{ "id": 702750, "title": "Tie at 0.446 for positions 4-6: real or display rounding?", "url": "/competitions/cafa-6-protein-function-prediction/discussion/702750", "author_name": "xaxipiruli", "post_date": "2026-05-26T13:11:49.548000", "votes": 3, "comment_count": 21, "views": 0, "content": "

Quick question on the rescored leaderboard: positions 4, 5 and 6 all display 0.446. Is this a genuine three-way tie at full precision, or just three-decimal display rounding over values that actually differ? Could you expose one or two more decimals for the final standings, or confirm the tie-breaking rule? Thanks!

", "messages": [ { "id": 3463270, "postDate": "2026-05-26T13:11:49.550Z", "content": "

Quick question on the rescored leaderboard: positions 4, 5 and 6 all display 0.446. Is this a genuine three-way tie at full precision, or just three-decimal display rounding over values that actually differ? Could you expose one or two more decimals for the final standings, or confirm the tie-breaking rule? Thanks!

", "rawMarkdown": "Quick question on the rescored leaderboard: positions 4, 5 and 6 all display 0.446. Is this a genuine three-way tie at full precision, or just three-decimal display rounding over values that actually differ? Could you expose one or two more decimals for the final standings, or confirm the tie-breaking rule? Thanks!", "votes": 3 }, { "id": 3463416, "postDate": "2026-05-26T20:55:10.263Z", "content": "

Please check the most current leaderboard, and expect a final post announcing the winners from the Kaggle staff over the next few days. And yes, the leaderboard display was rounded to 3 decimals, the actual value has more digits.

", "rawMarkdown": "Please check the most current leaderboard, and expect a final post announcing the winners from the Kaggle staff over the next few days. And yes, the leaderboard display was rounded to 3 decimals, the actual value has more digits.", "replies": [ { "id": 3463421, "postDate": "2026-05-26T21:05:51.423Z", "content": "

May i ask why are my highest scoring submissions not picked when in the previous rescoring they were picked ?

", "rawMarkdown": "May i ask why are my highest scoring submissions not picked when in the previous rescoring they were picked ?", "votes": 1, "replies": [ { "id": 3463430, "postDate": "2026-05-26T21:13:42.567Z", "content": "

The ones that are currently displayed on the leaderboard were only the highest-scoring 2 submissions per team based on the February evaluation (when submission closes), if your team did not hand-select your top 2. That's why you see only 2 entries per team right now on the leaderboard. Before this current leaderboard, we rescored all submissions of every team so the ranking was different.

", "rawMarkdown": "The ones that are currently displayed on the leaderboard were only the highest-scoring 2 submissions per team based on the February evaluation (when submission closes), if your team did not hand-select your top 2. That's why you see only 2 entries per team right now on the leaderboard. Before this current leaderboard, we rescored all submissions of every team so the ranking was different." }, { "id": 3463485, "postDate": "2026-05-27T02:37:43.910Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3463495, "postDate": "2026-05-27T03:03:20.663Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3463504, "postDate": "2026-05-27T03:33:39.583Z", "content": "

@ahphan What if my selected submissions failed even early rescores and my high scoring submissions were automatically selected by system for 15th may update? Why 2 subs that were selected 15th may were unselected and two random subs were selected again, as you told prev rescore and lb was final?

", "rawMarkdown": " @ahphan What if my selected submissions failed even early rescores and my high scoring submissions were automatically selected by system for 15th may update? Why 2 subs that were selected 15th may were unselected and two random subs were selected again, as you told prev rescore and lb was final?", "votes": 1 }, { "id": 3463511, "postDate": "2026-05-27T04:15:15.480Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3463512, "postDate": "2026-05-27T04:15:42.697Z", "content": "

hi, is the public lb position now the final position? How is the final private position? Is it the same as the private lb last time, or the same as the public lb shown now

", "rawMarkdown": "hi, is the public lb position now the final position? How is the final private position? Is it the same as the private lb last time, or the same as the public lb shown now", "votes": 2 }, { "id": 3463580, "postDate": "2026-05-27T10:07:41.147Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3463581, "postDate": "2026-05-27T10:07:59.743Z", "content": "

Same here. I also doubt that the two currently selected submissions were the ones automatically chosen at the deadline. The only thing I clearly remember is that, during the last week of the competition, I couldn’t beat my own public leaderboard score. So it does seem likely that the two selected submissions came from that period.

