{
  "id": 74357,
  "title": "second stage with pretrained model! help new kaggler",
  "url": "/competitions/quora-insincere-questions-classification/discussion/74357",
  "author_name": "",
  "post_date": "2018-12-11T13:49:57.335522400Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>In the second stage, only the test data is changed. Do we only need to load pretrained model i have saved as private data in kernel to make predictions instead of train again?\nmy code as follow:\nif file path exits:\nmodel.load\nelse:\nmodel.train\ncan it work in the 2 stage?</p>",
  "messages": [
    {
      "id": "437179",
      "postDate": "12/11/2018 13:49:57",
      "content": "<p>In the second stage, only the test data is changed. Do we only need to load pretrained model i have saved as private data in kernel to make predictions instead of train again?\nmy code as follow:\nif file path exits:\nmodel.load\nelse:\nmodel.train\ncan it work in the 2 stage?</p>",
      "rawMarkdown": "In the second stage, only the test data is changed. Do we only need to load pretrained model i have saved as private data in kernel to make predictions instead of train again?\nmy code as follow:\nif file path exits:\nmodel.load\nelse:\nmodel.train\ncan it work in the 2 stage?",
      "votes": null
    },
    {
      "id": "437245",
      "postDate": "12/11/2018 15:53:36",
      "content": "<p>Your entire kernel will be re-run, so it will train on the train data (same as before) and then predict on the new test data.</p>",
      "rawMarkdown": "Your entire kernel will be re-run, so it will train on the train data (same as before) and then predict on the new test data.",
      "votes": null
    },
    {
      "id": "437475",
      "postDate": "12/12/2018 01:38:53",
      "content": "<p>what if i load pretrained model i have saved as private data in kernel? like this\nif file path exits:\n   model.load\nelse:\n  model.train</p>",
      "rawMarkdown": "what if i load pretrained model i have saved as private data in kernel? like this\nif file path exits:\n   model.load\nelse:\n  model.train",
      "votes": null
    },
    {
      "id": "437491",
      "postDate": "12/12/2018 02:44:16",
      "content": "<p>I think it can not be use.</p>",
      "rawMarkdown": "I think it can not be use.",
      "votes": null
    },
    {
      "id": "437550",
      "postDate": "12/12/2018 05:53:48",
      "content": "<p>thank you,but in 2 stage, test dataset is very large,is time limit same as 2 hours?</p>",
      "rawMarkdown": "thank you,but in 2 stage, test dataset is very large,is time limit same as 2 hours?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 437245,
      "author_name": "peterhurford",
      "author_url": "",
      "post_date": "12/11/2018 15:53:36",
      "content": "<p>Your entire kernel will be re-run, so it will train on the train data (same as before) and then predict on the new test data.</p>",
      "votes": null,
      "replies": [
        {
          "id": 437475,
          "author_name": "liangdeli",
          "author_url": "",
          "post_date": "12/12/2018 01:38:53",
          "content": "<p>what if i load pretrained model i have saved as private data in kernel? like this\nif file path exits:\n   model.load\nelse:\n  model.train</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 437491,
          "author_name": "tw1994",
          "author_url": "",
          "post_date": "12/12/2018 02:44:16",
          "content": "<p>I think it can not be use.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 437550,
          "author_name": "liangdeli",
          "author_url": "",
          "post_date": "12/12/2018 05:53:48",
          "content": "<p>thank you,but in 2 stage, test dataset is very large,is time limit same as 2 hours?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "437179": "In the second stage, only the test data is changed. Do we only need to load pretrained model i have saved as private data in kernel to make predictions instead of train again?\nmy code as follow:\nif file path exits:\nmodel.load\nelse:\nmodel.train\ncan it work in the 2 stage?",
    "437245": "Your entire kernel will be re-run, so it will train on the train data (same as before) and then predict on the new test data.",
    "437475": "what if i load pretrained model i have saved as private data in kernel? like this\nif file path exits:\n   model.load\nelse:\n  model.train",
    "437491": "I think it can not be use.",
    "437550": "thank you,but in 2 stage, test dataset is very large,is time limit same as 2 hours?"
  },
  "source": "meta"
}