{ "id": 107642, "title": "Any last minute tips ??", "url": "/competitions/aptos2019-blindness-detection/discussion/107642", "author_name": "", "post_date": "2019-09-05T14:08:51.784416600Z", "votes": 1, "comment_count": 9, "views": 0, "content": "", "messages": [ { "id": "618806", "postDate": "09/05/2019 14:08:51", "content": "", "rawMarkdown": "", "votes": null }, { "id": "618838", "postDate": "09/05/2019 14:50:37", "content": "

Cross your fingers :)

", "rawMarkdown": "Cross your fingers :)", "votes": null }, { "id": "618840", "postDate": "09/05/2019 14:54:21", "content": "

Here take mine =) \n\"\"

", "rawMarkdown": "Here take mine =) \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F991320%2F2e52d06838d04317f0b3bf6e36f4bf9f%2Fsaving-tips-jar.jpg?generation=1567695249174064&alt=media)", "votes": null }, { "id": "618977", "postDate": "09/05/2019 17:44:00", "content": "

Make sure you have an inference kernel takes less than 34 minutes to run in case the Kaggle performance issues will not be resolved by the end of the competition (see @sheriytm's comment here).\nUnless someone can report he/she was able to successfully submit an inference kernel which takes longer than 34 minutes to commit?\nDouble-check even kernels that were successfuly submitted in the past (prior to the performance issues).

", "rawMarkdown": "Make sure you have an inference kernel takes less than 34 minutes to run in case the Kaggle performance issues will not be resolved by the end of the competition (see @sheriytm's comment [here](https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/107323#latest-617204)).\nUnless someone can report he/she was able to successfully submit an inference kernel which takes longer than 34 minutes to commit?\nDouble-check even kernels that were successfuly submitted in the past (prior to the performance issues).", "votes": null }, { "id": "619057", "postDate": "09/05/2019 19:43:29", "content": "

Will our kernels be rerun? Doesn't synchronous kernel competition mean it has already been run on the unseen dataset ?

", "rawMarkdown": "Will our kernels be rerun? Doesn't synchronous kernel competition mean it has already been run on the unseen dataset ?", "votes": null }, { "id": "619090", "postDate": "09/05/2019 20:10:57", "content": "

Actually I think you are right. I just re-read the blog post announcing the synchronous kernel format and it states:

\n\n
\n

Why Synchronous Kernels-only Competitions?\n We’re now pleased to announce general support for a synchronous Kernels-only format. When you submit from a Kernel, Kaggle will run the code against both the public test set and private test set in real-time. This small-but-substantial tweak improves the experience for participants, the host, and Kaggle:

\n \n \n
", "rawMarkdown": "Actually I think you are right. I just re-read the [blog post announcing the synchronous kernel format](http://blog.kaggle.com/2019/05/17/synchronous-kernels-only-competitions-real-ml-in-real-time/) and it states:\n> **Why Synchronous Kernels-only Competitions**?\nWe’re now pleased to announce general support for a synchronous Kernels-only format. When you submit from a Kernel, Kaggle will run the code against both the public test set and private test set in real-time. This small-but-substantial tweak improves the experience for participants, the host, and Kaggle:\n\n> - With a truly withheld test set, we are practicing proper, rigorous machine learning.\n> - We will be able to offer more varieties of competitions and intend to run many fewer confusing two-stage competitions.\n> - You will be able to see if your code runs successfully on the withheld test set and have the leeway to intervene if it fails.\n> - We will run all submissions against the private data, not just selected ones. Participants will get the complete and familiar public/private scores available in a traditional competition.\n> - **The final leaderboard can be released at the end of the competition, without the delay of rerunning Kernels**.`", "votes": null }, { "id": "619092", "postDate": "09/05/2019 20:14:52", "content": "

select 2 submissions, is important

", "rawMarkdown": "select 2 submissions, is important", "votes": null }, { "id": "619345", "postDate": "09/06/2019 05:25:41", "content": "

I guess your top LB score will not be the best private score

", "rawMarkdown": "I guess your top LB score will not be the best private score", "votes": null }, { "id": "619359", "postDate": "09/06/2019 05:51:42", "content": "

Role a dice and pick your 2 submissions.