\n

But at this point, after so many changes, I can’t really be sure anymore. That’s why I mentioned that the original leaderboard probably should have been preserved given how many things changed afterwards.

", "rawMarkdown": "Same here. I also doubt that the two currently selected submissions were the ones automatically chosen at the deadline. The only thing I clearly remember is that, during the last week of the competition, I couldn’t beat my own public leaderboard score. So it does seem likely that the two selected submissions came from that period.\n\nBut at this point, after so many changes, I can’t really be sure anymore. That’s why I mentioned that the original leaderboard probably should have been preserved given how many things changed afterwards.", "votes": 2 }, { "id": 3463601, "postDate": "2026-05-27T11:57:35.337Z", "content": "

The two currently selected submissions were chosen as follows:

\n\n

The most recent change is because of this:

\n", "rawMarkdown": "The two currently selected submissions were chosen as follows:\n- If your team hand-selected submissions, those 2 submissions remained your final submissions\n- If you hand-selected 1 submission, the other submission with the highest score in the February leaderboard (when submissions closed) was chosen by the system and locked in\n- If you hand-selected 0 submissions, two submissions with the highest scores in the February leaderboard were chosen by the system and locked in\n\nThe most recent change is because of this:\n- When we update the ground truth, Kaggle rescores everything of every team (i.e., all 41k submissions).\n- The previous leaderboard available around mid-May was based on the ground truth I described in this [post](https://www.kaggle.com/competitions/cafa-6-protein-function-prediction/discussion/699352#3457425). This remains our final ground truth data.\n- The current leaderboard ranking is only based on the 2 locked in submissions as described above, per Kaggle competition rules. If your ranking dropped between mid-May and current, that means you did have better scoring-submissions based on the May leaderboard but these were not the highest-scoring in the February leaderboard, so they are not part of the final ranking.\n- You will see errors in all of your submissions right now, except for the 2 submissions that were locked in. This is because we have invalidated the remaining submissions to exclude them from the final ranking. Now each team only has 2 entries, and the better one is shown on the public leaderboard." }, { "id": 3463602, "postDate": "2026-05-27T12:02:30.483Z", "content": "

does it mean that our final position is the same as the positions shown in the public lb now? Just by the two sbumissions chosen by system and does not taged with error? Our final score will be decided by the two not error submissions?😶

", "rawMarkdown": "does it mean that our final position is the same as the positions shown in the public lb now? Just by the two sbumissions chosen by system and does not taged with error? Our final score will be decided by the two not error submissions?😶\n", "votes": 1 }, { "id": 3463607, "postDate": "2026-05-27T12:08:48.240Z", "content": "

Yes, the submissions chosen by the system are the ones that are not tagged with errors. These two submissions decide your final score.

", "rawMarkdown": "Yes, the submissions chosen by the system are the ones that are not tagged with errors. These two submissions decide your final score." }, { "id": 3463608, "postDate": "2026-05-27T12:19:26.283Z", "content": "
\n

Yes, the submissions chosen by the system are the ones that are not tagged with errors. These two submissions decide your final score.

\n
\n

\"\"\nDear organizers,

\n

In another post, you mentioned that the current leaderboard represents the final ranking. At that time, the ranking appeared to be based on each team’s highest-scoring submission.

\n

However, if higher-scoring submissions are now excluded from the final ranking, then some of the current top-ranked solutions may actually be weaker than submissions that were not selected as final submissions. In other words, many submissions that scored higher than some of the current gold-medal solutions have been invalidated.

\n

In that case, can the current top-ranked solutions still be considered the best solutions? Wouldn’t this be somewhat inconsistent with the original purpose of holding this competition?\n\"\"\nMy current highest-scoring submission has a score of 0.44682, which is higher than the current third place.

