{ "id": 104028, "title": "If we have to give away prize based on public LB submissions right now", "url": "/competitions/generative-dog-images/discussion/104028", "author_name": "", "post_date": "2019-08-13T18:17:30.579171400Z", "votes": 14, "comment_count": 19, "views": 0, "content": "

Well that is a fictional matter : However if we go through the patterns of peoples submission , last submission date and gradual improvements with a little bit of margin of error below is the leader board \n\"\"\n\"\"

", "messages": [ { "id": "598564", "postDate": "08/13/2019 18:17:30", "content": "

Well that is a fictional matter : However if we go through the patterns of peoples submission , last submission date and gradual improvements with a little bit of margin of error below is the leader board \n\"\"\n\"\"

", "rawMarkdown": "Well that is a fictional matter : However if we go through the patterns of peoples submission , last submission date and gradual improvements with a little bit of margin of error below is the leader board \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234817%2Fbca213486d4cac2accc1ed0edd05d43a%2Fleaderboard.png?generation=1565720231556358&alt=media)\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234817%2F69034a7385a04962f0b4ce84425f1523%2Fleaderboard1.png?generation=1565720226954830&alt=media)", "votes": null }, { "id": "598568", "postDate": "08/13/2019 18:22:25", "content": "

You missed out the god - bestfitting. He has ~12 public LB. \nAnd don't forget, another legend - Giba, he is participating too.

", "rawMarkdown": "You missed out the god - bestfitting. He has ~12 public LB. \nAnd don't forget, another legend - Giba, he is participating too.", "votes": null }, { "id": "598582", "postDate": "08/13/2019 18:47:01", "content": "

Didnt miss them for sure . I will give it for bestfitting , May be he just started submitting from 29 score and now in range of 12 . In that case he is topping the chart . Whereas if you see the comparison between Yabea and Giba , you can see why I discarded Mr Giba , for this competition . \"\"

", "rawMarkdown": "Didnt miss them for sure . I will give it for bestfitting , May be he just started submitting from 29 score and now in range of 12 . In that case he is topping the chart . Whereas if you see the comparison between Yabea and Giba , you can see why I discarded Mr Giba , for this competition . ![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234817%2F017669c7309636c6eb6c3aea2b898579%2Fleaderboard3.png?generation=1565722017684227&alt=media)", "votes": null }, { "id": "598594", "postDate": "08/13/2019 18:56:11", "content": "

Nice list!

", "rawMarkdown": "Nice list!", "votes": null }, { "id": "598764", "postDate": "08/14/2019 02:24:38", "content": "

Some team have submitted a memorize GAN score so their really score is hidden.\nSo I think it will be more competitive for the final result.

", "rawMarkdown": "Some team have submitted a memorize GAN score so their really score is hidden.\nSo I think it will be more competitive for the final result.", "votes": null }, { "id": "598775", "postDate": "08/14/2019 02:39:02", "content": "

Thanks for sharing.

", "rawMarkdown": "Thanks for sharing.", "votes": null }, { "id": "598869", "postDate": "08/14/2019 06:48:20", "content": "

Nice chart! I think there are at least a few more people with results better than mine. And the current metric has a few flaws. IMO it biased towards diversity too much. It's possible to get a good score with lots of crappy pics. Hard to say how the 'secret' private model will behave at stage 2.

", "rawMarkdown": "Nice chart! I think there are at least a few more people with results better than mine. And the current metric has a few flaws. IMO it biased towards diversity too much. It's possible to get a good score with lots of crappy pics. Hard to say how the 'secret' private model will behave at stage 2.", "votes": null }, { "id": "598878", "postDate": "08/14/2019 07:06:19", "content": "

Yes they are . But how I have checked is ..gradual improvement and the timeline of posting . If you download the raw data from leaderboard .you can see some people posted a score of 31 or 15 in June and then didn't progress ..Your submissions shows a gradual improvement from 50 odd .

", "rawMarkdown": "Yes they are . But how I have checked is ..gradual improvement and the timeline of posting . If you download the raw data from leaderboard .you can see some people posted a score of 31 or 15 in June and then didn't progress ..Your submissions shows a gradual improvement from 50 odd .", "votes": null }, { "id": "598880", "postDate": "08/14/2019 07:10:08", "content": "

True, current metric is definitely biased towards diversity. That explains why my rushed up crappy conditional GAN performs better than a non-CGAN I crafted in several days.

