{ "id": 318793, "title": "Fifth score update", "url": "/competitions/g-research-crypto-forecasting/discussion/318793", "author_name": "", "post_date": "2022-04-13T22:48:21.738228400Z", "votes": 12, "comment_count": 16, "views": 0, "content": "

I've just published the fifth score update. It may take several minutes for all entries to show up on the leaderboard.

\n

Please note that this is the last interim update. The next score update will be the final update published when the competition closes.

\n

There was a higher failure rate than in previous runs (roughly 9%). I'm going to conduct a review to confirm nothing unexpected happened but will most likely not have any updates until early next week.

", "messages": [ { "id": "1754702", "postDate": "04/13/2022 22:48:21", "content": "

I've just published the fifth score update. It may take several minutes for all entries to show up on the leaderboard.

\n

Please note that this is the last interim update. The next score update will be the final update published when the competition closes.

\n

There was a higher failure rate than in previous runs (roughly 9%). I'm going to conduct a review to confirm nothing unexpected happened but will most likely not have any updates until early next week.

", "rawMarkdown": "I've just published the fifth score update. It may take several minutes for all entries to show up on the leaderboard. \n\nPlease note that this is the last interim update. The next score update will be the final update published when the competition closes.\n\nThere was a higher failure rate than in previous runs (roughly 9%). I'm going to conduct a review to confirm nothing unexpected happened but will most likely not have any updates until early next week.", "votes": null }, { "id": "1754708", "postDate": "04/13/2022 23:01:08", "content": "

Thanks for the job.

", "rawMarkdown": "Thanks for the job.", "votes": null }, { "id": "1754710", "postDate": "04/13/2022 23:06:27", "content": "

I guess the higher rate of failure is caused by the increased data and therefore also increased runtimes. I wonder how this will affect the final reruns. Speaking of which, there are 3 weeks left, how many reruns will there be until the final update?

", "rawMarkdown": "I guess the higher rate of failure is caused by the increased data and therefore also increased runtimes. I wonder how this will affect the final reruns. Speaking of which, there are 3 weeks left, how many reruns will there be until the final update?", "votes": null }, { "id": "1754919", "postDate": "04/14/2022 06:21:19", "content": "

Is it possible to fail due to more data in the final batch? My submissions never gets an error up to this point but I feat if it fails in the last batch :)

", "rawMarkdown": "Is it possible to fail due to more data in the final batch? My submissions never gets an error up to this point but I feat if it fails in the last batch :)", "votes": null }, { "id": "1755061", "postDate": "04/14/2022 09:23:56", "content": "

🙏I am still holding very little hope that our second submission could work in the last re-run… Looking forward to the results of your review. Thanks, Sohier!

", "rawMarkdown": "🙏I am still holding very little hope that our second submission could work in the last re-run... Looking forward to the results of your review. Thanks, Sohier!", "votes": null }, { "id": "1755503", "postDate": "04/14/2022 17:10:29", "content": "

Good question, I've added a note to the post. The next update will be the final one in three weeks.

", "rawMarkdown": "Good question, I've added a note to the post. The next update will be the final one in three weeks.", "votes": null }, { "id": "1758677", "postDate": "04/18/2022 00:02:18", "content": "

Was I removed from this competition?

", "rawMarkdown": "Was I removed from this competition?", "votes": null }, { "id": "1759174", "postDate": "04/18/2022 11:59:05", "content": "

Excuse me. Both of my submissions scores -1 after this update. But they all had normal scores during the past four updates. And one was in silver zone, the other in bronze zone.
\nCould you check if there are some mistakes happen in this update that can lead to my -1 score. As the behavior of my codes won't change so much if other conditions haven't change.
\nIt can't be the increased data. Because my submission didn't turn out with time error. My guess is something can lead to nan occur in this update. My submission will predict constant value if input with nan.

