{ "id": 527246, "title": "New to Kaggle: Understanding Competitions and Submissions", "url": "/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/527246", "author_name": "", "post_date": "2024-08-11T08:02:19.490586100Z", "votes": 1, "comment_count": 2, "views": 0, "content": "

I'm new to Kaggle and have a few questions regarding the competition process and submission requirements:

\n
    \n

  1. What does the requirement 'CPU Notebook <= 9 hours run-time, GPU Notebook <= 9 hours run-time' mean? Does it mean that your code, including training and testing, must complete within 9 hours?

    2. \n

  2. Are pre-trained models allowed in competitions?

    3. \n

  3. Do I need to submit the actual model file, or just the results?

    4. \n
", "messages": [ { "id": "2955502", "postDate": "08/11/2024 08:02:19", "content": "

I'm new to Kaggle and have a few questions regarding the competition process and submission requirements:

\n
    \n

  1. What does the requirement 'CPU Notebook <= 9 hours run-time, GPU Notebook <= 9 hours run-time' mean? Does it mean that your code, including training and testing, must complete within 9 hours?

    2. \n

  2. Are pre-trained models allowed in competitions?

    3. \n

  3. Do I need to submit the actual model file, or just the results?

    4. \n
", "rawMarkdown": "I'm new to Kaggle and have a few questions regarding the competition process and submission requirements:\n\n1. What does the requirement 'CPU Notebook <= 9 hours run-time, GPU Notebook <= 9 hours run-time' mean? Does it mean that your code, including training and testing, must complete within 9 hours?\n\n2. Are pre-trained models allowed in competitions?\n\n3. Do I need to submit the actual model file, or just the results?", "votes": null }, { "id": "2955562", "postDate": "08/11/2024 09:34:32", "content": "

1) No. you may train in a separate notebook and load the weights for inference. During submission your nb is allowed to run 9 h.
\n2) yes
\n3) this is a code competition, that means you need to make the submission with a running notebook.

\n

For more information look at the examples under “code”

", "rawMarkdown": "1) No. you may train in a separate notebook and load the weights for inference. During submission your nb is allowed to run 9 h.\n2) yes\n3) this is a code competition, that means you need to make the submission with a running notebook. \n\nFor more information look at the examples under “code”", "votes": null }, { "id": "2975975", "postDate": "09/01/2024 12:06:49", "content": "

FOR 3:
\nTypically, in Kaggle competitions, you only need to submit the results (predictions) of your model on the test dataset provided by the competition organizers. Here's how it works:

\n

Results Submission: You will submit a CSV file or a similar format containing your predictions. The format of this file will be specified in the competition rules.
\nModel File Submission: In most cases, you do not need to submit the actual model file (e.g., a .h5 file for Keras models, or a .pth file for PyTorch models). The organizers are only interested in your predictions.
\nExceptions: In some special competitions (e.g., code competitions or challenges with specific deployment requirements), you may be asked to submit your model or a runnable script that can reproduce your results. Again, the competition rules will specify this clearly if it is required.

", "rawMarkdown": "FOR 3:\nTypically, in Kaggle competitions, you only need to submit the results (predictions) of your model on the test dataset provided by the competition organizers. Here's how it works:\n\nResults Submission: You will submit a CSV file or a similar format containing your predictions. The format of this file will be specified in the competition rules.\nModel File Submission: In most cases, you do not need to submit the actual model file (e.g., a .h5 file for Keras models, or a .pth file for PyTorch models). The organizers are only interested in your predictions.\nExceptions: In some special competitions (e.g., code competitions or challenges with specific deployment requirements), you may be asked to submit your model or a runnable script that can reproduce your results. Again, the competition rules will specify this clearly if it is required.", "votes": null } ], "comments": [ { "id": 2955562, "author_name": "simonveitner", "author_url": "", "post_date": "08/11/2024 09:34:32", "content": "

1) No. you may train in a separate notebook and load the weights for inference. During submission your nb is allowed to run 9 h.
\n2) yes
\n3) this is a code competition, that means you need to make the submission with a running notebook.

\n

For more information look at the examples under “code”

", "votes": null, "replies": [] }, { "id": 2975975, "author_name": "aslkuscu", "author_url": "", "post_date": "09/01/2024 12:06:49", "content": "

FOR 3:
\nTypically, in Kaggle competitions, you only need to submit the results (predictions) of your model on the test dataset provided by the competition organizers. Here's how it works:

\n

Results Submission: You will submit a CSV file or a similar format containing your predictions. The format of this file will be specified in the competition rules.
\nModel File Submission: In most cases, you do not need to submit the actual model file (e.g., a .h5 file for Keras models, or a .pth file for PyTorch models). The organizers are only interested in your predictions.
\nExceptions: In some special competitions (e.g., code competitions or challenges with specific deployment requirements), you may be asked to submit your model or a runnable script that can reproduce your results. Again, the competition rules will specify this clearly if it is required.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2955502": "I'm new to Kaggle and have a few questions regarding the competition process and submission requirements:\n\n1. What does the requirement 'CPU Notebook <= 9 hours run-time, GPU Notebook <= 9 hours run-time' mean? Does it mean that your code, including training and testing, must complete within 9 hours?\n\n2. Are pre-trained models allowed in competitions?\n\n3. Do I need to submit the actual model file, or just the results?", "2955562": "1) No. you may train in a separate notebook and load the weights for inference. During submission your nb is allowed to run 9 h.\n2) yes\n3) this is a code competition, that means you need to make the submission with a running notebook. \n\nFor more information look at the examples under “code”", "2975975": "FOR 3:\nTypically, in Kaggle competitions, you only need to submit the results (predictions) of your model on the test dataset provided by the competition organizers. Here's how it works:\n\nResults Submission: You will submit a CSV file or a similar format containing your predictions. The format of this file will be specified in the competition rules.\nModel File Submission: In most cases, you do not need to submit the actual model file (e.g., a .h5 file for Keras models, or a .pth file for PyTorch models). The organizers are only interested in your predictions.\nExceptions: In some special competitions (e.g., code competitions or challenges with specific deployment requirements), you may be asked to submit your model or a runnable script that can reproduce your results. Again, the competition rules will specify this clearly if it is required." }, "source": "meta" }