", "rawMarkdown": "True, current metric is definitely biased towards diversity. That explains why my rushed up crappy conditional GAN performs better than a non-CGAN I crafted in several days.", "votes": null }, { "id": "598899", "postDate": "08/14/2019 07:50:02", "content": "

I guess this chart may miss a lot of teams...

", "rawMarkdown": "I guess this chart may miss a lot of teams...", "votes": null }, { "id": "599010", "postDate": "08/14/2019 11:29:19", "content": "

Yes, this chart is missing my traditional GAN submissions.

\n\n

Nice chart Nirjhar. Is this chart missing everyone who submitted a memorizer GAN (in addition to traditional GANs)?

", "rawMarkdown": "Yes, this chart is missing my traditional GAN submissions.\n\nNice chart Nirjhar. Is this chart missing everyone who submitted a memorizer GAN (in addition to traditional GANs)?", "votes": null }, { "id": "599051", "postDate": "08/14/2019 12:36:43", "content": "

Hi Chris , I took the cut off as < 61 . Probably the latest kernel of you with 50+ either not submitted or there is a timelag in raw data .

", "rawMarkdown": "Hi Chris , I took the cut off as < 61 . Probably the latest kernel of you with 50+ either not submitted or there is a timelag in raw data .", "votes": null }, { "id": "599061", "postDate": "08/14/2019 12:48:50", "content": "

Wow, definitely didn't expect to see my name up here in my first comp :D. Who knows what the final standings will be though.

", "rawMarkdown": "Wow, definitely didn't expect to see my name up here in my first comp :D. Who knows what the final standings will be though.", "votes": null }, { "id": "599072", "postDate": "08/14/2019 13:09:16", "content": "

Weird, there's probably a time delay. I submitted an LB 58 three days ago and an LB 55 yesterday. This morning I submitted an LB 47!! You should update this post after the deadline today (and again 24 hours later to capture delay).

\n\n

I'm surprised that there are only 90 teams with a traditional GAN better than 61. That means someone could submit your LB 60 kernel and get a bronze medal.

", "rawMarkdown": "Weird, there's probably a time delay. I submitted an LB 58 three days ago and an LB 55 yesterday. This morning I submitted an LB 47!! You should update this post after the deadline today (and again 24 hours later to capture delay).\n\nI'm surprised that there are only 90 teams with a traditional GAN better than 61. That means someone could submit your LB 60 kernel and get a bronze medal.", "votes": null }, { "id": "599076", "postDate": "08/14/2019 13:10:50", "content": "

It is just wishful thinking . But hope and wish are important to have .

", "rawMarkdown": "It is just wishful thinking . But hope and wish are important to have .", "votes": null }, { "id": "599141", "postDate": "08/14/2019 14:47:37", "content": "

I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.

", "rawMarkdown": "I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.", "votes": null }, { "id": "599151", "postDate": "08/14/2019 15:05:15", "content": "

Guess only the best score is kept in the LB?

", "rawMarkdown": "Guess only the best score is kept in the LB?", "votes": null }, { "id": "599169", "postDate": "08/14/2019 15:35:09", "content": "
\n

I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.

\n
\n\n

Ah, that explains it. If you submitted a memorizer GAN weeks ago scoring LB 7 (which 300+ teams did). And today you submit a traditional GAN scoring 50, the new submission doesn't get recorded.

\n\n

If that's the case, then the list in this post is missing many teams.

", "rawMarkdown": "> I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.\n\nAh, that explains it. If you submitted a memorizer GAN weeks ago scoring LB 7 (which 300+ teams did). And today you submit a traditional GAN scoring 50, the new submission doesn't get recorded.\n\nIf that's the case, then the list in this post is missing many teams.", "votes": null }, { "id": "599256", "postDate": "08/14/2019 18:16:04", "content": "

\"\"

", "rawMarkdown": "![](https://i.kym-cdn.com/news_feeds/icons/mobile/000/038/929/118.jpg)", "votes": null }, { "id": "599258", "postDate": "08/14/2019 18:18:14", "content": "

Even though my 30-ish score is legit, I'm 99% sure I won't be in the top 5, I may also miss the gold medal by a few. \nI will be sharing my solution tomorrow morning though :)

", "rawMarkdown": "Even though my 30-ish score is legit, I'm 99% sure I won't be in the top 5, I may also miss the gold medal by a few. \nI will be sharing my solution tomorrow morning though :)", "votes": null } ], "comments": [ { "id": 598568, "author_name": "timetraveller98", "author_url": "", "post_date": "08/13/2019 18:22:25", "content": "

You missed out the god - bestfitting. He has ~12 public LB. \nAnd don't forget, another legend - Giba, he is participating too.