", "rawMarkdown": "Excuse me. Both of my submissions scores -1 after this update. But they all had normal scores during the past four updates. And one was in silver zone, the other in bronze zone.\nCould you check if there are some mistakes happen in this update that can lead to my -1 score. As the behavior of my codes won't change so much if other conditions haven't change.\nIt can't be the increased data. Because my submission didn't turn out with time error. My guess is something can lead to nan occur in this update. My submission will predict constant value if input with nan.", "votes": null }, { "id": "1759312", "postDate": "04/18/2022 14:25:13", "content": "

So it looks like this competition is really hard to win. There are a lot of entries with high scores. I am curious to see the final update.

", "rawMarkdown": "So it looks like this competition is really hard to win. There are a lot of entries with high scores. I am curious to see the final update.", "votes": null }, { "id": "1762623", "postDate": "04/20/2022 20:39:47", "content": "

I took a look at the failures and was unable to find any root cause other than the increasingly large test set. Most of the new errors boiled down to the notebook either running out of memory or hitting the time limit.

", "rawMarkdown": "I took a look at the failures and was unable to find any root cause other than the increasingly large test set. Most of the new errors boiled down to the notebook either running out of memory or hitting the time limit.", "votes": null }, { "id": "1762973", "postDate": "04/21/2022 06:44:57", "content": "

Thanks for the effort.

\n

Would it be possible to publish an updated supplemental_train.csv?

\n

With this, we could follow the competition and see where we are failing.

\n

Calculating the pearson correlation coefficient over a large and changing sample may be more unstable than calculating the average C.Pearson for each week. Perhaps that is why some participants have gone down or up a lot in the last week.

\n

Best regards

", "rawMarkdown": "Thanks for the effort.\n\nWould it be possible to publish an updated supplemental_train.csv? \n\nWith this, we could follow the competition and see where we are failing.\n\nCalculating the pearson correlation coefficient over a large and changing sample may be more unstable than calculating the average C.Pearson for each week. Perhaps that is why some participants have gone down or up a lot in the last week.\n\nBest regards", "votes": null }, { "id": "1763041", "postDate": "04/21/2022 08:20:50", "content": "

In this case, are you going to relax the time limitation or keep it?

", "rawMarkdown": "In this case, are you going to relax the time limitation or keep it?", "votes": null }, { "id": "1763246", "postDate": "04/21/2022 11:34:13", "content": "

It seems the number of teams is decreased ? Why? Is this my misunderstanding ?

", "rawMarkdown": "It seems the number of teams is decreased ? Why? Is this my misunderstanding ?", "votes": null }, { "id": "1763970", "postDate": "04/22/2022 02:44:52", "content": "

@sohier Kaggle should implement the ability to see how long a submission took to run. We are given how long a notebook takes to run but that doesn't include the submission api. Also, maybe include what the max ram reached was. This would ensure that less failures occur and wouldn't take too long to implement.

", "rawMarkdown": "sohier Kaggle should implement the ability to see how long a submission took to run. We are given how long a notebook takes to run but that doesn't include the submission api. Also, maybe include what the max ram reached was. This would ensure that less failures occur and wouldn't take too long to implement.", "votes": null }, { "id": "1764599", "postDate": "04/22/2022 16:08:52", "content": "

@petersk20
\nAgree with you. It's very important.

", "rawMarkdown": "petersk20 \nAgree with you. It's very important.", "votes": null }, { "id": "1767994", "postDate": "04/25/2022 21:13:26", "content": "

@sohier

\n

Will the rank of medals be according to the competitors who finished or those who will finish?

", "rawMarkdown": "sohier \n\nWill the rank of medals be according to the competitors who finished or those who will finish?", "votes": null }, { "id": "1768887", "postDate": "04/26/2022 18:27:48", "content": "

Hope you can bless us with a blog post on your solution! 😃 Did your team use some variation of the XGB+MLP+AE you are known for?