", "rawMarkdown": "> Yes, the submissions chosen by the system are the ones that are not tagged with errors. These two submissions decide your final score.\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17190502%2Fbc863a33a136f91fd145cfce59b0b3dd%2F.png?generation=1779884357260291&alt=media)\nDear organizers,\n\nIn another post, you mentioned that the current leaderboard represents the final ranking. At that time, the ranking appeared to be based on each team’s highest-scoring submission.\n\nHowever, if higher-scoring submissions are now excluded from the final ranking, then some of the current top-ranked solutions may actually be weaker than submissions that were not selected as final submissions. In other words, many submissions that scored higher than some of the current gold-medal solutions have been invalidated.\n\nIn that case, can the current top-ranked solutions still be considered the best solutions? Wouldn’t this be somewhat inconsistent with the original purpose of holding this competition?\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17190502%2Fd58c158b2a0fa566d97751ef61982573%2F.png?generation=1779884596261043&alt=media)\nMy current highest-scoring submission has a score of 0.44682, which is higher than the current third place.\n", "votes": 1 }, { "id": 3463610, "postDate": "2026-05-27T12:36:11.260Z", "content": "
\n

In another post, you mentioned that the current leaderboard represents the final ranking. At that time, the ranking appeared to be based on each team’s highest-scoring submission.

\n
\n

Yes, at that time, the ranking included all 41k submissions (meaning all submissions for all teams). However, each team could only have 2 final submissions, so the most recent leaderboard implemented that.

\n
\n

In other words, many submissions that scored higher than some of the current gold-medal solutions have been invalidated.

\n
\n

Yes, that is possible with your example, however remember that the competition was only meant to allow 2 submissions per team. If we did not invalidate other submissions that were not the 2 locked in, teams that didn't hand-select their submissions could have more than 2 chances to be placed in the final ranking, which is not fair in the competition. That's why we implemented the \"2 per team based on February leaderboard\" mechanism, to give every team only 2 chances to be ranked in the final ranking.

", "rawMarkdown": ">In another post, you mentioned that the current leaderboard represents the final ranking. At that time, the ranking appeared to be based on each team’s highest-scoring submission.\n\nYes, at that time, the ranking included all 41k submissions (meaning all submissions for all teams). However, each team could only have 2 final submissions, so the most recent leaderboard implemented that.\n\n>In other words, many submissions that scored higher than some of the current gold-medal solutions have been invalidated.\n\nYes, that is possible with your example, however remember that the competition was only meant to allow 2 submissions per team. If we did not invalidate other submissions that were not the 2 locked in, teams that didn't hand-select their submissions could have more than 2 chances to be placed in the final ranking, which is not fair in the competition. That's why we implemented the \"2 per team based on February leaderboard\" mechanism, to give every team only 2 chances to be ranked in the final ranking. " }, { "id": 3463632, "postDate": "2026-05-27T14:13:07.437Z", "content": "

Thanks for the clarification.

\n

A bit disappointing on my end: as I'm not the team leader I couldn't hand-select, so the two auto-locked entries were the overfit February picks. The mid-May rescore on the held-out ground truth had moved us to #2; with proper hand-selection I'd probably have landed somewhere between #7 and #12, which is the honest position.

\n

Everything checks out. Thanks for the detailed write-up and congrats to the winners.

", "rawMarkdown": "Thanks for the clarification.\n\nA bit disappointing on my end: as I'm not the team leader I couldn't hand-select, so the two auto-locked entries were the overfit February picks. The mid-May rescore on the held-out ground truth had moved us to #2; with proper hand-selection I'd probably have landed somewhere between #7 and #12, which is the honest position.\n\nEverything checks out. Thanks for the detailed write-up and congrats to the winners." }, { "id": 3463711, "postDate": "2026-05-27T18:05:56.220Z", "content": "

We understand the frustration, so if you or anyone is still interested in showcasing your method, LAFA will be the platform to do so and we look forward to seeing more method being hosted on LAFA

", "rawMarkdown": "We understand the frustration, so if you or anyone is still interested in showcasing your method, LAFA will be the platform to do so and we look forward to seeing more method being hosted on [LAFA](https://www.kaggle.com/competitions/cafa-6-protein-function-prediction/discussion/702254)", "votes": 1 }, { "id": 3463829, "postDate": "2026-05-28T03:10:15.767Z", "content": "

Hi organizers, will there be a CAFA7?

