I am new to Kaggle and its competitions. I have submitted the notebook three times, but it always says 'Your submission notebook may not have started the inference server that is called to obtain predictions. This could mean you forgot to start it, or the notebook crashed.' after the submission. What does it mean? And what can I do to avoid it? The scoring lasts for 10 minutes or more before failure.
", "messages": [ { "id": "3037115", "postDate": "11/05/2024 11:07:54", "content": "I am new to Kaggle and its competitions. I have submitted the notebook three times, but it always says 'Your submission notebook may not have started the inference server that is called to obtain predictions. This could mean you forgot to start it, or the notebook crashed.' after the submission. What does it mean? And what can I do to avoid it? The scoring lasts for 10 minutes or more before failure.
", "rawMarkdown": "I am new to Kaggle and its competitions. I have submitted the notebook three times, but it always says 'Your submission notebook may not have started the inference server that is called to obtain predictions. This could mean you forgot to start it, or the notebook crashed.' after the submission. What does it mean? And what can I do to avoid it? The scoring lasts for 10 minutes or more before failure.", "votes": null }, { "id": "3037142", "postDate": "11/05/2024 12:03:54", "content": "This issue is likely due to not properly setting up the inference server with the following code:
\ninference_server = kaggle_evaluation.jane_street_inference_server.JSInferenceServer(predict)\n\n os.getenv():\n inference_server.serve()\n:\n inference_server.run_local_gateway(\n (\n ,\n ,\n )\n )\nI recommend following the official submission guidelines to avoid this issue.
", "rawMarkdown": "This issue is likely due to not properly setting up the inference server with the following code:\n\n```python\ninference_server = kaggle_evaluation.jane_street_inference_server.JSInferenceServer(predict)\n\nif os.getenv('KAGGLE_IS_COMPETITION_RERUN'):\n inference_server.serve()\nelse:\n inference_server.run_local_gateway(\n (\n '/kaggle/input/jane-street-real-time-market-data-forecasting/test.parquet',\n '/kaggle/input/jane-street-real-time-market-data-forecasting/lags.parquet',\n )\n )\n```\n\nI recommend following the official submission guidelines to avoid this issue.", "votes": null }, { "id": "3037275", "postDate": "11/05/2024 14:54:07", "content": "Thank you for your advice! I have just found a strange thing. The submission can be done if I use less num_epochs. However, when the num_epochs is huge, the submission will fail. I then suspect that the notebook should not run for long?
", "rawMarkdown": "Thank you for your advice! I have just found a strange thing. The submission can be done if I use less num_epochs. However, when the num_epochs is huge, the submission will fail. I then suspect that the notebook should not run for long?", "votes": null }, { "id": "3037281", "postDate": "11/05/2024 14:59:29", "content": "And I have just checked my notebook. I have already written the same code at the end of the notebook. It just can not work normally.
", "rawMarkdown": "And I have just checked my notebook. I have already written the same code at the end of the notebook. It just can not work normally.", "votes": null }, { "id": "3037286", "postDate": "11/05/2024 15:07:29", "content": "Yes, that's correct. For the final submission, there's a time limit, so it's a good idea to test your predictions using the test.parquet file in the dataset. Ideally, aim for a prediction time under 0.15 seconds on this dataset to help ensure a smooth submission in the notebook.
", "rawMarkdown": "Yes, that's correct. For the final submission, there's a time limit, so it's a good idea to test your predictions using the test.parquet file in the dataset. Ideally, aim for a prediction time under 0.15 seconds on this dataset to help ensure a smooth submission in the notebook.", "votes": null }, { "id": "3037287", "postDate": "11/05/2024 15:07:36", "content": "You're likely training the model before starting the server. There's the following note in the demo submission notebook:
\n\n", "rawMarkdown": "You're likely training the model before starting the server. There's the following note in the [demo submission notebook](https://www.kaggle.com/code/ryanholbrook/jane-street-rmf-demo-submission):\n\n>When your notebook is run on the hidden test set, inference_server.serve must be called within 15 minutes of the notebook starting or the gateway will throw an error. If you need more than 15 minutes to load your model you can do so during the very first predict call, which does not have the usual 1 minute response deadline.", "votes": null }, { "id": "3037292", "postDate": "11/05/2024 15:11:44", "content": "When your notebook is run on the hidden test set, inference_server.serve must be called within 15 minutes of the notebook starting or the gateway will throw an error. If you need more than 15 minutes to load your model you can do so during the very first predict call, which does not have the usual 1 minute response deadline.
\n
In fact, I haven't personally encountered this issue myself. Most of the challenges I face are within the '''predict''' function. My suggestion is to simply copy & edit the official notebook as a base for submission. This way, you can avoid some technical errors and focus more on the model itself.
", "rawMarkdown": "In fact, I haven't personally encountered this issue myself. Most of the challenges I face are within the '''predict''' function. My suggestion is to simply copy & edit the official notebook as a base for submission. This way, you can avoid some technical errors and focus more on the model itself.", "votes": null }, { "id": "3037297", "postDate": "11/05/2024 15:13:46", "content": "Thank you for your advice!
", "rawMarkdown": "Thank you for your advice!", "votes": null }, { "id": "3037300", "postDate": "11/05/2024 15:15:24", "content": "Thank you!