", "rawMarkdown": "Hope you can bless us with a blog post on your solution! 😃 Did your team use some variation of the XGB+MLP+AE you are known for?", "votes": null } ], "comments": [ { "id": 1754708, "author_name": "zhongtian93", "author_url": "", "post_date": "04/13/2022 23:01:08", "content": "

Thanks for the job.

", "votes": null, "replies": [] }, { "id": 1754710, "author_name": "fritzcremer", "author_url": "", "post_date": "04/13/2022 23:06:27", "content": "

I guess the higher rate of failure is caused by the increased data and therefore also increased runtimes. I wonder how this will affect the final reruns. Speaking of which, there are 3 weeks left, how many reruns will there be until the final update?

", "votes": null, "replies": [ { "id": 1755503, "author_name": "sohier", "author_url": "", "post_date": "04/14/2022 17:10:29", "content": "

Good question, I've added a note to the post. The next update will be the final one in three weeks.

", "votes": null, "replies": [] }, { "id": 1767994, "author_name": "coreacasa", "author_url": "", "post_date": "04/25/2022 21:13:26", "content": "

@sohier

\n

Will the rank of medals be according to the competitors who finished or those who will finish?

", "votes": null, "replies": [] } ] }, { "id": 1754919, "author_name": "bturan19", "author_url": "", "post_date": "04/14/2022 06:21:19", "content": "

Is it possible to fail due to more data in the final batch? My submissions never gets an error up to this point but I feat if it fails in the last batch :)

", "votes": null, "replies": [] }, { "id": 1755061, "author_name": "gogo827jz", "author_url": "", "post_date": "04/14/2022 09:23:56", "content": "

🙏I am still holding very little hope that our second submission could work in the last re-run… Looking forward to the results of your review. Thanks, Sohier!

", "votes": null, "replies": [ { "id": 1768887, "author_name": "valindor", "author_url": "", "post_date": "04/26/2022 18:27:48", "content": "

Hope you can bless us with a blog post on your solution! 😃 Did your team use some variation of the XGB+MLP+AE you are known for?

", "votes": null, "replies": [] } ] }, { "id": 1758677, "author_name": "valindor", "author_url": "", "post_date": "04/18/2022 00:02:18", "content": "

Was I removed from this competition?

", "votes": null, "replies": [] }, { "id": 1759174, "author_name": "littlewc", "author_url": "", "post_date": "04/18/2022 11:59:05", "content": "

Excuse me. Both of my submissions scores -1 after this update. But they all had normal scores during the past four updates. And one was in silver zone, the other in bronze zone.
\nCould you check if there are some mistakes happen in this update that can lead to my -1 score. As the behavior of my codes won't change so much if other conditions haven't change.
\nIt can't be the increased data. Because my submission didn't turn out with time error. My guess is something can lead to nan occur in this update. My submission will predict constant value if input with nan.

", "votes": null, "replies": [] }, { "id": 1759312, "author_name": "", "author_url": "", "post_date": "04/18/2022 14:25:13", "content": "

So it looks like this competition is really hard to win. There are a lot of entries with high scores. I am curious to see the final update.

", "votes": null, "replies": [] }, { "id": 1762623, "author_name": "sohier", "author_url": "", "post_date": "04/20/2022 20:39:47", "content": "

I took a look at the failures and was unable to find any root cause other than the increasingly large test set. Most of the new errors boiled down to the notebook either running out of memory or hitting the time limit.

", "votes": null, "replies": [ { "id": 1763041, "author_name": "gogo827jz", "author_url": "", "post_date": "04/21/2022 08:20:50", "content": "

In this case, are you going to relax the time limitation or keep it?

", "votes": null, "replies": [] } ] }, { "id": 1762973, "author_name": "ricardocsag", "author_url": "", "post_date": "04/21/2022 06:44:57", "content": "

Thanks for the effort.

\n

Would it be possible to publish an updated supplemental_train.csv?

\n

With this, we could follow the competition and see where we are failing.

