As always, the time series API has memory issues. This time, there are also time-related challenges. Therefore, as I created a notebook explaining the new time series API, which served as the basis for building this simulator.
\nhttps://www.kaggle.com/code/chumajin/janestreet-simulator-for-time-series-api
\nYou can use it by simply changing the period in #1 and the predict function in #3 in the public notebook to your own!
\nIf there are any mistakes, I would appreciate it if you could point them out.
\n-- updated 2024125 --
\nFrom this topic, I remake this simulator using the evaluation API directly. Notebook has been updated. But, not change how to use this.
\nEnjoy this competition ~ .
", "messages": [ { "id": "3023431", "postDate": "10/20/2024 14:35:46", "content": "As always, the time series API has memory issues. This time, there are also time-related challenges. Therefore, as I created a notebook explaining the new time series API, which served as the basis for building this simulator.
\nhttps://www.kaggle.com/code/chumajin/janestreet-simulator-for-time-series-api
\nYou can use it by simply changing the period in #1 and the predict function in #3 in the public notebook to your own!
\nIf there are any mistakes, I would appreciate it if you could point them out.
\n-- updated 2024125 --
\nFrom this topic, I remake this simulator using the evaluation API directly. Notebook has been updated. But, not change how to use this.
\nEnjoy this competition ~ .
", "rawMarkdown": "As always, the time series API has memory issues. This time, there are also time-related challenges. Therefore, as I created a notebook explaining the new time series API, which served as the basis for building this simulator.\n\nhttps://www.kaggle.com/code/chumajin/janestreet-simulator-for-time-series-api\n\nYou can use it by simply changing the period in #1 and the predict function in #3 in the public notebook to your own!\n\nIf there are any mistakes, I would appreciate it if you could point them out.\n\n\n-- updated 2024125 --\n\nFrom this [topic](https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/542022#3026354), I remake this simulator using the evaluation API directly. Notebook has been updated. But, not change how to use this. \n\nEnjoy this competition ~ .", "votes": null }, { "id": "3023920", "postDate": "10/21/2024 06:56:34", "content": "Thank you @chumajin. Your work is very insightful. For my TF model the approximate time is close to the true submission time. I think it can help us to control timeout issues.
", "rawMarkdown": "Thank you @chumajin. Your work is very insightful. For my TF model the approximate time is close to the true submission time. I think it can help us to control timeout issues.", "votes": null }, { "id": "3023984", "postDate": "10/21/2024 08:18:47", "content": "@ulrich07 Thanks for the more information! I'm glad it works well!
", "rawMarkdown": "ulrich07 Thanks for the more information! I'm glad it works well!", "votes": null }, { "id": "3024103", "postDate": "10/21/2024 10:30:40", "content": "I hope kaggle could add more dummy data for debug purpose … Currently, there is only 1 time_id of data.
", "rawMarkdown": "I hope kaggle could add more dummy data for debug purpose ... Currently, there is only 1 time_id of data.", "votes": null }, { "id": "3024245", "postDate": "10/21/2024 13:08:15", "content": "I have made a kernel using the given train data to generate synthetic test for debugging. I hope it will be helpful for you. Link: https://www.kaggle.com/code/shiyili/js24-rmf-generate-synthetic-test-data
", "rawMarkdown": "I have made a kernel using the given train data to generate synthetic test for debugging. I hope it will be helpful for you. Link: https://www.kaggle.com/code/shiyili/js24-rmf-generate-synthetic-test-data", "votes": null }, { "id": "3024246", "postDate": "10/21/2024 13:08:57", "content": "here is how I actually use it in my only inference pipeline: https://www.kaggle.com/code/shiyili/js2024-rmf-gru-inference
", "rawMarkdown": "here is how I actually use it in my only inference pipeline: https://www.kaggle.com/code/shiyili/js2024-rmf-gru-inference", "votes": null }, { "id": "3024274", "postDate": "10/21/2024 13:30:22", "content": "@yuanzhezhou You are absolutely right. In this regard, the previous competition version was better, wasn't it…
\nBecause of this, I created a simulation-version.