", "votes": null, "replies": [ { "id": 598582, "author_name": "phoenix9032", "author_url": "", "post_date": "08/13/2019 18:47:01", "content": "

Didnt miss them for sure . I will give it for bestfitting , May be he just started submitting from 29 score and now in range of 12 . In that case he is topping the chart . Whereas if you see the comparison between Yabea and Giba , you can see why I discarded Mr Giba , for this competition . \"\"

", "votes": null, "replies": [] } ] }, { "id": 598594, "author_name": "mightyrains", "author_url": "", "post_date": "08/13/2019 18:56:11", "content": "

Nice list!

", "votes": null, "replies": [] }, { "id": 598764, "author_name": "rhonin0", "author_url": "", "post_date": "08/14/2019 02:24:38", "content": "

Some team have submitted a memorize GAN score so their really score is hidden.\nSo I think it will be more competitive for the final result.

", "votes": null, "replies": [] }, { "id": 598775, "author_name": "suneelpatel", "author_url": "", "post_date": "08/14/2019 02:39:02", "content": "

Thanks for sharing.

", "votes": null, "replies": [] }, { "id": 598869, "author_name": "dvorobiev", "author_url": "", "post_date": "08/14/2019 06:48:20", "content": "

Nice chart! I think there are at least a few more people with results better than mine. And the current metric has a few flaws. IMO it biased towards diversity too much. It's possible to get a good score with lots of crappy pics. Hard to say how the 'secret' private model will behave at stage 2.

", "votes": null, "replies": [ { "id": 598878, "author_name": "phoenix9032", "author_url": "", "post_date": "08/14/2019 07:06:19", "content": "

Yes they are . But how I have checked is ..gradual improvement and the timeline of posting . If you download the raw data from leaderboard .you can see some people posted a score of 31 or 15 in June and then didn't progress ..Your submissions shows a gradual improvement from 50 odd .

", "votes": null, "replies": [] }, { "id": 598880, "author_name": "timetraveller98", "author_url": "", "post_date": "08/14/2019 07:10:08", "content": "

True, current metric is definitely biased towards diversity. That explains why my rushed up crappy conditional GAN performs better than a non-CGAN I crafted in several days.

", "votes": null, "replies": [] } ] }, { "id": 598899, "author_name": "jiujing23333", "author_url": "", "post_date": "08/14/2019 07:50:02", "content": "

I guess this chart may miss a lot of teams...

", "votes": null, "replies": [ { "id": 599010, "author_name": "cdeotte", "author_url": "", "post_date": "08/14/2019 11:29:19", "content": "

Yes, this chart is missing my traditional GAN submissions.

\n\n

Nice chart Nirjhar. Is this chart missing everyone who submitted a memorizer GAN (in addition to traditional GANs)?

", "votes": null, "replies": [] }, { "id": 599051, "author_name": "phoenix9032", "author_url": "", "post_date": "08/14/2019 12:36:43", "content": "

Hi Chris , I took the cut off as < 61 . Probably the latest kernel of you with 50+ either not submitted or there is a timelag in raw data .

", "votes": null, "replies": [] }, { "id": 599072, "author_name": "cdeotte", "author_url": "", "post_date": "08/14/2019 13:09:16", "content": "

Weird, there's probably a time delay. I submitted an LB 58 three days ago and an LB 55 yesterday. This morning I submitted an LB 47!! You should update this post after the deadline today (and again 24 hours later to capture delay).

\n\n

I'm surprised that there are only 90 teams with a traditional GAN better than 61. That means someone could submit your LB 60 kernel and get a bronze medal.

", "votes": null, "replies": [] }, { "id": 599141, "author_name": "yukia18", "author_url": "", "post_date": "08/14/2019 14:47:37", "content": "

I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.

", "votes": null, "replies": [] }, { "id": 599151, "author_name": "mightyrains", "author_url": "", "post_date": "08/14/2019 15:05:15", "content": "

Guess only the best score is kept in the LB?

", "votes": null, "replies": [] }, { "id": 599169, "author_name": "cdeotte", "author_url": "", "post_date": "08/14/2019 15:35:09", "content": "
\n

I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.

\n
\n\n

Ah, that explains it. If you submitted a memorizer GAN weeks ago scoring LB 7 (which 300+ teams did). And today you submit a traditional GAN scoring 50, the new submission doesn't get recorded.

\n\n

If that's the case, then the list in this post is missing many teams.