\n

Calculating the pearson correlation coefficient over a large and changing sample may be more unstable than calculating the average C.Pearson for each week. Perhaps that is why some participants have gone down or up a lot in the last week.

\n

Best regards

", "votes": null, "replies": [] }, { "id": 1763246, "author_name": "meisa0", "author_url": "", "post_date": "04/21/2022 11:34:13", "content": "

It seems the number of teams is decreased ? Why? Is this my misunderstanding ?

", "votes": null, "replies": [] }, { "id": 1763970, "author_name": "petersk20", "author_url": "", "post_date": "04/22/2022 02:44:52", "content": "

@sohier Kaggle should implement the ability to see how long a submission took to run. We are given how long a notebook takes to run but that doesn't include the submission api. Also, maybe include what the max ram reached was. This would ensure that less failures occur and wouldn't take too long to implement.

", "votes": null, "replies": [] }, { "id": 1764599, "author_name": "naby96", "author_url": "", "post_date": "04/22/2022 16:08:52", "content": "

@petersk20
\nAgree with you. It's very important.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1754702": "I've just published the fifth score update. It may take several minutes for all entries to show up on the leaderboard. \n\nPlease note that this is the last interim update. The next score update will be the final update published when the competition closes.\n\nThere was a higher failure rate than in previous runs (roughly 9%). I'm going to conduct a review to confirm nothing unexpected happened but will most likely not have any updates until early next week.", "1754708": "Thanks for the job.", "1754710": "I guess the higher rate of failure is caused by the increased data and therefore also increased runtimes. I wonder how this will affect the final reruns. Speaking of which, there are 3 weeks left, how many reruns will there be until the final update?", "1754919": "Is it possible to fail due to more data in the final batch? My submissions never gets an error up to this point but I feat if it fails in the last batch :)", "1755061": "🙏I am still holding very little hope that our second submission could work in the last re-run... Looking forward to the results of your review. Thanks, Sohier!", "1755503": "Good question, I've added a note to the post. The next update will be the final one in three weeks.", "1758677": "Was I removed from this competition?", "1759174": "Excuse me. Both of my submissions scores -1 after this update. But they all had normal scores during the past four updates. And one was in silver zone, the other in bronze zone.\nCould you check if there are some mistakes happen in this update that can lead to my -1 score. As the behavior of my codes won't change so much if other conditions haven't change.\nIt can't be the increased data. Because my submission didn't turn out with time error. My guess is something can lead to nan occur in this update. My submission will predict constant value if input with nan.", "1759312": "So it looks like this competition is really hard to win. There are a lot of entries with high scores. I am curious to see the final update.", "1762623": "I took a look at the failures and was unable to find any root cause other than the increasingly large test set. Most of the new errors boiled down to the notebook either running out of memory or hitting the time limit.", "1762973": "Thanks for the effort.\n\nWould it be possible to publish an updated supplemental_train.csv? \n\nWith this, we could follow the competition and see where we are failing.\n\nCalculating the pearson correlation coefficient over a large and changing sample may be more unstable than calculating the average C.Pearson for each week. Perhaps that is why some participants have gone down or up a lot in the last week.\n\nBest regards", "1763041": "In this case, are you going to relax the time limitation or keep it?", "1763246": "It seems the number of teams is decreased ? Why? Is this my misunderstanding ?", "1763970": "sohier Kaggle should implement the ability to see how long a submission took to run. We are given how long a notebook takes to run but that doesn't include the submission api. Also, maybe include what the max ram reached was. This would ensure that less failures occur and wouldn't take too long to implement.", "1764599": "petersk20 \nAgree with you. It's very important.", "1767994": "sohier \n\nWill the rank of medals be according to the competitors who finished or those who will finish?", "1768887": "Hope you can bless us with a blog post on your solution! 😃 Did your team use some variation of the XGB+MLP+AE you are known for?" }, "source": "meta" }