{
  "topic": {
    "id": 553294,
    "title": "Starter notebook with LLMs",
    "authorName": "Tong Hui Kang",
    "commentCount": 12,
    "votes": 29,
    "postDate": "2024-12-25T04:53:44.817000"
  },
  "comments": [
    {
      "id": 3083202,
      "authorName": "Tong Hui Kang",
      "votes": 3,
      "postDate": "2024-12-29T06:23:58.603000",
      "content": "<p>Apparently I can't even get o1 to work on the training examples</p>\n<p><a href=\"https://www.kaggle.com/code/huikang/select-patch-verify-with-o1\" target=\"_blank\">https://www.kaggle.com/code/huikang/select-patch-verify-with-o1</a></p>"
    },
    {
      "id": 3081150,
      "authorName": "aaaditya56",
      "votes": 1,
      "postDate": "2024-12-26T10:02:47.457000",
      "content": "<p>Thank you so much for this great resource!</p>"
    },
    {
      "id": 3080655,
      "authorName": "Khoi Nguyen",
      "votes": 1,
      "postDate": "2024-12-25T14:21:28.960000",
      "content": "<p>You're doing God's work here, out of all the possible formats a single notebook is probably the worst way to do this.</p>"
    },
    {
      "id": 3080681,
      "authorName": "Tong Hui Kang",
      "votes": -1,
      "postDate": "2024-12-25T15:11:52.263000",
      "content": "<p>The notebook is just a Python script split into different cells, I don't blame the notebook format.</p>\n<p>I think the API is unnecessary, I don't see why we can't we just put our submissions in a csv, and let the evaluation happen in a separate process. This way I don't need to write one of my tutorial notebooks, instead of focusing on how to use the API we have the freedom to do whatever we want as long as we create a <code>submission.csv</code></p>"
    },
    {
      "id": 3081905,
      "authorName": "Steven Liu",
      "votes": 2,
      "postDate": "2024-12-27T11:38:21.400000",
      "content": "<ol>\n<li>Make sure you are using Settings &gt; Accelerator &gt; GPU L4 x 4</li>\n<li>Make sure to press File &gt; Input &gt; Qwen 2.5 Coder, Transfromers, 32b-instruct-awq</li>\n<li>Make sure you install the vLLM Installation Fix patch: <a href=\"https://www.kaggle.com/code/abdullahmeda/load-72b-awq-model-using-vllm-on-l4-x4/comments#3042241\" target=\"_blank\">https://www.kaggle.com/code/abdullahmeda/load-72b-awq-model-using-vllm-on-l4-x4/comments#3042241</a></li>\n</ol>\n<p>============== Alternative ============== <br>\nI tried to run your notebook but it doesn't work on kaggle… however, your generosity is inspiring 🙌  so here's another notebook which will run on kaggle which gives a function <code>send_llm</code> you can call to send responses to an LLM running completely within kaggle itself!<br>\n<a href=\"https://www.kaggle.com/code/thornkey/notebookdd75bed624\" target=\"_blank\">https://www.kaggle.com/code/thornkey/notebookdd75bed624</a></p>\n<p>(Note you will need your own HuggingFace token, but that's free 🙂 )<br>\nHeaps of kudos to <a href=\"https://github.com/casualcomputer/llm_google_colab\" target=\"_blank\">https://github.com/casualcomputer/llm_google_colab</a> which I distilled to make this notebook.<br>\nSee you at the finish line! 🙂</p>"
    },
    {
      "id": 3082195,
      "authorName": "Tong Hui Kang",
      "votes": 1,
      "postDate": "2024-12-27T19:30:57.770000",
      "content": "<p>When you submit the notebook, it cannot use the Internet at all.</p>\n<p>Your method could run the notebook, but it could not be used for submission.</p>\n<p>I added a note there on how to run the notebook <a href=\"https://www.kaggle.com/code/huikang/starter-notebook-select-patch-verify/comments#3082193\" target=\"_blank\">https://www.kaggle.com/code/huikang/starter-notebook-select-patch-verify/comments#3082193</a></p>"
    },
    {
      "id": 3082215,
      "authorName": "Steven Liu",
      "votes": 0,
      "postDate": "2024-12-27T19:58:36.133000",
      "content": "<p>Fair enough! I was feeling a little frustrated yesterday trying to recreate this, haha - thanks for sharing!</p>"
    },
    {
      "id": 3087934,
      "authorName": "Justin Connor Davis",
      "votes": 0,
      "postDate": "2025-01-04T05:22:43.383000",
      "content": "<p>Awesome work, this is a great resource to reference!</p>"
    },
    {
      "id": 3080867,
      "authorName": "mhdaw",
      "votes": 0,
      "postDate": "2024-12-25T21:07:24.637000",
      "content": "<p>why LLM based subs get a negative score? </p>"
    },
    {
      "id": 3097647,
      "authorName": "Ginto",
      "votes": 0,
      "postDate": "2025-01-15T15:29:27.500000",
      "content": "<p>Now, do you know why？</p>"
    },
    {
      "id": 3086494,
      "authorName": "",
      "votes": 0,
      "postDate": "2025-01-02T11:53:07.313000",
      "content": ""
    },
    {
      "id": 3082573,
      "authorName": "Chong Gu (Why_stable)",
      "votes": 0,
      "postDate": "2024-12-28T11:31:56.540000",
      "content": "<p>thanks for sharing</p>"
    }
  ],
  "index": {
    "id": "553294",
    "title": "Starter notebook with LLMs",
    "authorName": "",
    "commentCount": "12",
    "votes": "29",
    "postDate": "2024-12-25 04:53:44.817000"
  },
  "competition": "konwinski-prize"
}