", "rawMarkdown": "Hi organizers, will there be a CAFA7?", "votes": 1 } ] }, { "id": 3463513, "postDate": "2026-05-27T04:16:48.523Z", "content": "

If we do now handlely chose two, how is the final postion choses the top solutions as the score of the final postion

", "rawMarkdown": "If we do now handlely chose two, how is the final postion choses the top solutions as the score of the final postion", "votes": 1, "replies": [ { "id": 3463597, "postDate": "2026-05-27T11:45:34.570Z", "content": "

I don't think you can hand-select submissions as final submissions anymore, that option was only available before submission closes

", "rawMarkdown": "I don't think you can hand-select submissions as final submissions anymore, that option was only available before submission closes" }, { "id": 3463600, "postDate": "2026-05-27T11:57:27.490Z", "content": "

yes, and is the public lb shown now is the final position?I just noticed other submissions were taged error in the submission list😃

", "rawMarkdown": "yes, and is the public lb shown now is the final position?I just noticed other submissions were taged error in the submission list😃", "votes": 1 } ] } ] } ], "comments": [ { "id": 3463416, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-26T20:55:10.263000", "content": "

Please check the most current leaderboard, and expect a final post announcing the winners from the Kaggle staff over the next few days. And yes, the leaderboard display was rounded to 3 decimals, the actual value has more digits.

", "votes": 0, "replies": [ { "id": 3463421, "author_name": "sadek01", "author_url": "", "post_date": "2026-05-26T21:05:51.423000", "content": "

May i ask why are my highest scoring submissions not picked when in the previous rescoring they were picked ?

", "votes": 1, "replies": [ { "id": 3463430, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-26T21:13:42.567000", "content": "

The ones that are currently displayed on the leaderboard were only the highest-scoring 2 submissions per team based on the February evaluation (when submission closes), if your team did not hand-select your top 2. That's why you see only 2 entries per team right now on the leaderboard. Before this current leaderboard, we rescored all submissions of every team so the ranking was different.

", "votes": 0, "replies": [] }, { "id": 3463485, "author_name": "", "author_url": "", "post_date": "2026-05-27T02:37:43.910000", "content": "", "votes": 0, "replies": [] }, { "id": 3463495, "author_name": "", "author_url": "", "post_date": "2026-05-27T03:03:20.663000", "content": "", "votes": 0, "replies": [] }, { "id": 3463504, "author_name": "Baidalin Adilzhan [dsmlkz]", "author_url": "", "post_date": "2026-05-27T03:33:39.583000", "content": "

@ahphan What if my selected submissions failed even early rescores and my high scoring submissions were automatically selected by system for 15th may update? Why 2 subs that were selected 15th may were unselected and two random subs were selected again, as you told prev rescore and lb was final?

", "votes": 1, "replies": [] }, { "id": 3463511, "author_name": "", "author_url": "", "post_date": "2026-05-27T04:15:15.480000", "content": "", "votes": 0, "replies": [] }, { "id": 3463512, "author_name": "Born4", "author_url": "", "post_date": "2026-05-27T04:15:42.697000", "content": "

hi, is the public lb position now the final position? How is the final private position? Is it the same as the private lb last time, or the same as the public lb shown now

", "votes": 2, "replies": [] }, { "id": 3463580, "author_name": "", "author_url": "", "post_date": "2026-05-27T10:07:41.147000", "content": "", "votes": 0, "replies": [] }, { "id": 3463581, "author_name": "xaxipiruli", "author_url": "", "post_date": "2026-05-27T10:07:59.743000", "content": "

Same here. I also doubt that the two currently selected submissions were the ones automatically chosen at the deadline. The only thing I clearly remember is that, during the last week of the competition, I couldn’t beat my own public leaderboard score. So it does seem likely that the two selected submissions came from that period.

\n

But at this point, after so many changes, I can’t really be sure anymore. That’s why I mentioned that the original leaderboard probably should have been preserved given how many things changed afterwards.

", "votes": 2, "replies": [] }, { "id": 3463601, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-27T11:57:35.337000", "content": "

The two currently selected submissions were chosen as follows:

\n\n

The most recent change is because of this:

\n", "votes": 0, "replies": [] }, { "id": 3463602, "author_name": "Born4", "author_url": "", "post_date": "2026-05-27T12:02:30.483000", "content": "

does it mean that our final position is the same as the positions shown in the public lb now? Just by the two sbumissions chosen by system and does not taged with error? Our final score will be decided by the two not error submissions?😶

", "votes": 1, "replies": [] }, { "id": 3463607, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-27T12:08:48.240000", "content": "

Yes, the submissions chosen by the system are the ones that are not tagged with errors. These two submissions decide your final score.