", "votes": null, "replies": [] } ] }, { "id": 599061, "author_name": "jadeblue", "author_url": "", "post_date": "08/14/2019 12:48:50", "content": "

Wow, definitely didn't expect to see my name up here in my first comp :D. Who knows what the final standings will be though.

", "votes": null, "replies": [ { "id": 599076, "author_name": "phoenix9032", "author_url": "", "post_date": "08/14/2019 13:10:50", "content": "

It is just wishful thinking . But hope and wish are important to have .

", "votes": null, "replies": [] } ] }, { "id": 599256, "author_name": "theoviel", "author_url": "", "post_date": "08/14/2019 18:16:04", "content": "

\"\"

", "votes": null, "replies": [ { "id": 599258, "author_name": "theoviel", "author_url": "", "post_date": "08/14/2019 18:18:14", "content": "

Even though my 30-ish score is legit, I'm 99% sure I won't be in the top 5, I may also miss the gold medal by a few. \nI will be sharing my solution tomorrow morning though :)

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "598564": "Well that is a fictional matter : However if we go through the patterns of peoples submission , last submission date and gradual improvements with a little bit of margin of error below is the leader board \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234817%2Fbca213486d4cac2accc1ed0edd05d43a%2Fleaderboard.png?generation=1565720231556358&alt=media)\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234817%2F69034a7385a04962f0b4ce84425f1523%2Fleaderboard1.png?generation=1565720226954830&alt=media)", "598568": "You missed out the god - bestfitting. He has ~12 public LB. \nAnd don't forget, another legend - Giba, he is participating too.", "598582": "Didnt miss them for sure . I will give it for bestfitting , May be he just started submitting from 29 score and now in range of 12 . In that case he is topping the chart . Whereas if you see the comparison between Yabea and Giba , you can see why I discarded Mr Giba , for this competition . ![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234817%2F017669c7309636c6eb6c3aea2b898579%2Fleaderboard3.png?generation=1565722017684227&alt=media)", "598594": "Nice list!", "598764": "Some team have submitted a memorize GAN score so their really score is hidden.\nSo I think it will be more competitive for the final result.", "598775": "Thanks for sharing.", "598869": "Nice chart! I think there are at least a few more people with results better than mine. And the current metric has a few flaws. IMO it biased towards diversity too much. It's possible to get a good score with lots of crappy pics. Hard to say how the 'secret' private model will behave at stage 2.", "598878": "Yes they are . But how I have checked is ..gradual improvement and the timeline of posting . If you download the raw data from leaderboard .you can see some people posted a score of 31 or 15 in June and then didn't progress ..Your submissions shows a gradual improvement from 50 odd .", "598880": "True, current metric is definitely biased towards diversity. That explains why my rushed up crappy conditional GAN performs better than a non-CGAN I crafted in several days.", "598899": "I guess this chart may miss a lot of teams...", "599010": "Yes, this chart is missing my traditional GAN submissions.\n\nNice chart Nirjhar. Is this chart missing everyone who submitted a memorizer GAN (in addition to traditional GANs)?", "599051": "Hi Chris , I took the cut off as < 61 . Probably the latest kernel of you with 50+ either not submitted or there is a timelag in raw data .", "599061": "Wow, definitely didn't expect to see my name up here in my first comp :D. Who knows what the final standings will be though.", "599072": "Weird, there's probably a time delay. I submitted an LB 58 three days ago and an LB 55 yesterday. This morning I submitted an LB 47!! You should update this post after the deadline today (and again 24 hours later to capture delay).\n\nI'm surprised that there are only 90 teams with a traditional GAN better than 61. That means someone could submit your LB 60 kernel and get a bronze medal.", "599076": "It is just wishful thinking . But hope and wish are important to have .", "599141": "I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.", "599151": "Guess only the best score is kept in the LB?", "599169": "> I think it is not time delay. A submission log is recoded into the raw data of LB when you take best score at that time.\n\nAh, that explains it. If you submitted a memorizer GAN weeks ago scoring LB 7 (which 300+ teams did). And today you submit a traditional GAN scoring 50, the new submission doesn't get recorded.\n\nIf that's the case, then the list in this post is missing many teams.", "599256": "![](https://i.kym-cdn.com/news_feeds/icons/mobile/000/038/929/118.jpg)", "599258": "Even though my 30-ish score is legit, I'm 99% sure I won't be in the top 5, I may also miss the gold medal by a few. \nI will be sharing my solution tomorrow morning though :)" }, "source": "meta" }