose gpu does not support it.\n\nI haven't tried it on this competition as yet. There is a leak via the external data. We have a new leader with plb 0.58. Hopefully we will get some confirmation from Kaggle regarding the status of this competition and whether it will need to be abandoned.",
    "402263": "It's been about 6 months since I tried anything with NVIDIA CUDA. My experience is that it's very finicky about which environments it will work with. Maybe this new package will make it easier to work with.",
    "402469": "Thanks for keeping us up-to-date @pandabambu"
  },
  "source": "meta"
}