", "votes": 0, "replies": [] }, { "id": 3463608, "author_name": "HUG", "author_url": "", "post_date": "2026-05-27T12:19:26.283000", "content": "
\n

Yes, the submissions chosen by the system are the ones that are not tagged with errors. These two submissions decide your final score.

\n
\n

\"\"\nDear organizers,

\n

In another post, you mentioned that the current leaderboard represents the final ranking. At that time, the ranking appeared to be based on each team’s highest-scoring submission.

\n

However, if higher-scoring submissions are now excluded from the final ranking, then some of the current top-ranked solutions may actually be weaker than submissions that were not selected as final submissions. In other words, many submissions that scored higher than some of the current gold-medal solutions have been invalidated.

\n

In that case, can the current top-ranked solutions still be considered the best solutions? Wouldn’t this be somewhat inconsistent with the original purpose of holding this competition?\n\"\"\nMy current highest-scoring submission has a score of 0.44682, which is higher than the current third place.

", "votes": 1, "replies": [] }, { "id": 3463610, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-27T12:36:11.260000", "content": "
\n

In another post, you mentioned that the current leaderboard represents the final ranking. At that time, the ranking appeared to be based on each team’s highest-scoring submission.

\n
\n

Yes, at that time, the ranking included all 41k submissions (meaning all submissions for all teams). However, each team could only have 2 final submissions, so the most recent leaderboard implemented that.

\n
\n

In other words, many submissions that scored higher than some of the current gold-medal solutions have been invalidated.

\n
\n

Yes, that is possible with your example, however remember that the competition was only meant to allow 2 submissions per team. If we did not invalidate other submissions that were not the 2 locked in, teams that didn't hand-select their submissions could have more than 2 chances to be placed in the final ranking, which is not fair in the competition. That's why we implemented the \"2 per team based on February leaderboard\" mechanism, to give every team only 2 chances to be ranked in the final ranking.

", "votes": 0, "replies": [] }, { "id": 3463632, "author_name": "xaxipiruli", "author_url": "", "post_date": "2026-05-27T14:13:07.437000", "content": "

Thanks for the clarification.

\n

A bit disappointing on my end: as I'm not the team leader I couldn't hand-select, so the two auto-locked entries were the overfit February picks. The mid-May rescore on the held-out ground truth had moved us to #2; with proper hand-selection I'd probably have landed somewhere between #7 and #12, which is the honest position.

\n

Everything checks out. Thanks for the detailed write-up and congrats to the winners.

", "votes": 0, "replies": [] }, { "id": 3463711, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-27T18:05:56.220000", "content": "

We understand the frustration, so if you or anyone is still interested in showcasing your method, LAFA will be the platform to do so and we look forward to seeing more method being hosted on LAFA

", "votes": 1, "replies": [] }, { "id": 3463829, "author_name": "HUG", "author_url": "", "post_date": "2026-05-28T03:10:15.767000", "content": "

Hi organizers, will there be a CAFA7?

", "votes": 1, "replies": [] } ] }, { "id": 3463513, "author_name": "Born4", "author_url": "", "post_date": "2026-05-27T04:16:48.523000", "content": "

If we do now handlely chose two, how is the final postion choses the top solutions as the score of the final postion

", "votes": 1, "replies": [ { "id": 3463597, "author_name": "An Phan", "author_url": "", "post_date": "2026-05-27T11:45:34.570000", "content": "

I don't think you can hand-select submissions as final submissions anymore, that option was only available before submission closes

", "votes": 0, "replies": [] }, { "id": 3463600, "author_name": "Born4", "author_url": "", "post_date": "2026-05-27T11:57:27.490000", "content": "

yes, and is the public lb shown now is the final position?I just noticed other submissions were taged error in the submission list😃

", "votes": 1, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3463270": "Quick question on the rescored leaderboard: positions 4, 5 and 6 all display 0.446. Is this a genuine three-way tie at full precision, or just three-decimal display rounding over values that actually differ? Could you expose one or two more decimals for the final standings, or confirm the tie-breaking rule? Thanks!", "3463416": "Please check the most current leaderboard, and expect a final post announcing the winners from the Kaggle staff over the next few days. And yes, the leaderboard display was rounded to 3 decimals, the actual value has more digits." } }