", "rawMarkdown": "Role a dice and pick your 2 submissions.", "votes": null }, { "id": "619360", "postDate": "09/06/2019 05:51:50", "content": "

Pray that there is no substantial shake-up πŸ™ πŸ™

", "rawMarkdown": "Pray that there is no substantial shake-up πŸ™ πŸ™", "votes": null } ], "comments": [ { "id": 618838, "author_name": "valanm", "author_url": "", "post_date": "09/05/2019 14:50:37", "content": "

Cross your fingers :)

", "votes": null, "replies": [] }, { "id": 618840, "author_name": "drhabib", "author_url": "", "post_date": "09/05/2019 14:54:21", "content": "

Here take mine =) \n\"\"

", "votes": null, "replies": [] }, { "id": 618977, "author_name": "pyotam", "author_url": "", "post_date": "09/05/2019 17:44:00", "content": "

Make sure you have an inference kernel takes less than 34 minutes to run in case the Kaggle performance issues will not be resolved by the end of the competition (see @sheriytm's comment here).\nUnless someone can report he/she was able to successfully submit an inference kernel which takes longer than 34 minutes to commit?\nDouble-check even kernels that were successfuly submitted in the past (prior to the performance issues).

", "votes": null, "replies": [ { "id": 619057, "author_name": "saurabh7", "author_url": "", "post_date": "09/05/2019 19:43:29", "content": "

Will our kernels be rerun? Doesn't synchronous kernel competition mean it has already been run on the unseen dataset ?

", "votes": null, "replies": [] }, { "id": 619090, "author_name": "pyotam", "author_url": "", "post_date": "09/05/2019 20:10:57", "content": "

Actually I think you are right. I just re-read the blog post announcing the synchronous kernel format and it states:

\n\n
\n

Why Synchronous Kernels-only Competitions?\n We’re now pleased to announce general support for a synchronous Kernels-only format. When you submit from a Kernel, Kaggle will run the code against both the public test set and private test set in real-time. This small-but-substantial tweak improves the experience for participants, the host, and Kaggle:

\n \n \n
", "votes": null, "replies": [] } ] }, { "id": 619092, "author_name": "jesucristo", "author_url": "", "post_date": "09/05/2019 20:14:52", "content": "

select 2 submissions, is important

", "votes": null, "replies": [] }, { "id": 619345, "author_name": "kupchanski", "author_url": "", "post_date": "09/06/2019 05:25:41", "content": "

I guess your top LB score will not be the best private score

", "votes": null, "replies": [] }, { "id": 619359, "author_name": "naivelamb", "author_url": "", "post_date": "09/06/2019 05:51:42", "content": "

Role a dice and pick your 2 submissions.

", "votes": null, "replies": [] }, { "id": 619360, "author_name": "bibek777", "author_url": "", "post_date": "09/06/2019 05:51:50", "content": "

Pray that there is no substantial shake-up πŸ™ πŸ™

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "618806": "", "618838": "Cross your fingers :)", "618840": "Here take mine =) \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F991320%2F2e52d06838d04317f0b3bf6e36f4bf9f%2Fsaving-tips-jar.jpg?generation=1567695249174064&alt=media)", "618977": "Make sure you have an inference kernel takes less than 34 minutes to run in case the Kaggle performance issues will not be resolved by the end of the competition (see @sheriytm's comment [here](https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/107323#latest-617204)).\nUnless someone can report he/she was able to successfully submit an inference kernel which takes longer than 34 minutes to commit?\nDouble-check even kernels that were successfuly submitted in the past (prior to the performance issues).", "619057": "Will our kernels be rerun? Doesn't synchronous kernel competition mean it has already been run on the unseen dataset ?", "619090": "Actually I think you are right. I just re-read the [blog post announcing the synchronous kernel format](http://blog.kaggle.com/2019/05/17/synchronous-kernels-only-competitions-real-ml-in-real-time/) and it states:\n> **Why Synchronous Kernels-only Competitions**?\nWe’re now pleased to announce general support for a synchronous Kernels-only format. When you submit from a Kernel, Kaggle will run the code against both the public test set and private test set in real-time. This small-but-substantial tweak improves the experience for participants, the host, and Kaggle:\n\n> - With a truly withheld test set, we are practicing proper, rigorous machine learning.\n> - We will be able to offer more varieties of competitions and intend to run many fewer confusing two-stage competitions.\n> - You will be able to see if your code runs successfully on the withheld test set and have the leeway to intervene if it fails.\n> - We will run all submissions against the private data, not just selected ones. Participants will get the complete and familiar public/private scores available in a traditional competition.\n> - **The final leaderboard can be released at the end of the competition, without the delay of rerunning Kernels**.`", "619092": "select 2 submissions, is important", "619345": "I guess your top LB score will not be the best private score", "619359": "Role a dice and pick your 2 submissions.", "619360": "Pray that there is no substantial shake-up πŸ™ πŸ™" }, "source": "meta" }