", "rawMarkdown": "Thank you!", "votes": null } ], "comments": [ { "id": 3037142, "author_name": "xuliangxu", "author_url": "", "post_date": "11/05/2024 12:03:54", "content": "This issue is likely due to not properly setting up the inference server with the following code:
\ninference_server = kaggle_evaluation.jane_street_inference_server.JSInferenceServer(predict)\n\n os.getenv():\n inference_server.serve()\n:\n inference_server.run_local_gateway(\n (\n ,\n ,\n )\n )\nI recommend following the official submission guidelines to avoid this issue.
", "votes": null, "replies": [ { "id": 3037275, "author_name": "candyshan", "author_url": "", "post_date": "11/05/2024 14:54:07", "content": "Thank you for your advice! I have just found a strange thing. The submission can be done if I use less num_epochs. However, when the num_epochs is huge, the submission will fail. I then suspect that the notebook should not run for long?
", "votes": null, "replies": [ { "id": 3037286, "author_name": "xuliangxu", "author_url": "", "post_date": "11/05/2024 15:07:29", "content": "Yes, that's correct. For the final submission, there's a time limit, so it's a good idea to test your predictions using the test.parquet file in the dataset. Ideally, aim for a prediction time under 0.15 seconds on this dataset to help ensure a smooth submission in the notebook.
", "votes": null, "replies": [] }, { "id": 3037287, "author_name": "shuthdar", "author_url": "", "post_date": "11/05/2024 15:07:36", "content": "You're likely training the model before starting the server. There's the following note in the demo submission notebook:
\n\n", "votes": null, "replies": [ { "id": 3037300, "author_name": "candyshan", "author_url": "", "post_date": "11/05/2024 15:15:24", "content": "When your notebook is run on the hidden test set, inference_server.serve must be called within 15 minutes of the notebook starting or the gateway will throw an error. If you need more than 15 minutes to load your model you can do so during the very first predict call, which does not have the usual 1 minute response deadline.
\n
Thank you!
", "votes": null, "replies": [] } ] } ] }, { "id": 3037281, "author_name": "candyshan", "author_url": "", "post_date": "11/05/2024 14:59:29", "content": "And I have just checked my notebook. I have already written the same code at the end of the notebook. It just can not work normally.
", "votes": null, "replies": [ { "id": 3037292, "author_name": "xuliangxu", "author_url": "", "post_date": "11/05/2024 15:11:44", "content": "In fact, I haven't personally encountered this issue myself. Most of the challenges I face are within the '''predict''' function. My suggestion is to simply copy & edit the official notebook as a base for submission. This way, you can avoid some technical errors and focus more on the model itself.
", "votes": null, "replies": [ { "id": 3037297, "author_name": "candyshan", "author_url": "", "post_date": "11/05/2024 15:13:46", "content": "Thank you for your advice!
", "votes": null, "replies": [] } ] } ] } ] } ], "raw_markdown_by_id": { "3037115": "I am new to Kaggle and its competitions. I have submitted the notebook three times, but it always says 'Your submission notebook may not have started the inference server that is called to obtain predictions. This could mean you forgot to start it, or the notebook crashed.' after the submission. What does it mean? And what can I do to avoid it? The scoring lasts for 10 minutes or more before failure.", "3037142": "This issue is likely due to not properly setting up the inference server with the following code:\n\n```python\ninference_server = kaggle_evaluation.jane_street_inference_server.JSInferenceServer(predict)\n\nif os.getenv('KAGGLE_IS_COMPETITION_RERUN'):\n inference_server.serve()\nelse:\n inference_server.run_local_gateway(\n (\n '/kaggle/input/jane-street-real-time-market-data-forecasting/test.parquet',\n '/kaggle/input/jane-street-real-time-market-data-forecasting/lags.parquet',\n )\n )\n```\n\nI recommend following the official submission guidelines to avoid this issue.", "3037275": "Thank you for your advice! I have just found a strange thing. The submission can be done if I use less num_epochs. However, when the num_epochs is huge, the submission will fail. I then suspect that the notebook should not run for long?", "3037281": "And I have just checked my notebook. I have already written the same code at the end of the notebook. It just can not work normally.", "3037286": "Yes, that's correct. For the final submission, there's a time limit, so it's a good idea to test your predictions using the test.parquet file in the dataset. Ideally, aim for a prediction time under 0.15 seconds on this dataset to help ensure a smooth submission in the notebook.", "3037287": "You're likely training the model before starting the server. There's the following note in the [demo submission notebook](https://www.kaggle.com/code/ryanholbrook/jane-street-rmf-demo-submission):\n\n>When your notebook is run on the hidden test set, inference_server.serve must be called within 15 minutes of the notebook starting or the gateway will throw an error. If you need more than 15 minutes to load your model you can do so during the very first predict call, which does not have the usual 1 minute response deadline.", "3037292": "In fact, I haven't personally encountered this issue myself. Most of the challenges I face are within the '''predict''' function. My suggestion is to simply copy & edit the official notebook as a base for submission. This way, you can avoid some technical errors and focus more on the model itself.", "3037297": "Thank you for your advice!", "3037300": "Thank you!" }, "source": "meta" }