@shiyili It's similar to mine! Thank you for sharing.
", "rawMarkdown": "shiyili It's similar to mine! Thank you for sharing.", "votes": null }, { "id": "3040115", "postDate": "11/08/2024 17:47:24", "content": "Thank you for providing this. It helps a lot it local debugging!!!
", "rawMarkdown": "Thank you for providing this. It helps a lot it local debugging!!!", "votes": null } ], "comments": [ { "id": 3023920, "author_name": "ulrich07", "author_url": "", "post_date": "10/21/2024 06:56:34", "content": "Thank you @chumajin. Your work is very insightful. For my TF model the approximate time is close to the true submission time. I think it can help us to control timeout issues.
", "votes": null, "replies": [ { "id": 3023984, "author_name": "chumajin", "author_url": "", "post_date": "10/21/2024 08:18:47", "content": "@ulrich07 Thanks for the more information! I'm glad it works well!
", "votes": null, "replies": [] } ] }, { "id": 3024103, "author_name": "yuanzhezhou", "author_url": "", "post_date": "10/21/2024 10:30:40", "content": "I hope kaggle could add more dummy data for debug purpose … Currently, there is only 1 time_id of data.
", "votes": null, "replies": [ { "id": 3024245, "author_name": "shiyili", "author_url": "", "post_date": "10/21/2024 13:08:15", "content": "I have made a kernel using the given train data to generate synthetic test for debugging. I hope it will be helpful for you. Link: https://www.kaggle.com/code/shiyili/js24-rmf-generate-synthetic-test-data
", "votes": null, "replies": [ { "id": 3024246, "author_name": "shiyili", "author_url": "", "post_date": "10/21/2024 13:08:57", "content": "here is how I actually use it in my only inference pipeline: https://www.kaggle.com/code/shiyili/js2024-rmf-gru-inference
", "votes": null, "replies": [ { "id": 3024276, "author_name": "chumajin", "author_url": "", "post_date": "10/21/2024 13:31:01", "content": "@shiyili It's similar to mine! Thank you for sharing.
", "votes": null, "replies": [] } ] } ] }, { "id": 3024274, "author_name": "chumajin", "author_url": "", "post_date": "10/21/2024 13:30:22", "content": "@yuanzhezhou You are absolutely right. In this regard, the previous competition version was better, wasn't it…
\nBecause of this, I created a simulation-version.
Thank you for providing this. It helps a lot it local debugging!!!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3023431": "As always, the time series API has memory issues. This time, there are also time-related challenges. Therefore, as I created a notebook explaining the new time series API, which served as the basis for building this simulator.\n\nhttps://www.kaggle.com/code/chumajin/janestreet-simulator-for-time-series-api\n\nYou can use it by simply changing the period in #1 and the predict function in #3 in the public notebook to your own!\n\nIf there are any mistakes, I would appreciate it if you could point them out.\n\n\n-- updated 2024125 --\n\nFrom this [topic](https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/542022#3026354), I remake this simulator using the evaluation API directly. Notebook has been updated. But, not change how to use this. \n\nEnjoy this competition ~ .", "3023920": "Thank you @chumajin. Your work is very insightful. For my TF model the approximate time is close to the true submission time. I think it can help us to control timeout issues.", "3023984": "ulrich07 Thanks for the more information! I'm glad it works well!", "3024103": "I hope kaggle could add more dummy data for debug purpose ... Currently, there is only 1 time_id of data.", "3024245": "I have made a kernel using the given train data to generate synthetic test for debugging. I hope it will be helpful for you. Link: https://www.kaggle.com/code/shiyili/js24-rmf-generate-synthetic-test-data", "3024246": "here is how I actually use it in my only inference pipeline: https://www.kaggle.com/code/shiyili/js2024-rmf-gru-inference", "3024274": "yuanzhezhou You are absolutely right. In this regard, the previous competition version was better, wasn't it...\nBecause of this, I created a simulation-version.", "3024276": "shiyili It's similar to mine! Thank you for sharing.", "3040115": "Thank you for providing this. It helps a lot it local debugging!!!" }, "source": "meta" }