{
  "id": 400994,
  "title": "How do you submit your notebooks? ",
  "url": "/competitions/birdclef-2023/discussion/400994",
  "author_name": "",
  "post_date": "2023-04-11T08:55:23.185821700Z",
  "votes": 3,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Can you directly upload and submit the local prediction result submission.csv file? Do I need to perform model training in the notebook, and then submit the prediction results of the soundscape_29201.ogg file?</p>",
  "messages": [
    {
      "id": "2217899",
      "postDate": "04/11/2023 08:55:23",
      "content": "<p>Can you directly upload and submit the local prediction result submission.csv file? Do I need to perform model training in the notebook, and then submit the prediction results of the soundscape_29201.ogg file?</p>",
      "rawMarkdown": "Can you directly upload and submit the local prediction result submission.csv file? Do I need to perform model training in the notebook, and then submit the prediction results of the soundscape_29201.ogg file?",
      "votes": null
    },
    {
      "id": "2218475",
      "postDate": "04/11/2023 18:15:15",
      "content": "<p>Hey Xiango, you do not have to train your model(s) in the submission notebook - this would bee too time consuming and the submission notebook in this case is only allowed to use CPU.<br>\nTherefore you have to train your model(s) locally or in other notebooks and attach the models as a dataset to your submission notebook which is then able to load the models for inference. </p>\n<p>Nevertheless, your preprocessing needs to be included in the submission notebook in order to preprocess the data just like you did for training.</p>\n<p>Just take a look at the published \"inference\" notebooks in the Code section.</p>\n<p>Hope this helps!</p>\n<p>Best,<br>\nJan</p>",
      "rawMarkdown": "Hey Xiango, you do not have to train your model(s) in the submission notebook - this would bee too time consuming and the submission notebook in this case is only allowed to use CPU.\nTherefore you have to train your model(s) locally or in other notebooks and attach the models as a dataset to your submission notebook which is then able to load the models for inference. \n\nNevertheless, your preprocessing needs to be included in the submission notebook in order to preprocess the data just like you did for training.\n\nJust take a look at the published \"inference\" notebooks in the Code section.\n\nHope this helps!\n\nBest,\nJan",
      "votes": null
    },
    {
      "id": "2220551",
      "postDate": "04/13/2023 13:48:14",
      "content": "<p>Hi Jan, </p>\n<p>I have a followup question, is it possible to persist our model in input folder. Some of the infer notebooks try to save the model checkpoints into the input folder, but when I try to write something \"/kaggle/input/birdclef-2023\" folder access denied.</p>\n<p>Meanwhile as per your suggestion I have split my notebooks into [infer] and [submission].Trying to download the model from output file and added it as a dataset to submission notebook. </p>",
      "rawMarkdown": "Hi Jan, \n\nI have a followup question, is it possible to persist our model in input folder. Some of the infer notebooks try to save the model checkpoints into the input folder, but when I try to write something \"/kaggle/input/birdclef-2023\" folder access denied.\n\nMeanwhile as per your suggestion I have split my notebooks into [infer] and [submission].Trying to download the model from output file and added it as a dataset to submission notebook.",
      "votes": null
    },
    {
      "id": "2221281",
      "postDate": "04/14/2023 05:40:39",
      "content": "<p>Today, when I tried to add my model as a dataset, it's acutally adding it to the input folder of the competitions - so that clarifies by question Thanks Jan &amp; Xiango</p>",
      "rawMarkdown": "Today, when I tried to add my model as a dataset, it's acutally adding it to the input folder of the competitions - so that clarifies by question Thanks Jan & Xiango",
      "votes": null
    },
    {
      "id": "2222056",
      "postDate": "04/14/2023 20:42:34",
      "content": "<p>Hey Rajesh, happy it helped!</p>\n<p>Best of luck,<br>\nJan</p>",
      "rawMarkdown": "Hey Rajesh, happy it helped!\n\nBest of luck,\nJan",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2218475,
      "author_name": "janbrederecke",
      "author_url": "",
      "post_date": "04/11/2023 18:15:15",
      "content": "<p>Hey Xiango, you do not have to train your model(s) in the submission notebook - this would bee too time consuming and the submission notebook in this case is only allowed to use CPU.<br>\nTherefore you have to train your model(s) locally or in other notebooks and attach the models as a dataset to your submission notebook which is then able to load the models for inference. </p>\n<p>Nevertheless, your preprocessing needs to be included in the submission notebook in order to preprocess the data just like you did for training.</p>\n<p>Just take a look at the published \"inference\" notebooks in the Code section.</p>\n<p>Hope this helps!</p>\n<p>Best,<br>\nJan</p>",
      "votes": null,
      "replies": [
        {
          "id": 2220551,
          "author_name": "rajeshradhakrishnan",
          "author_url": "",
          "post_date": "04/13/2023 13:48:14",
          "content": "<p>Hi Jan, </p>\n<p>I have a followup question, is it possible to persist our model in input folder. Some of the infer notebooks try to save the model checkpoints into the input folder, but when I try to write something \"/kaggle/input/birdclef-2023\" folder access denied.</p>\n<p>Meanwhile as per your suggestion I have split my notebooks into [infer] and [submission].Trying to download the model from output file and added it as a dataset to submission notebook. </p>",
          "votes": null,
          "replies": [
            {
              "id": 2221281,
              "author_name": "rajeshradhakrishnan",
              "author_url": "",
              "post_date": "04/14/2023 05:40:39",
              "content": "<p>Today, when I tried to add my model as a dataset, it's acutally adding it to the input folder of the competitions - so that clarifies by question Thanks Jan &amp; Xiango</p>",
              "votes": null,
              "replies": [
                {
                  "id": 2222056,
                  "author_name": "janbrederecke",
                  "author_url": "",
                  "post_date": "04/14/2023 20:42:34",
                  "content": "<p>Hey Rajesh, happy it helped!</p>\n<p>Best of luck,<br>\nJan</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2217899": "Can you directly upload and submit the local prediction result submission.csv file? Do I need to perform model training in the notebook, and then submit the prediction results of the soundscape_29201.ogg file?",
    "2218475": "Hey Xiango, you do not have to train your model(s) in the submission notebook - this would bee too time consuming and the submission notebook in this case is only allowed to use CPU.\nTherefore you have to train your model(s) locally or in other notebooks and attach the models as a dataset to your submission notebook which is then able to load the models for inference. \n\nNevertheless, your preprocessing needs to be included in the submission notebook in order to preprocess the data just like you did for training.\n\nJust take a look at the published \"inference\" notebooks in the Code section.\n\nHope this helps!\n\nBest,\nJan",
    "2220551": "Hi Jan, \n\nI have a followup question, is it possible to persist our model in input folder. Some of the infer notebooks try to save the model checkpoints into the input folder, but when I try to write something \"/kaggle/input/birdclef-2023\" folder access denied.\n\nMeanwhile as per your suggestion I have split my notebooks into [infer] and [submission].Trying to download the model from output file and added it as a dataset to submission notebook.",
    "2221281": "Today, when I tried to add my model as a dataset, it's acutally adding it to the input folder of the competitions - so that clarifies by question Thanks Jan & Xiango",
    "2222056": "Hey Rajesh, happy it helped!\n\nBest of luck,\nJan"
  },
  "source": "meta"
}